EffectsandParadoxes

TIME · Proven · Edward Lorenz, 1963

Butterfly Effect

A rounding error in the third decimal place destroyed the weather forecast.

In plain words

Some things in the world are very touchy about how they start. Change the start by a tiny amount and the ending comes out completely different. The weather is like this. A puff of air today can, weeks later, be the difference between a sunny day and a storm. Scientists call this chaos. It does not mean the weather is random. Each step follows clear rules. The trouble is that no one can measure today's air perfectly, and the tiny bits we miss grow bigger every day. So forecasts for next month are guesses, even with the best computers.

An everyday example

Stand a pencil on its tip and let go. Which way it falls depends on a wobble too small to see. Do it ten times and it falls ten different ways. Now watch a leaf drop into a stream. Same start, as far as you can tell, but a different path every time. That is chaos on your kitchen table.

The short version

Lorenz reran a weather simulation using 0.506 instead of 0.506127 and got a completely different season. In some systems the tiniest difference at the start grows exponentially, so the future is not just hard to predict, it is mathematically locked away from us.

How it actually works

Sensitive dependence on initial conditions belongs to deterministic systems, not random ones. The equations fix the future exactly, but two starting states separated by a small distance move apart exponentially, at a rate set by the largest Lyapunov exponent. The reciprocal of that exponent, the Lyapunov time, is the interval over which the gap grows by a factor of about 2.7. Measuring the initial state ten times more precisely therefore extends useful prediction by only about 2.3 Lyapunov times. Lorenz estimated in a 1969 paper in Tellus that errors in the smallest scales of atmospheric motion grow within hours and cascade upward, imposing a limit of about two weeks on detailed weather prediction.

The 1961 accident happened on a Royal McBee LGP-30 running a twelve-variable model. Lorenz wanted to extend a run, so he typed in values from an earlier printout, which showed three decimal places, while the machine held six. One variable went in as 0.506 rather than 0.506127, about one part in four thousand. The two runs tracked each other briefly, then diverged completely. The 1963 paper reduced the problem to three ordinary differential equations for convection in a heated fluid layer, with parameters 10, 28 and 8/3, and showed solutions that were bounded, never repeated, and traced the shape now called the Lorenz attractor. Forecasters now run ensembles of perturbed starts; the European Centre for Medium-Range Weather Forecasts has done so since 1992.

Edward Lorenz, a meteorologist at the Massachusetts Institute of Technology, published 'Deterministic Nonperiodic Flow' in the Journal of the Atmospheric Sciences in 1963. In its closing discussion he quoted a colleague's remark that one flap of a seagull's wings could alter the course of the weather forever. The butterfly arrived on 29 December 1972, when Lorenz addressed the American Association for the Advancement of Science in Washington; he had not supplied a title, so the session organiser Philip Merilees wrote one: 'Predictability: Does the Flap of a Butterfly's Wings in Brazil Set Off a Tornado in Texas?' Henri Poincaré had found the same sensitivity in the three-body problem in the 1890s.

Where you have seen it

Weather forecasts stay useful for about a week and then fall apart, no matter how good the satellites get.

In films, books and games

The film The Butterfly Effect (2004) takes the name for a story about a man changing his own past. Jurassic Park (1993) has Ian Malcolm explain chaos with a drop of water on Ellie's hand. Ray Bradbury's short story A Sound of Thunder (1952) has a time traveller crush a butterfly and come home to a changed world.

The uncomfortable bit

It never claimed a butterfly causes a typhoon. It claims we can never gather enough information to rule it out.

Source: Lorenz, 'Deterministic Nonperiodic Flow', Journal of the Atmospheric Sciences, 1963

TIME · Thought experiment · Pulp fiction, 1930s

Grandfather Paradox

Go back, stop your grandparents from meeting, and you were never born to go back.

In plain words

Suppose a time machine exists. You go back sixty years and stop your grandfather from ever meeting your grandmother. Then your parent is never born, so you are never born. But if you were never born, nobody went back to stop the meeting, so it happened after all, and you were born. Round and round it goes. Each half of the story cancels out the other half. This is the biggest argument against travel into the past. The universe, it seems, cannot allow a change that wipes out its own cause.

An everyday example

Try it with a note. Write: whoever reads this must tear it up before it is written. The rule eats itself. Or think of a WhatsApp message that, when it arrives, deletes the phone that sent it. If the phone is gone, the message was never sent, so the phone is fine, so the message arrives. Nothing settles.

The short version

The classic argument against travelling into the past. Any change you make in the past can erase the reason you travelled in the first place, which erases the change, which restores the reason. The loop has no stable answer.

Where you have seen it

Every time travel film picks one rule to survive this, and most quietly break their own rule by the third act.

In films, books and games

Back to the Future (1985) plays it for laughs: Marty stops his parents meeting and watches himself fade from a photograph. The Futurama episode Roswell That Ends Well (2001) has Fry get his grandfather killed and then become his own grandfather.

The uncomfortable bit

Physics has an escape hatch. The Novikov self consistency principle says only self consistent timelines are physically allowed, so your gun would jam. Every single time.

Source: Schachner, 'Ancestral Voices', Astounding Stories, 1933; named in Barjavel, Le Voyageur Imprudent, 1943

TIME · Thought experiment · Robert Heinlein, 1941

Bootstrap Paradox

A book with no author. It exists only because someone carried it into its own past.

In plain words

A time traveller brings something from the future into the past. That thing then gets passed forward through the years until it reaches the traveller, who takes it back again. The question is where it came from. Nobody made it. It just goes round in a circle with no start. This can be an object, a song, or a piece of knowledge. The strange part is that nothing in the story breaks a rule. Every step looks fine on its own. It is the whole loop that has no beginning, and our minds are not built to accept that.

An everyday example

Your friend sends you a joke on WhatsApp. Later you send the same joke to them, and they say that is where they first heard it. Trace it backwards. If you got it from them and they got it from you, the joke has no author. In real life someone started it. In a time loop, nobody did.

The short version

You travel back and hand a young composer the sheet music of his most famous symphony. He writes it down, it becomes famous, you learn it, you carry it back. Nobody ever composed it.

Where you have seen it

The old man gives you the blueprints for the time machine that you will later hand to him.

In films, books and games

In Somewhere in Time (1980) an old woman gives the hero a pocket watch, which he carries into the past and gives to her young self. The Doctor Who episode Before the Flood (2015) opens with the Doctor explaining the loop to camera using Beethoven's Fifth. Predestination (2014) is the idea applied to a person.

The uncomfortable bit

No physical law is broken here. The information simply has no origin, and that is the part nobody can stomach.

Source: Heinlein, 'By His Bootstraps', Astounding Science Fiction, 1941

TIME · Proven · Albert Einstein, 1905

Twin Paradox

One twin flies away and comes home younger than the brother who stayed.

In plain words

Time does not tick at the same speed for everyone. The faster you move, the slower your clock runs compared with someone standing still. This is not a fault in the clock. Your heartbeat, your ageing, everything slows. So picture two twins. One stays on Earth. The other goes on a very fast rocket trip and comes back. The rocket twin has aged less. Two things puzzle people. First, that it happens at all. Second, since each twin sees the other moving, why it is not equal. The answer is that only the rocket twin turned around, and that turn breaks the tie.

An everyday example

Two friends each start a stopwatch. One sits on the bench. The other runs laps around the field and comes back. In real life the watches match. Now speed the runner up to nearly the speed of light. The runner's watch shows less time has passed. The runner is now a little younger than the friend on the bench.

The short version

Time runs slower for anything moving fast relative to you. The travelling twin turns around, changes frames, and ages less. It sounds symmetric but it is not, and the difference is real and measurable.

How it actually works

Special relativity says a moving clock runs slow by the factor gamma, one over the square root of one minus v squared over c squared, and each twin sees the other as the moving one, which is the apparent contradiction. The twins are not in the same situation. The one who stays occupies a single inertial frame throughout. The one who travels occupies at least two, an outbound frame and a returning one, and must accelerate to pass between them. In the geometry of special relativity the straight worldline between departure and reunion is the one with the most elapsed time, so any detour ages less.

Joseph Hafele and Richard Keating tested it in October 1971 with caesium atomic clocks flown around the world on airliners, first eastward then westward, against United States Naval Observatory clocks. Two effects combined: special-relativistic slowing, set by the aircraft's speed added to or subtracted from the Earth's rotation, and general-relativistic speeding up at altitude. The eastward prediction was a loss of 40 nanoseconds, give or take 23; the clocks lost 59, give or take 10. The westward prediction was a gain of 275, give or take 21; the clocks gained 273, give or take 7. Both papers appeared in Science on 14 July 1972. GPS satellite clocks run slow by about 7 microseconds a day from motion and fast by about 45 from weaker gravity, a net gain of 38 microseconds, which uncorrected would shift position fixes by roughly ten kilometres a day.

Einstein's 1905 paper 'Zur Elektrodynamik bewegter Körper' in Annalen der Physik already contained the core: a clock carried around a closed path returns lagging behind one that stayed. Paul Langevin turned it into a story of a traveller in a projectile in a 1911 lecture published in Scientia, returning after two of his own years to find two centuries had passed on Earth. Einstein answered the charge of contradiction in a 1918 dialogue in Die Naturwissenschaften. The cosmonaut Sergei Krikalev, with 803 days in orbit, is about 20 milliseconds younger than he would otherwise be.

Where you have seen it

In 1971 two physicists flew atomic clocks around the world on commercial flights and the clocks disagreed with the ones left at home, exactly as predicted.

In films, books and games

Interstellar (2014) is built on it: Cooper returns to find his daughter Murph an old woman while he has barely aged. Planet of the Apes (1968) has astronauts land on an Earth two thousand years older than the one they left. Joe Haldeman's novel The Forever War (1974) sends soldiers home to a society centuries ahead of them.

The uncomfortable bit

Astronauts on long missions genuinely return a fraction of a second younger than the rest of us. Time travel to the future already works. It is just very slow.

Source: Einstein, Annalen der Physik, 1905; Hafele and Keating, Science 177, 1972

TIME · Proven · Einstein, applied 1978

GPS Time Dilation

Your maps app would be wrong by ten kilometres a day if we ignored relativity.

In plain words

The satellites that tell your phone where it is carry very accurate clocks. Two strange things happen to those clocks. Because the satellites move fast, their clocks run a little slow. Because they are far from Earth, where gravity is weaker, their clocks run a little fast. The second effect is bigger, so overall the clocks race ahead. The gap is tiny, far less than a blink each day, but your phone works by timing signals from the satellites. Left uncorrected, the error in your position would grow by kilometres every day. Engineers fix it on purpose, using Einstein's rules.

An everyday example

Open the maps app on your phone and watch the blue dot. It sits within a few metres of where you stand. That dot is worked out by timing radio signals from satellites about twenty thousand kilometres up. Light covers thirty centimetres in a billionth of a second, so even a tiny clock slip would move the dot down the street.

The short version

Satellite clocks run slow because they move fast, and run fast because gravity is weaker up there. The second effect wins. Net result, they gain about 38 microseconds every day.

Where you have seen it

Every ride hailing pin, every food delivery tracker, every fishing boat off Mangaluru is quietly running a correction built from Einstein's equations.

In films, books and games

Interstellar (2014) puts gravitational time dilation on the big screen: an hour on Miller's planet, deep in a black hole's pull, costs the crew left in orbit seven years, the same effect the GPS clocks are corrected for, only enormously exaggerated.

The uncomfortable bit

The first GPS satellite launched in 1977 with the relativity correction built in but switched off, so the prediction could be tested in orbit. About three weeks of tracking confirmed the offset and the circuit was switched on. It has been on ever since.

Source: Ashby, 'Relativity in the Global Positioning System', Living Reviews in Relativity, 2003

CHANCE · Proven · Richard von Mises, 1939

Birthday Problem

In a room of 23 people the odds of a shared birthday cross fifty percent.

In plain words

Put 23 people in a room. There is a better than even chance that two of them share a birthday. That feels far too few, because there are 365 days. The catch is that you are not asking whether someone matches you. You are asking whether any two people match each other. With 23 people there are 253 different pairs, and each pair gets its own chance. All those small chances add up quickly. The puzzle shows how bad our gut is at counting pairs. We think about ourselves, while the maths thinks about everyone at once.

An everyday example

Try it in class tomorrow. Ask everyone to write their birthday on a slip and compare. In a class of 30 you will find a match about seven times out of ten. Try it at a wedding or a cricket dinner too. Two full cricket teams and one umpire make 23, so a shared birthday on the pitch is a coin flip.

The short version

Your brain counts the 22 people who could match you. The maths counts every possible pair, which is 253. Once you count pairs instead of people, the number stops being shocking.

How it actually works

The calculation takes the people one at a time and asks that each newcomer avoid every birthday already taken. The second person has 364 free dates out of 365, the third 363, and so on. The chance that all 23 miss each other is the product of those fractions, 0.493, so the chance of at least one match is 0.507. The pair count gives the same answer from the other side: each of the 253 pairs fails to match with probability 364 in 365, and that fraction raised to the power 253 is 0.4995. Intuition answers a different question. The chance that one of 22 others shares your particular birthday is 5.9 percent, and pushing that specific match past even odds takes 253 people, not 23.

The 2014 football World Cup made a public test. Each of the 32 squads had exactly 23 players, so the theory predicted a shared birthday in about 16 squads. The BBC checked the lists and found 16 with at least one pair of players born on the same day of the year. Larger groups climb quickly: 30 people give 70.6 percent, 40 give 89.1, 50 give 97.0, and 57 is the first size to pass 99 percent. The same arithmetic drives the birthday attack in cryptography: a hash function with 2 to the power 64 possible outputs yields a collision after roughly 2 to the power 32 random attempts, about four billion, because the attacker is counting pairs.

The problem is usually credited to Richard von Mises, who treated it in a 1939 paper in the journal of the Faculty of Sciences of Istanbul University. Rouse Ball and Coxeter's Mathematical Recreations and Essays records that Harold Davenport had worked it out in about 1927 and did not publish, assuming it was already known. William Feller included it in An Introduction to Probability Theory and Its Applications in 1950. The cryptographic use dates from Gideon Yuval's 1979 paper How to Swindle Rabin, in Cryptologia, which used the pair count to forge signatures against a scheme by Michael Rabin.

Where you have seen it

A normal Indian classroom of 40 students has an 89 percent chance of a shared birthday. In a class of 60 it is 99.4 percent.

The uncomfortable bit

Even at 70 people you are at 99.9 percent, not 100. Certainty needs 367 people, and that gap between almost sure and sure is where most bad betting happens.

Source: von Mises, 'Uber Aufteilungs- und Besetzungswahrscheinlichkeiten', 1939

This entry has a playable demo when scripts are on.

CHANCE · Proven · Steve Selvin, 1975

Monty Hall Problem

Switch doors and you double your chances. Thousands of mathematicians insisted otherwise.

In plain words

There are three doors. Behind one is a car. Behind the other two are goats. You pick a door. The host knows where the car is, and he opens one of the other doors to show you a goat. Then he asks if you want to keep your door or swap to the last closed one. Most people think it makes no difference. It does. Swapping wins two times out of three, because your first pick was probably wrong, and the host just showed you where the car is not.

An everyday example

Play it with a friend, three cups and a coin. Your friend hides the coin under a cup. You point at one. Your friend lifts an empty cup that you did not point at. Now decide: keep or swap. Do twenty rounds keeping and twenty rounds swapping and count the wins. The swapping rounds win about twice as often.

The short version

Three doors, one car. You pick one. The host, who knows what is behind each door, opens a different door showing a goat and offers you the swap. Staying wins one third of the time. Switching wins two thirds.

How it actually works

The trap is the moment after the host opens a door. Two doors remain, so it looks like a coin toss. But the host is not free. He can never open your door and he can never open the car, so his choice carries information. Your first pick is right one time in three, and in those games switching loses. It is wrong two times in three, and in every one of those games the host has exactly one goat door he is allowed to open. The door he leaves shut is the car. Switching wins exactly when your first guess was wrong, which is two thirds of the time.

The hundred-door version cures most people. Pick one door out of a hundred. The host, knowing where the car is, opens ninety-eight goat doors and leaves one closed. Your original door still has its one percent. The other ninety-nine percent has nowhere to go but the door he chose not to open. Nobody feels fifty fifty with a hundred doors, and the three-door game is the same game with smaller numbers.

Steve Selvin posed the problem in a 1975 letter to The American Statistician and named it after the host of the game show Let's Make a Deal. It became famous in September 1990 when Marilyn vos Savant gave the correct answer in Parade magazine and received around ten thousand letters, close to a thousand of them from people with doctorates, most insisting she was wrong. Paul Erdős, by the account of the mathematician Andrew Vázsonyi, accepted it only after watching a computer simulation. The result depends on the rules: the host must always open a goat door and always offer the swap. If he opened a door at random and it happened to show a goat, the two remaining doors really would be even.

Where you have seen it

When a magazine columnist published the correct answer in 1990, she received around ten thousand letters disagreeing, many from people with doctorates.

In films, books and games

In the film 21 (2008) a maths professor puts the three doors to his class, and the student who gets it right is recruited to his card-counting team. In the sitcom Brooklyn Nine-Nine, the episode Skyfire Cycle (2016) has Captain Holt and his husband Kevin fall out over the answer.

The uncomfortable bit

The whole thing hinges on the host knowing where the car is. If he opened a door at random, switching gains you nothing.

Source: Selvin, 'A Problem in Probability', The American Statistician, 1975; vos Savant, Parade, 1990

This entry has a playable demo when scripts are on.

CHANCE · Proven · Edward Simpson, 1951

Simpson's Paradox

A treatment can win in every single group and still lose overall.

In plain words

Sometimes numbers tell one story when you look at groups, and the opposite story when you add the groups together. A hospital might have better results than another for mild cases and better results for serious cases, yet worse results overall. That happens when it takes far more of the serious cases, which pull down its average. Both sets of numbers are correct. Neither is a lie. The picture flips because a hidden factor, how many of each kind of case each side got, was left out. Whenever two people argue with the same numbers, this is often why.

An everyday example

Two batsmen. At home Rohan averages 50 and Sameer averages 45. Away, Rohan averages 30 and Sameer averages 25. Rohan wins both. But Rohan played mostly away and Sameer mostly at home, so Sameer's overall average comes out higher. Check it: give Rohan 2 home and 8 away innings, Sameer 8 home and 2 away, and work out the totals.

The short version

Split the data one way and A beats B. Combine it and B beats A. Both results are arithmetically correct. The reversal comes from a hidden variable sitting behind how the groups were sized.

How it actually works

The reversal needs two conditions at once. The groups must differ in their success rates, and the two sides being compared must be spread unevenly across those groups. When both hold, each side's overall rate is a weighted average, with the weights set by where its members went rather than by how well they did. A side concentrated in the hard groups inherits their low rate even while beating its rival inside every one of them.

In autumn 1973 Berkeley admitted 44 percent of the 8,442 men who applied to graduate study and 35 percent of the 4,321 women. Peter Bickel, Eugene Hammel and J. William O'Connell examined the figures department by department. In the six largest, the admission rates were: A, 62 percent of 825 men and 82 percent of 108 women; B, 63 percent of 560 men and 68 percent of 25 women; C, 37 percent of 325 men and 34 percent of 593 women; D, 33 percent of 417 men and 35 percent of 375 women; E, 28 percent of 191 men and 24 percent of 393 women; F, 6 percent of 373 men and 7 percent of 341 women. Women led in four of the six. Pooled, those departments admitted 45 percent of men and 30 percent of women, because 93 percent of the women had applied to C, D, E and F while half of the men had applied to A and B.

George Udny Yule set out the arithmetic of pooling dissimilar groups in a 1903 Biometrika paper. Edward H. Simpson published The Interpretation of Interaction in Contingency Tables in the Journal of the Royal Statistical Society, Series B, in 1951; Colin Blyth attached Simpson's name to the effect in 1972. A medical instance in the British Medical Journal in 1986, by Charig and colleagues, compared open surgery with keyhole removal of kidney stones: open surgery succeeded more often on small stones, 93 against 87 percent, and on large stones, 73 against 69 percent, yet less often overall, 78 against 83 percent, because it had been reserved for the larger stones.

Where you have seen it

Berkeley's 1973 admissions data looked biased against women until it was read department by department, where the picture flipped. Women had applied in larger numbers to the departments that rejected almost everybody.

The uncomfortable bit

Two honest people with the exact same dataset can reach opposite conclusions and neither is lying. Ask how the data was sliced before you argue about what it means.

Source: Simpson, JRSS Series B, 1951; Bickel, Hammel and O'Connell, Science 187, 1975 (the Berkeley data)

CHANCE · Proven · Simon Newcomb, 1881

Benford's Law

In real world numbers, the first digit is a 1 about thirty percent of the time.

In plain words

Take a big pile of real numbers, like the populations of towns or the amounts on shop bills. Look only at the first digit of each. You would expect 1 to 9 to show up equally. They do not. About three in ten numbers start with a 1, and only about one in twenty starts with a 9. The reason is that things grow by multiplying. A town going from 1,000 to 2,000 people has to double, which takes a long time. Going from 9,000 to 10,000 is a small step. So numbers spend much more of their lives starting with 1.

An everyday example

Open a newspaper and note the first digit of every number on the sports and business pages. Keep a tally from 1 to 9. After fifty numbers the 1 column will be clearly tallest and the 9 column shortest. Try it again with the lengths of rivers in an atlas and the same slope appears.

The short version

River lengths, share prices, populations, electricity bills. The leading digit is not evenly spread. 1 leads about 30 percent of the time and 9 only about 5 percent, because quantities grow multiplicatively, not in equal steps.

How it actually works

The law states that the leading digit d appears with probability equal to the base ten logarithm of one plus one over d: 30.1 percent for 1, 17.6 for 2, 12.5 for 3, falling to 4.6 for 9. Quantities that grow by percentages are spread roughly evenly on a logarithmic scale, where the stretch from 1 to 2 covers 30.1 percent of each decade while the stretch from 9 to 10 covers 4.6 percent. A town of 1,000 people growing at 3 percent a year takes about 23 years to reach 2,000 and under 4 years to go from 9,000 to 10,000, so a census at a random moment is far more likely to catch it with a leading 1.

Benford's 1938 paper tallied 20,229 numbers from twenty sources, among them river drainage areas, city populations, physical constants, street addresses and baseball statistics. Overall the digit 1 led 30.6 percent of the time and the digit 9 4.7 percent, against the predicted 30.1 and 4.6. The powers of two show the pattern: the first ten, from 1 to 512, begin with 1 three times, and of the first thousand exactly 301 do, because each doubling adds the same fixed amount, 0.301, to the logarithm.

Simon Newcomb, the astronomer, published the observation in the American Journal of Mathematics in 1881, after noticing that the early pages of shared logarithm tables were dirtier than the later pages, and he stated the logarithmic formula. Frank Benford, a physicist at General Electric, rediscovered it and published The Law of Anomalous Numbers in the Proceedings of the American Philosophical Society in 1938. Mark Nigrini, in a 1992 doctoral thesis at the University of Cincinnati, proposed the digit frequencies as a test for tax evasion, and his 2012 book Benford's Law: Applications for Forensic Accounting, Auditing, and Fraud Detection is the standard audit manual. In 2011 Bernhard Rauch and colleagues ran the test on the deficit and debt figures that European Union governments had reported to Eurostat from 1999 to 2009; Greece's deviated most.

Where you have seen it

Forensic accountants and election auditors run this test first because it is cheap and it flags the books worth opening.

The uncomfortable bit

Fraudsters spread their digits evenly because even feels random to us. Reality is lumpy, and that is exactly how they get caught.

Source: Newcomb, American Journal of Mathematics, 1881; Benford, Proceedings of the American Philosophical Society, 1938

CHANCE · Proven · Named after Monte Carlo, 1913

Gambler's Fallacy

The coin has no memory. Your gut refuses to believe this.

In plain words

Toss a coin and get heads five times in a row. Most people feel that tails is now more likely. It is not. The coin does not know what happened before. Each toss is a fresh fifty fifty. The mistake is to think chance has to even things out in the short run. It does not. In the long run the numbers balance, but only because new tosses swamp the old ones, not because any toss is pushed the other way. This wrong feeling is called the gambler's fallacy, and it costs people real money every day.

An everyday example

Play snakes and ladders with a friend. After three sixes in a row, the friend groans that the luck is used up. It is not. The dice has no memory. Keep a tally over a hundred rolls and you will see runs of the same number that look impossible but are perfectly normal. Lottery players who avoid last week's numbers make the same mistake.

The short version

After a run of one outcome we feel the other is due. Independent events carry nothing forward. The next toss is fifty fifty whether the last ten were heads or a hundred were.

Where you have seen it

At a Monte Carlo table in 1913 the roulette ball landed on black 26 times in a row. Players lost fortunes piling onto red, convinced the streak had to break.

In films, books and games

Tom Stoppard's play Rosencrantz and Guildenstern Are Dead (1966, filmed 1990) opens with a coin coming up heads more than ninety times running while the two men argue about what that means for the next toss. Dostoevsky's The Gambler (1866) is the fallacy lived out at the roulette table.

The uncomfortable bit

The inverse ruins people too. In a genuinely random sequence, long streaks are expected. Their absence is the actual red flag.

Source: Tversky and Kahneman, 'Belief in the Law of Small Numbers', Psychological Bulletin, 1971

This entry has a playable demo when scripts are on.

CHANCE · Proven · Abraham Wald, 1943

Survivorship Bias

Armour the parts of the plane with no bullet holes.

In plain words

When we study only the things that made it, we miss the things that did not, and that can teach us the wrong lesson. Planes that came home from a battle showed bullet holes in the wings and tail. The tempting answer was to armour those spots. The right answer was the opposite. Those planes flew home with those holes, so those spots could take a hit. The planes hit in the engine never came back to be counted. The pattern we see is shaped by the ones that survived, so it hides the real danger.

An everyday example

A famous cricketer says he never went to coaching classes, so coaching is useless. Nobody interviews the thousand children who skipped coaching and never got picked. Your grandmother's steel tiffin box has lasted fifty years, so things were built better then. The flimsy ones from that era broke long ago and went to the scrap dealer.

The short version

Wartime analysts studied returning bombers and wanted to reinforce where the damage clustered. A statistician pointed out the obvious inversion. Those were the survivable hits. The planes shot in the engines never came back to be measured.

How it actually works

The bias arises whenever a selection step removes cases before they are counted and the removal is not random. Whatever caused a case to drop out is absent from the sample, and any measurement on the remainder is silently conditional on having survived. Wald assumed that enemy fire struck aircraft at points spread roughly in proportion to surface area, so the pattern of hits on the aircraft that took off was known even though only the returning ones could be inspected. Where the returning aircraft showed fewer hits on a section than its area predicted, the shortfall measured the hits that had brought aircraft down. The sections to armour are those with the largest deficits, not the largest counts.

On a rough count, if the engine section is a fifth of a bomber's exposed surface and holds only a tenth of the holes found on the aircraft that came home, then about half of all engine hits were on aircraft that did not. A 1987 study in the Journal of the American Veterinary Medical Association by Wayne Whitney and Cheryl Mehlhaff recorded 132 cats brought to a New York clinic after falling from buildings; 90 percent survived, and the injury rate fell for falls above seven storeys. The finding was read as showing that cats relax at terminal velocity, but the sample held only cats owners thought worth taking to a vet; cats killed outright were never counted.

Abraham Wald left Vienna for the United States in 1938. From 1942 he worked in the Statistical Research Group at Columbia University, a wartime unit led by W. Allen Wallis. His aircraft work was written up in 1943 as a series of memoranda titled A Method of Estimating Plane Vulnerability Based on Damage of Survivors, classified until the Center for Naval Analyses reissued them in 1980. The memoranda contain equations and tables; the silhouette peppered with red dots that circulates online is a later illustration. Wald died in December 1950 when the aircraft carrying him on a lecture tour of India crashed in the Nilgiri hills.

Where you have seen it

Every college dropout billionaire story, every ancient building that proves they built better back then, every fund that only shows you its surviving schemes.

The uncomfortable bit

The data you can easily collect is almost never the data you need. Ask what did not make it into the room.

Source: Wald, 'A Method of Estimating Plane Vulnerability Based on Damage of Survivors', Statistical Research Group, 1943

LOGIC · Open question · Plutarch, first century

Ship of Theseus

Replace every plank, one at a time. Is it still the same ship?

In plain words

An old wooden ship is kept in a harbour. Each year a rotten plank is taken out and a new one is fitted. After many years, not one original piece of wood is left. People then ask whether it is still the same ship. Every single change was small, and after each one everyone agreed it was the same ship. Yet nothing of the original remains. The puzzle is about what makes a thing itself. It might be the material, the shape, the name, or the unbroken story. People have argued about this for two thousand years and still do not agree.

An everyday example

Your school cricket team has had every player replaced over ten years, but it plays under the same name and people speak of its proud history. Your old bicycle has a new chain, new tyres, a new seat and a new frame. Deciding whether it is still your old bicycle is harder than it sounds. Try naming the exact moment it stopped being the same one.

The short version

A ship is preserved by swapping rotten planks for new ones until no original wood remains. Every single step preserved the ship. The end result shares no material with the start.

How it actually works

The puzzle works because the word same bundles several tests that normally agree. One is material: the same object is made of the same stuff. Another is continuity: an object persists if it changes gradually while keeping its form and function, with no moment at which it stops and something else starts. Plank by plank replacement satisfies continuity and breaks material. Hobbes's second ship, built from the discarded planks, satisfies material and breaks continuity. No step in either process is a plausible place to say the ship ceased to exist, so the tests are pulled apart and no further fact arbitrates between them.

The Ise Grand Shrine in Japan has been rebuilt from new timber on an adjacent plot every twenty years since the reign of Empress Jito in the seventh century; the 62nd rebuilding was completed in 2013. Human tissue turns over at rates measured by Ron Sender and Ron Milo in Nature Medicine in 2021: about 330 billion cells a day out of roughly 37 trillion. The replacement is not total. Carbon-14 dating of DNA by Jonas Frisen's group at the Karolinska Institute, published in Cell in 2005, showed that the neurons of the cerebral cortex are as old as the person, and the proteins of the eye's lens are never renewed.

Plutarch set the case down in his Life of Theseus, written around 75 CE. He records that the Athenians kept the thirty-oared ship in which Theseus was said to have returned from Crete, replacing decayed timbers as needed, until the time of Demetrius of Phalerum at the end of the fourth century BC, and that it had become a stock example among philosophers 'for the logical question of things that grow'. Thomas Hobbes added the second ship in De Corpore in 1655. David Wiggins's Sameness and Substance in 1980 argued that identity must be judged under a sortal such as ship, and Derek Parfit's Reasons and Persons in 1984 applied the puzzle to persons and concluded that identity is not what matters in survival.

Where you have seen it

Almost every cell in your body has been replaced multiple times. You still answer to the same name.

In films, books and games

The Marvel series WandaVision (2021) has two Visions argue the puzzle out loud in its final episode. Only Fools and Horses gave it a famous form in the episode Heroes and Villains (1996): Trigger's broom, which has had seventeen new heads and fourteen new handles.

The uncomfortable bit

Now build a second ship from all the discarded original planks. Two ships. Which one is the real one, and what exactly are you pointing at when you say "same"?

Source: Plutarch, Life of Theseus, c. 75 CE; Hobbes, De Corpore, 1655 (the second ship)

LOGIC · Open question · Eubulides, fourth century BCE

Liar Paradox

This sentence is false. If it is true, it is false.

In plain words

Read this sentence: this sentence is false. Now try to decide whether it is true. If it is true, then what it says is right, so it is false. If it is false, then what it says is wrong, so it must be true. You can go round forever and never land anywhere. The sentence talks about itself, and that is what breaks it. The same trap appears any time a rule or a statement has to judge itself. Thinkers have tried for centuries to fix it, and every fix so far creates a new problem somewhere else.

An everyday example

Write on a card: the sentence on the other side is true. On the back write: the sentence on the other side is false. Hand it to a friend and ask which side is right. They will flip it over and over. Or a simpler version: a friend says I am lying right now. Ask them if that was a lie.

The short version

A statement that refers to its own truth value cannot settle into either state. Two and a half thousand years of philosophy have produced patches, hierarchies and truth value gaps, but no clean fix everybody accepts.

Where you have seen it

Anyone who says they always lie. Any self referential rule in a legal document. Any system asked to judge itself.

In films, books and games

In the Star Trek episode I, Mudd (1967) Kirk and Harry Mudd crash an android with it: everything Mudd says is a lie, and Mudd says he is lying. In Portal 2 (2011) GLaDOS tries it on Wheatley, and it fails because he is too simple to notice that there is a paradox.

The uncomfortable bit

This is not a word game. The same self reference sits at the heart of Godel's incompleteness theorems and the halting problem, and it puts hard limits on what maths and computers can ever do.

Source: Attributed to Eubulides, 4th century BCE; Tarski, 'The Concept of Truth in Formalized Languages', 1933

LOGIC · Resolved · Zeno of Elea, 450 BCE

Zeno's Achilles

The fastest runner can never catch the slowest, so motion is impossible.

In plain words

Achilles is fast. A tortoise is slow. Give the tortoise a head start and race. By the time Achilles gets to where the tortoise started, it has crawled a bit further. He runs to that new spot. It has moved again. This goes on forever, so it seems Achilles can never catch up. Of course he does, and easily. The trick is that the gaps get smaller and smaller, and the time for each gap shrinks too. Adding up endlessly many shrinking pieces can give a small, ordinary total. Zeno was right that there are endless steps. He was wrong that endless steps take endless time.

An everyday example

Stand at one end of a room. Walk halfway to the wall. Then walk half the distance that is left. Then half of that. Each step is a real step, and there are endless steps, yet you reach the wall in a few seconds because the steps shrink so fast. Try it and see how quickly the halves vanish.

The short version

Give the tortoise a head start. By the time Achilles reaches its position, it has moved on a little. Repeat forever. Infinitely many gaps means he never passes it, which is obviously wrong, and pinning down exactly why took two thousand years.

Where you have seen it

The same shape as a loading bar that goes 90 percent, 99 percent, 99.9 percent and feels like it will never finish.

In films, books and games

Tom Stoppard's play Jumpers (1972) has its philosopher hero try to act out the race with a real tortoise and a hare.

The uncomfortable bit

Infinitely many steps can add up to a finite time. Calculus solved it. But Zeno's real question, whether space and time can be split forever, is still live in modern physics.

Source: Aristotle, Physics VI:9, c. 350 BCE, our main source for Zeno; Salmon, Zeno's Paradoxes, 1970

LOGIC · Proven · David Hilbert, 1924

Hilbert's Hotel

A hotel with infinite rooms, all full, can always take one more guest.

In plain words

A hotel has endless rooms, numbered 1, 2, 3 and on forever, and every room is taken. A new guest arrives. In a normal full hotel there is no space. Here the manager asks each guest to move up one room. The guest in 1 goes to 2, 2 goes to 3, and so on. Nobody is left out, because there is no last room, and room 1 is now empty. The hotel was full and it still fitted someone in. The lesson is that endless does not behave like large. Adding to it, or even doubling it, leaves it exactly the same size.

An everyday example

Line your friends up in a queue that never ends. To let a new friend in at the front, everyone takes one step back. Nobody falls off the end, because there is no end. At home, write the counting numbers 1, 2, 3 in one column and the even numbers 2, 4, 6 next to them. Every row pairs up, so there are just as many evens as there are numbers.

The short version

Ask every guest to shift one room down. Room 1 is now free. An infinite bus arrives, so move everyone to double their room number and all the odd rooms open up. Full and infinite are not the same word.

How it actually works

The hotel works because an infinite set can be matched one to one with a proper part of itself. Richard Dedekind made that property the definition of infinite in 1888. A set is countable when its members can be listed in order. Georg Cantor proved in 1874, in Crelle's Journal, that the algebraic numbers are countable but the real numbers are not: any listing of the reals leaves some real out. In 1891 he gave the diagonal argument. Write any list of infinite decimals, then build a new one whose nth digit differs from the nth digit of the nth entry. That number is on no line of the list, so the reals outnumber the counting numbers.

When infinitely many buses each carry infinitely many passengers, the fundamental theorem of arithmetic supplies the room plan. Send the existing guest in room n to room 2 to the power n. Send seat m on bus 1 to 3 to the power m, seat m on bus 2 to 5 to the power m, and so on through the primes. Seat 4 on bus 1 gets room 81, seat 2 on bus 2 gets room 25. Every arrival has a distinct room because every whole number has one prime factorisation. A bus with one passenger for every real number between 0 and 1 cannot be housed by any plan, because the room numbers form a countable set.

The hotel appears in Hilbert's Göttingen lecture course of 1924 titled Über das Unendliche, whose notes were published by Springer in 2013. George Gamow's One Two Three... Infinity of 1947 made the story famous, as Helge Kragh traced in a 2014 paper. Cantor's 1874 paper ran in Crelle's Journal, volume 77, and the 1891 diagonal paper in the first volume of the Jahresbericht of the German Mathematical Society. Whether an infinity sits between the counting numbers and the reals, Cantor's continuum hypothesis of 1878, was shown independent of the standard axioms by Kurt Gödel in 1940 and Paul Cohen in 1963.

Where you have seen it

The reason there are exactly as many even numbers as there are whole numbers, even though one is a subset of the other.

The uncomfortable bit

Not all infinities are equal. The infinity of decimal numbers is strictly larger than the infinity of counting numbers, and no clever rearrangement fits them into the hotel.

Source: Hilbert lecture 'Uber das Unendliche', 1924; popularised in Gamow, One Two Three... Infinity, 1947

LOGIC · Proven · Lewis Fry Richardson, 1951

Coastline Paradox

A country's coastline has no fixed length. It depends on the size of your ruler.

In plain words

Ask how long a coastline is and the honest answer is: it depends on how closely you look. Measure with a giant ruler and you go straight across the bays. Use a shorter ruler and you follow the bays and get a longer number. Use a shorter one still and you go round every rock, longer again. Unlike a straight road, a coast has wiggles inside wiggles all the way down to the pebbles. The length keeps growing as the ruler shrinks and never settles on a final value. So any coastline figure is really a choice about how much detail to count.

An everyday example

Take a map of Goa and a piece of thread. Lay the thread along the coast, keeping it as straight as you can, and measure it. Now do it again, pushing the thread into every bay and around every headland. The second thread is much longer. Do it a third time on a bigger map and it grows again.

The short version

Measure with a hundred kilometre stick and you skip the bays. Measure with a one metre stick and every inlet and rock adds distance. As the ruler shrinks, the length grows without settling anywhere.

Where you have seen it

India's official coastline jumped from 7,516 km to 11,098 km in a 2024 revision. No land was added. The Survey of India simply remapped it at a finer scale.

The uncomfortable bit

Every border, every map, every measurement quietly hides the ruler someone chose. The number that sounds like a hard fact is a decision.

Source: Richardson, 'The Problem of Contiguity', General Systems Yearbook, 1961; Mandelbrot, 'How Long Is the Coast of Britain?', Science 156, 1967. India figure: Ministry of Home Affairs coastline revision, 2024

This entry has a playable demo when scripts are on.

LOGIC · Proven · Banach and Tarski, 1924

Banach-Tarski Paradox

Cut a solid ball into five pieces and reassemble them into two identical balls.

In plain words

A theorem in maths says you can take a solid ball, split it into five pieces, and move the pieces around, without stretching or squashing, so that they form two solid balls the same size as the first. This sounds like making something from nothing. Nobody can do it with a real ball, and the reason is the pieces. They are not chunks you could cut with a knife. They are scattered sprays of points so wild that they do not have a size at all. Because the pieces have no size, the usual rule that sizes add up does not apply, and the trick goes through.

An everyday example

Think of the counting numbers going on forever. Split them into odds and evens. Each half still goes on forever, and each half can be relabelled 1, 2, 3 and so on. From one endless set you got two full endless sets. The ball puzzle does something like this in three dimensions, using the endless number of points inside a ball.

The short version

Rigid motions only. No stretching, no adding matter. It is a genuine theorem of set theory, and it is completely valid maths.

How it actually works

A solid ball can be split into finitely many pieces which, moved only by rotations and translations, fit together as two balls each the size of the original. The construction starts with two rotations of the sphere, such as rotations by arccos(1/3) about two perpendicular axes, chosen so that no combination of them ever returns to the start. The words in those two rotations form a free group, and a free group can be divided into parts that, after one rotation each, rebuild two whole copies of the group. The axiom of choice is used to select one point from each orbit on the sphere. The pieces are not measurable: volume is never assigned to them, so volume is neither conserved nor broken.

Five pieces suffice, and Raphael Robinson proved in 1947 that four cannot. The result cannot be carried out on matter because the pieces are dense, scattered sets of points, and selecting them needs an uncountable number of arbitrary choices that no cutting procedure can make. A physical ball contains about 10 to the power 23 atoms, a finite number, and any finite set is measurable. The paradox also fails in the plane: Banach showed in 1923 that area can be extended to every subset of the plane in a way rigid motions preserve, because the plane's motion group contains no free subgroup on two generators.

Felix Hausdorff took the first step in 1914 in Grundzüge der Mengenlehre, splitting the sphere minus a countable set into three pieces where each was congruent to the other two combined. Stefan Banach and Alfred Tarski extended it to the solid ball in Fundamenta Mathematicae, volume 6, 1924. John von Neumann identified the source in 1929, defining amenable groups, and showed the paradox arises exactly where a group is not amenable. Robert Solovay proved in 1970 that in a model of set theory without the full axiom of choice every set of reals is measurable. In 1994 Randall Dougherty and Matthew Foreman found a version whose pieces have the Baire property and needs no choice.

Where you have seen it

It has no physical demonstration and never will, which is exactly what makes it interesting.

The uncomfortable bit

The pieces are infinitely detailed clouds of points with no definable volume, and the proof needs the axiom of choice. Some mathematicians treat this result as an argument against that axiom.

Source: Banach and Tarski, 'Sur la decomposition des ensembles de points en parties respectivement congruentes', Fundamenta Mathematicae, 1924

LOGIC · Proven · Emile Borel, 1913

Infinite Monkey Theorem

A monkey typing forever will eventually produce the complete works of Shakespeare.

In plain words

Sit a monkey at a keyboard and let it press keys at random forever. Sooner or later, the theorem says, it will type out any text you name, including all of Shakespeare's plays. This is not because monkeys are clever. It is because random tries that go on without end must eventually hit every possible string of letters. The catch is the word forever. Even a short sentence takes so long to appear by chance that the wait dwarfs the age of the universe. So the idea is true on paper and hopeless in practice. It is really a lesson about how huge the number of possibilities is.

An everyday example

Try a tiny version. Close your eyes and press three letter keys on a phone keyboard, aiming for the word cat. Each press has 26 letters to choose from, so there are more than seventeen thousand possible three letter strings. You would expect thousands of tries for one short word. Now picture that for a whole play.

The short version

Given infinite time and random keystrokes, every finite text appears with probability one. The maths is airtight, which is why the line has survived a century as shorthand for randomness eventually producing order.

Where you have seen it

Quoted in almost every argument about evolution, encryption and large scale randomness.

In films, books and games

The Simpsons episode Last Exit to Springfield (1993) has Mr Burns keep a thousand monkeys at a thousand typewriters, and one types It was the best of times, it was the blurst of times. Douglas Adams' novel The Hitchhiker's Guide to the Galaxy (1979) has a horde of monkeys turn up with a script of Hamlet they want to discuss.

The uncomfortable bit

Two mathematicians ran the numbers in 2024 using the actual monkey population and the projected lifespan of the universe. There is not remotely enough time. The theorem is true and useless in the same breath.

Source: Borel, 'Mecanique Statistique et Irreversibilite', 1913; Woodcock and Falletta, Franklin Open, 2024 (the finite-resources calculation)

COSMOS · Thought experiment · Erwin Schrödinger, 1935

Schrödinger's Cat

Both alive and dead until someone opens the box.

In plain words

A cat is shut in a box with a tiny bit of radioactive stuff. If one atom breaks apart, a poison is released and the cat dies. If not, the cat lives. The rules of very small things say the atom is in both states, broken and not broken, until someone looks. So, by those same rules, the cat is both dead and alive until you open the box. Schrödinger made this up to show how silly that sounds when you apply the rules of atoms to something as big as a cat. Nobody has ever hurt a cat over it.

An everyday example

Think of a coin spinning on a table. While it spins, it is not really heads or tails yet. Only when it falls flat, or you slap your hand on it, does it become one or the other. Atoms act like the spinning coin. The puzzle is why a cat, made of trillions of atoms, never seems to spin like that.

The short version

A cat is sealed in a box with a mechanism triggered by a random quantum event. Until an observation happens, the maths describes the system as a mix of both outcomes rather than one hidden answer.

How it actually works

A tiny amount of radioactive substance has a one in two chance of one atom decaying within an hour. A Geiger counter registers the decay, trips a hammer and breaks a flask of hydrocyanic acid. Quantum theory writes the atom after an hour as a superposition of decayed and undecayed, and because the counter, hammer, flask and cat are physical systems obeying the same linear equations, the superposition passes up the chain: the state of the whole box is a sum of decayed atom with dead cat and intact atom with live cat. The theory supplies no rule saying where along the chain a definite outcome appears.

The modern answer is decoherence. A cat is never isolated: air molecules and thermal photons scatter off it, each scattering records which branch the body is in, and the interference terms between live and dead leak into the environment beyond recovery. Wojciech Zurek estimated in 1991 that a one gram object at room temperature, in a superposition of positions one centimetre apart, would lose its interference in about 10 to the power minus 23 seconds even if its thermal relaxation took the age of the universe. Small cat states have been built: in 1996 David Wineland's group at NIST placed a beryllium ion in two positions 80 nanometres apart, and in 2023 a team at ETH Zurich reported a superposition of a 16 microgram sapphire crystal. Decoherence explains why interference is never seen, not which outcome occurs.

Einstein, Podolsky and Rosen published their argument in Physical Review on 15 May 1935, and Schrödinger and Einstein then corresponded. On 8 August Einstein proposed a keg of gunpowder whose quantum state after a year would be a blend of exploded and unexploded, and the cat is Schrödinger's sharper version. It appeared in his three-part paper in Naturwissenschaften, volume 23, in November and December 1935. The decoherence programme began with H. Dieter Zeh's 1970 paper in Foundations of Physics, was made quantitative by Zurek in Physical Review D in 1981 and 1982, and reached a wide audience through Zurek's 1991 Physics Today article.

Where you have seen it

Shorthand for anything unresolved, usually used by people who think it means "we do not know yet". It means something stranger.

In films, books and games

In The Big Bang Theory, the episode The Tangerine Factor (2008) has Sheldon use the cat to tell Penny that her date with Leonard is both good and bad until she tries it. Terry Pratchett's novel Lords and Ladies (1992) gives the cat Greebo a third state: bloody furious. The film Coherence (2013) builds its whole plot on the idea.

The uncomfortable bit

Schrödinger invented it as mockery. He thought applying quantum rules to a cat exposed how absurd the standard interpretation was. Ninety years later it is the poster the interpretation is famous for.

Source: Schrodinger, 'Die gegenwartige Situation in der Quantenmechanik', Naturwissenschaften, 1935

COSMOS · Proven · Thomas Young, 1801

Double Slit Experiment

Watch which slit the particle goes through and it stops behaving like a wave.

In plain words

Shine light, or fire tiny particles like electrons, at a wall with two narrow slits. On a screen behind, you get a pattern of stripes, the kind of pattern waves make when they overlap. That happens even if you send the particles one at a time, as if each one somehow passes through both slits. Now put a detector at the slits to see which slit each particle uses. The stripes vanish. You get two plain bands instead. Simply checking the path changes the result. This is the clearest sign that the smallest things do not behave like tiny balls.

An everyday example

Drop two pebbles into a still pond at once. The ripples cross and make a pattern of high and low spots. The stripes on the screen are that pattern, made of light or electrons. You can see the wave version at home with a laser pointer and a strand of hair held in the beam: the wall shows bands of light and dark, not one dot.

The short version

Fire electrons one at a time at two slits and an interference pattern builds up, as if each one went through both. Add a detector at the slits and the pattern collapses into two plain bands.

How it actually works

A wave through two openings a distance d apart lands bright where the two paths differ by a whole number of wavelengths. Louis de Broglie proposed in 1924 that a particle of momentum p carries a wavelength of Planck's constant divided by p. In quantum theory the amplitude for reaching a point on the screen is the sum of the amplitudes for each slit, and the probability is the squared size of that sum, which contains a cross term. A detector at the slits correlates the particle with a record of its path; the two amplitudes then belong to distinguishable situations and cannot add, so the cross term vanishes.

Clinton Davisson and Lester Germer, at Bell Labs in 1927, fired electrons of 54 electronvolts at a nickel crystal and found a strong reflection at 50 degrees. Akira Tonomura's group at Hitachi in 1989 sent electrons through an electron biprism at 50 kilovolts, a wavelength near 5.4 picometres, with the source so weak that on average fewer than one electron was in flight along the 1.5 metre path at any moment. Each landed as a single dot, and the fringes emerged after tens of thousands had arrived. In 2019 Yaakov Fein, Markus Arndt and colleagues in Vienna reported in Nature Physics the interference of oligoporphyrin molecules of up to 2,000 atoms and masses above 25,000 daltons, with de Broglie wavelengths near 53 femtometres, in a two metre interferometer.

Thomas Young described his interference work in the Bakerian lecture of 1801, published in 1802, and the split beam experiment with a card in Philosophical Transactions in 1804. George Paget Thomson observed electron diffraction independently in 1927 and shared the 1937 Nobel Prize with Davisson. Claus Jönsson performed the first electron double slit in Tübingen in 1961. Pier Giorgio Merli, Gian Franco Missiroli and Giulio Pozzi in Bologna filmed single-electron arrival in 1976, and Tonomura's 1989 paper appeared in the American Journal of Physics. Markus Arndt, Anton Zeilinger and colleagues first diffracted carbon-60 fullerenes in Nature in 1999.

Where you have seen it

This single experiment is the cleanest evidence that measurement is not a passive act at the smallest scales.

The uncomfortable bit

Consciousness has nothing to do with it. Detection means physical interaction, and interaction is what destroys the interference. The mystical reading of this experiment is the most repeated science myth on the internet.

Source: Young, Philosophical Transactions of the Royal Society, 1804; Tonomura et al., American Journal of Physics, 1989 (single electrons)

COSMOS · Open question · Enrico Fermi, 1950

Fermi Paradox

Hundreds of billions of stars, billions of years of head start. So where is everybody?

In plain words

Our galaxy has hundreds of billions of stars, and many are far older than the Sun. If life and clever creatures are at all common, some should have had a huge head start. Even slow spaceships could spread across the whole galaxy in a few million years, which is a blink against its age. So the sky ought to be full of visitors, signals or leftover machines. Instead we see and hear nothing. That silence is the puzzle. Either life is much rarer than it looks, or something stops civilisations before they spread, or they are out there and keeping quiet.

An everyday example

You move into a huge new housing colony with thousands of flats. The lights are on, but in a whole year you never hear a voice, see a car, or get a knock at the door. You would start asking questions. That is how astronomers feel about the galaxy. Look up on a clear night and count the stars. Each one could be somebody's home.

The short version

Even at slow sub light speeds, a civilisation could cross the galaxy in a few million years, which is nothing against the age of the universe. The galaxy should show signs of company. It looks empty.

How it actually works

The question was asked over lunch at Los Alamos in the summer of 1950. Enrico Fermi, Edward Teller, Emil Konopinski and Herbert York had been joking about a New Yorker cartoon blaming flying saucers for missing city bins, and some time later Fermi asked, out of nowhere, where everybody was. Eric Jones reconstructed the conversation from the three surviving witnesses in a 1985 Los Alamos report. Fermi's own view, as they remembered it, was that interstellar travel is probably impractical. The modern paradox is really Michael Hart's, from a 1975 paper arguing that the absence of aliens on Earth is strong evidence that there are none in the galaxy, sharpened by Frank Tipler in 1980.

The arithmetic is what makes it bite. The Milky Way holds somewhere between one hundred and four hundred billion stars, and results from the Kepler telescope suggest that rocky planets in temperate orbits are common rather than rare. The galaxy is about thirteen billion years old and the Earth four and a half. A species with ships at one percent of light speed, no exotic physics needed, could cross the hundred thousand light years of the galactic disc in ten million years. With probes that build copies of themselves from asteroid material, Tipler's estimate for filling the galaxy was a few tens of millions of years. Against thirteen billion years, that is an afternoon. One expansionist species, ever, and the galaxy would be full.

The proposed answers sort into a few families. Life or intelligence is far rarer than the star counts suggest, the Rare Earth argument of Ward and Brownlee in 2000. Something stops civilisations before they spread, Robin Hanson's Great Filter of 1998, which is only comforting if the filter is behind us. They are here and deliberately quiet, John Ball's zoo hypothesis of 1973. Or there is no paradox. In 2018 Sandberg, Drexler and Ord reran the Drake equation with honest uncertainty on every factor instead of single point estimates, and found a large probability that we are alone in the observable universe. On that reading, an empty sky is not surprising at all.

Where you have seen it

Every serious argument about SETI, about colonising Mars, and about how long technological civilisations last.

In films, books and games

The film Contact (1997), from Carl Sagan's 1985 novel, is the hopeful side: the first signal arrives. Liu Cixin's Three-Body Problem novels (from 2008) and the Netflix series 3 Body Problem (2024) give the darkest answer: everyone stays silent because speaking gets you killed.

The uncomfortable bit

The Great Filter is the grim version. Some step between dead chemistry and a galactic civilisation is almost impossible. If it is behind us we got absurdly lucky. If it is ahead of us, that is worse news.

Source: Hart, Quarterly Journal of the Royal Astronomical Society, 1975; Fermi's 1950 remark recorded in Jones, Los Alamos LA-10311-MS, 1985

COSMOS · Resolved · Heinrich Olbers, 1823

Olbers' Paradox

In an infinite eternal universe the night sky should be blindingly bright.

In plain words

If the universe went on forever, had always existed, and was filled with stars, then every line you drew from your eye out into space would end on a star. Near or far, some star would sit in every direction. The whole night sky would glow as bright as the Sun. It plainly does not. It is dark with dots. So one of the ifs must be wrong. The answer is that the universe is not endlessly old. It began about 13.8 billion years ago, so light from beyond that distance has not reached us, and the stretching of space dims much of what has.

An everyday example

Stand in a thick old forest and look sideways. Every line of sight ends on a tree trunk, so you see a wall of wood. Now stand in a forest planted only last year, with thin saplings. You can see right through it. The night sky is like the young forest: the stars are there, but the far ones have not had time to send their light.

The short version

Look in any direction and your line of sight should eventually hit a star. Every point of the sky should burn like the surface of the sun. It does not, and the darkness demands an explanation.

Where you have seen it

One of the rare cases where the answer is visible from any rooftop, if you know what the darkness means.

In films, books and games

Edgar Allan Poe's prose poem Eureka (1848) guessed the answer decades early: the far stars look dark because their light has not had time to reach us.

The uncomfortable bit

The night is dark because the universe had a beginning and is expanding. Light from the most distant stars has not arrived yet, and much of what has arrived is stretched beyond what our eyes can see. Darkness is the evidence.

Source: Olbers, 1823; Harrison, Darkness at Night, Harvard University Press, 1987

COSMOS · Thought experiment · Nick Bostrom, 2003

Simulation Hypothesis

If civilisations can simulate minds, almost every mind that exists is simulated.

In plain words

Computer games already show worlds full of characters. Suppose that one day a civilisation could run a computer so powerful that the characters inside it think and feel. If that ever becomes possible and is done many times, then simulated minds would outnumber real ones by a huge margin. And if most minds that ever exist are simulated, the odds say you are probably one of them. The argument does not claim we live in a simulation. It says one of three things must be true: nobody ever reaches that power, nobody who has it uses it, or this is probably not the base world.

An everyday example

Think of a game like Minecraft. The villagers inside have no way of knowing there is a screen and a player. Now picture a future game where the villagers can think. If a million copies of that game are running and there is only one real world, a thinking villager who bets it is in the real world is almost always wrong.

The short version

Bostrom's argument is a fork, not a claim. Either civilisations reliably die before that power, or they get it and choose not to use it, or we are almost certainly inside a simulation. At least one of the three must be true.

Where you have seen it

Endlessly repeated as though it were a prediction. It is a trilemma, and the first two branches are just as interesting.

In films, books and games

The Matrix (1999) is the obvious one: humanity lives inside a machine-built world. Free Guy (2021) follows a game character who works out what he is. The Rick and Morty episode M. Night Shaym-Aliens! (2014) traps Rick in a simulation inside a simulation. Nick Bostrom published the argument itself in 2003.

The uncomfortable bit

It is nearly unfalsifiable, which is why most physicists file it under philosophy. Ask what evidence would change your mind. If nothing would, it is not a scientific claim.

Source: Bostrom, 'Are You Living in a Computer Simulation?', Philosophical Quarterly, 2003

COSMOS · Open question · Ludwig Boltzmann, 1890s

Boltzmann Brains

It is easier for a universe to accidentally assemble one thinking brain than a whole cosmos.

In plain words

Given a very long time, the random jiggling of particles can throw up almost anything, even a working brain floating in empty space, complete with made-up memories of a life it never lived. That sounds absurd, but a lone brain is a far smaller thing than a whole galaxy. Building a small thing by chance is much likelier than building a big one. So in some models of the universe, chance brains should outnumber real people. Physicists do not believe in these brains. They use them as an alarm: if a theory predicts too many of them, the theory is probably wrong.

An everyday example

Shake a box of Lego for a billion years. Once in a while two bricks click together. Very rarely a small toy car forms by itself. A full Lego city forming by accident is far, far rarer than the car. The brain is the car. The universe is the city. If your idea of how the box shakes predicts a sky full of accidental cars, and you see none, doubt the idea.

The short version

Given enough time, random fluctuations can produce anything, including a conscious brain complete with false memories, floating alone in space. Such a brain is vastly less improbable than an entire ordered universe.

Where you have seen it

Cosmologists use it as a stress test. If your model of the universe predicts more of these brains than real observers, your model has a problem.

The uncomfortable bit

This is not a spooky story. It is a working tool, and several otherwise reasonable cosmological models have been rejected because they fail it.

Source: Boltzmann, Nature, 1895; Albrecht and Sorbo, Physical Review D, 2004

COSMOS · Open question · Donald Kessler, 1978

Kessler Syndrome

One collision in orbit sets off the next, until low Earth orbit closes for a century.

In plain words

Space around Earth is getting crowded. Thousands of satellites circle up there, along with old rocket parts and broken bits. Everything moves incredibly fast, many times faster than a bullet. If two objects crash, they do not just stop. They shatter into thousands of new pieces, each still racing along. Those pieces can hit other satellites, making even more pieces. Past a certain point the crashes feed themselves and never stop. Then a belt of flying junk would wrap the planet, and launching anything through it would be too dangerous for decades. The worry is that we might cross that point without noticing.

An everyday example

Picture a busy roundabout with no brakes on any car. One bump sends two cars spinning into four more, and soon the whole roundabout is a pile-up that keeps growing. Now picture the cars never slowing down, for years. That is orbit. On a clear night, look for a steady dot that crosses the whole sky in a few minutes. That is a satellite, and thousands more are up there with it.

The short version

Debris from a smash creates more debris, each fragment travelling at several kilometres per second. Past a certain density the cascade becomes self sustaining, and a shell of shrapnel makes launches unviable.

Where you have seen it

A 2009 satellite collision and several anti satellite weapon tests have already added thousands of trackable fragments.

In films, books and games

The film Gravity (2013) is the syndrome as a disaster movie: a destroyed satellite sends a wave of debris around the planet every ninety minutes. The WALL-E (2008) opening shows Earth wrapped in a shell of old satellites. The anime Planetes (2003) follows a crew whose whole job is collecting orbital junk.

The uncomfortable bit

A screw travelling at orbital speed hits harder than a bullet. The thing that could lock humanity out of space is not a rival power. It is our own litter.

Source: Kessler and Cour-Palais, 'Collision Frequency of Artificial Satellites', Journal of Geophysical Research, 1978

MIND · Proven · Fiona Broome, 2009

Mandela Effect

Large groups of strangers remember the same thing that never happened.

In plain words

Memory feels like a video you play back. It is not. Every time you remember something, your brain rebuilds it from bits and fills the gaps with whatever fits. If the same gap gets filled the same way by many people, a false memory can spread across a whole crowd. The name comes from many people clearly remembering news of Nelson Mandela's death in prison, which never happened. He was released in 1990 and died in 2013. The effect is not proof of parallel worlds. It is proof that memory is a story we retell, not a recording.

An everyday example

Ask ten friends to draw Pikachu's tail from memory. Many will add a black tip that is not there. Ask them what Darth Vader says in The Empire Strikes Back (1980). Most will quote Luke, I am your father. The real line is No, I am your father. Then play the scene and watch their faces.

The short version

Named after a widespread false memory of Nelson Mandela dying in prison in the 1980s. Memory is not playback. It is reconstruction, and it fills gaps with what feels right, so plausible errors spread and stabilise.

How it actually works

Frederic Bartlett showed in 1932 that memory works by reconstruction: people who retold an unfamiliar folk tale shortened it and bent its details toward the familiar. Elizabeth Loftus turned that into an experimental programme. Her misinformation effect is the finding that information met after an event gets folded into the memory of it, and that people cannot tell the added part from the original. Schemas fill gaps with the typical case, so a name is recalled with the more common spelling and a character gains the accessory his type usually wears. Source confusion then lets a detail from a parody or a forum post be filed as first-hand. People who share a schema make the same error independently, and once it is discussed online it becomes a source that confirms itself.

Loftus and John Palmer showed 45 students films of car crashes in 1974 and asked how fast the cars were going when they smashed, collided, bumped, hit or contacted each other. Mean estimates ran from 40.5 miles per hour for smashed to 31.8 for contacted. A week later 32 per cent of the smashed group said they had seen broken glass, against 14 per cent of the hit group; there was none. Stan and Jan Berenstain published their first book in 1962, but -stein is far commoner as an ending than -stain, so the regular form wins. Deepasri Prasad and Wilma Bainbridge tested 40 well-known images in Psychological Science in 2022 and found seven, including the Monopoly man and Pikachu's tail, misremembered the same way with high confidence.

Fiona Broome, a paranormal writer, found at Dragon Con in Atlanta in 2009 that others shared her memory of Nelson Mandela dying in prison in the 1980s. Mandela was released in February 1990 after 27 years and died in December 2013. Broome launched a website under the name that year; the term spread through Reddit and YouTube from about 2015. Loftus's key papers are Loftus and Palmer in the Journal of Verbal Learning and Verbal Behavior in 1974 and Loftus, Miller and Burns in 1978 on the stop sign experiment.

Where you have seen it

The children's book series spelled Berenstain that half the internet remembers as Berenstein. The Monopoly mascot that most people picture wearing a monocle he never had.

In films, books and games

The Mandela Effect (2019) is a film built around the idea, following a father who becomes convinced reality has been rewritten.

The uncomfortable bit

The internet supercharges it. Once a mistaken version is written down and shared, it becomes the reference other people check their memory against.

Source: Coined by Broome, 2009; Prasad and Bainbridge, 'The Visual Mandela Effect', Psychological Science, 2022

MIND · Contested · Dunning and Kruger, 1999

Dunning Kruger Effect

The least skilled are the most confident, because judging skill needs the skill.

In plain words

People who are bad at something often think they are good at it. They are not showing off. To know how well you did, you need the same knowledge you needed to do the task in the first place. If you do not have it, you cannot see your own mistakes, so you feel fine. People who are very good tend to think everyone finds it easy, so they rate themselves a little low. The surprise is that the worst performers were the most sure of themselves. Later checks found the gap is smaller than the internet claims.

An everyday example

A boy learns three chords on a guitar and tells his class he can play. His cousin has practised for six years and says she is still learning. He cannot hear his own mistakes yet, because hearing them takes the skill he has not built. Try it at school: everyone writes down the mark they expect before a test. Compare guesses with results afterwards.

The short version

In the original study, low performers overestimated their ability while high performers slightly underestimated theirs. The proposed explanation is that incompetence hides itself, since the same knowledge is needed to do a task and to judge it.

How it actually works

The argument of the 1999 paper is that competence and self-assessment draw on the same knowledge. Someone without the rules that produce good grammar, valid proofs or good jokes produces poor work and cannot see that it is poor, because the errors look like correct answers. Kruger and Dunning called this a dual burden. Participants took a test, then estimated their percentile rank in ability and on the test. In the grammar study they later graded five other people's papers and re-estimated: top performers raised their estimates once they saw how badly others had done, bottom performers did not move. Finally, ten minutes of training in logic made bottom-quartile participants better at the task and at judging their own answers.

The participants were Cornell undergraduates: 65 rated 30 jokes against a panel of professional comedians, 45 answered 20 logic questions from a law-school admissions test, 84 took a 20-item grammar test, and 140 the training study. The abstract's headline figure: those in the bottom quartile, whose scores put them on average at the 12th percentile, estimated themselves to be at the 62nd. The top quartile ran the other way, placing themselves roughly ten to fifteen points below their true rank. Nearly everyone guessed between the 55th and 75th percentiles wherever they actually sat, the shape the critics fastened on.

Justin Kruger and David Dunning published Unskilled and Unaware of It in the Journal of Personality and Social Psychology in December 1999. Joachim Krueger and Ross Mueller argued in the same journal in 2002 that most of the pattern is regression to the mean: with any two imperfectly correlated measures, the lowest scorers on one sit closer to the average on the other. Edward Nuhfer and colleagues showed in Numeracy in 2016 and 2017 that random numbers plotted in the paper's format, self-estimate minus score against score, give the same graph, because the score appears on both axes. Gilles Gignac and Marcin Zajenkowski tested 929 people in Intelligence in 2020 and found the effect largely vanished under proper statistics. Dunning's reply is that regression cannot produce the training result.

Where you have seen it

Quoted in every online argument, usually as an insult, usually by someone who has not read the paper.

The uncomfortable bit

Statisticians have shown that a good part of the pattern appears even in random data, because of how the two measurements are compared. The effect is real but far smaller and messier than the meme version.

Source: Kruger and Dunning, Journal of Personality and Social Psychology, 1999; Nuhfer et al., Numeracy, 2017 (the statistical critique)

MIND · Proven · Arnold Zwicky, 2005

Frequency Illusion

You learn a word on Tuesday and it is suddenly everywhere.

In plain words

You notice something new, a word, a song, a type of car. Then it seems to pop up all the time. It feels like the world has changed. It has not. Those things were always there. Before, your brain ignored them because they did not matter to you. Now they matter, so your brain flags every one. Your brain also keeps a tally of the hits and forgets the misses, so the pattern looks stronger than it is. The surprise is how strongly it feels like a sign or a coincidence when it is only your attention changing.

An everyday example

Your friend buys yellow trainers. On the bus the next morning you spot three pairs of yellow trainers. Nobody bought them overnight. Try it at home: pick a boring thing, like red number plates, and decide to count them for one day. By the evening it will feel like the town is full of them.

The short version

Also called the Baader Meinhof phenomenon. Two ordinary mechanisms stack. Selective attention makes you notice the thing, then confirmation bias logs each new sighting as remarkable while ignoring the years you filtered it out.

Where you have seen it

You research a car model and the roads fill with it. You learn a Kannada word and hear it three times that week.

The uncomfortable bit

This is the same machinery behind believing an app is listening to you. Nothing changed in the world. Your filter changed.

Source: Zwicky, Language Log, 2005; the 'Baader-Meinhof' name from a 1994 St. Paul Pioneer Press reader letter

MIND · Contested · Darley and Latane, 1968

Bystander Effect

The more people watching, the less likely anyone steps in.

In plain words

When something bad happens in a crowd, each person thinks someone else will help. Everyone waits for someone else, so nobody moves. There are two reasons. First, the job of helping is split between many people, so each person feels only a small piece of it. Second, people look at each other to work out if it is serious, and everyone looks calm because everyone is waiting. In a lab this shows up clearly. In real life, with cameras watching real fights, people usually do step in. The surprise is that a crowd can make helping slower, but rarely stops it.

An everyday example

A girl drops her books in a crowded school corridor. Thirty pupils walk past. In an empty corridor, the one pupil who walks past almost always stops. If you ever need help in a crowd, point at one person and say, you in the blue shirt, call someone. Giving one person the job breaks the spell.

The short version

Responsibility spreads thin across a crowd, and each person reads the stillness of the others as evidence that nothing is wrong. In controlled lab settings the effect shows up reliably.

Where you have seen it

Taught for fifty years using a 1964 murder in New York where 38 witnesses supposedly did nothing. Later investigation showed that account was badly wrong. Several people did act.

In films, books and games

The last episode of Seinfeld, The Finale (1998), sends the four friends to jail after they stand filming a carjacking and crack jokes instead of helping.

The uncomfortable bit

A 2019 study of real CCTV footage of public conflicts found someone intervened in about nine out of ten cases. The lab effect is real. The bleak story built on top of it is not.

Source: Darley and Latane, Journal of Personality and Social Psychology, 1968; Philpot et al., American Psychologist, 2019 (the CCTV study)

MIND · Proven · Lambert and Jakobovits, 1962

Semantic Satiation

Say any word thirty times and it dissolves into noise.

In plain words

Pick any word and say it out loud again and again. After twenty or thirty times, it stops sounding like a word. It turns into a strange noise, and you may wonder if you are even saying it right. This happens because the part of your brain that links the sound to its meaning gets tired when it is used over and over without a break. The sound still comes in, but the meaning switches off for a moment. It wears off within seconds. People are surprised that a word they have known all their lives can go blank.

An everyday example

Try it right now with the word spoon. Say it steadily, once a second, for half a minute. Somewhere around the twentieth time it will feel like a nonsense sound. Then stop, talk about something else for a moment, and say it once more. The meaning comes straight back. It also works with your own name.

The short version

Repeated firing of the same neural pattern briefly exhausts it. The sound survives, the meaning drops out, and for a moment you hear your own language as a foreigner does.

Where you have seen it

Any word you have written so many times in one afternoon that you start doubting the spelling.

The uncomfortable bit

Therapists use it deliberately. Repeating a feared or shaming word until it goes flat is a real technique for draining its charge.

Source: Jakobovits and Lambert, Journal of Experimental Psychology, 1962

MIND · Proven · Walter Kennedy, 1961

Nocebo Effect

The placebo's evil twin. Expect harm and you will feel harm.

In plain words

Most people know about placebo: a sugar pill can make you feel better if you believe it is medicine. Nocebo is the reverse. If you believe something will make you feel ill, it often does, even when the thing itself does nothing. Your body is not faking. Worry can bring on real headaches, sickness and pain. This is why people who are given a harmless pill and a list of warnings often report the exact problems on the list. The surprise is that a warning can cause the very thing it warns about.

An everyday example

Two classmates drink the same lemonade. One is told it is a new energy drink that can give you a headache. An hour later she has a headache and the other does not. Nothing was in the drink except sugar and lemon. Simply reading the side effects on a medicine box before you take it can make you notice every tiny twinge.

The short version

Believing something will hurt you produces measurable symptoms with no active cause. Nausea, headaches, pain, all from an inert pill and a warning.

Where you have seen it

In drug trials, the group taking sugar pills reports the exact side effects listed on the leaflet at surprisingly high rates.

The uncomfortable bit

It puts doctors in a genuine bind. Full disclosure of side effects is an ethical duty, and it demonstrably causes some of those side effects.

Source: Kennedy, 'The Nocebo Reaction', Medical World, 1961; Barsky et al., JAMA, 2002

MIND · Contested · Masahiro Mori, 1970

Uncanny Valley

The closer a face gets to human without arriving, the more it repels us.

In plain words

We like things that look a bit like people. A teddy bear with a face is cute. A cartoon robot is fun. As a thing looks more and more human, we like it more. Then, just before it looks fully real, something flips. A doll or a computer face that is almost human, but not quite, feels creepy. The eyes are a bit dead or the skin does not move right. Our brains spot the small errors and treat them as a warning. The surprise is that getting closer to real makes it worse, not better, until it becomes perfect.

An everyday example

Look at a shop mannequin from across the road and it can pass for a person for a second. Walk up close and the frozen face becomes unsettling. A wax statue in a museum does the same, and the closer the wax face is to a real celebrity, the more people say it gives them a shiver.

The short version

Comfort rises as something becomes more humanlike, then plunges sharply just short of realistic, then recovers. A cartoon robot is charming. A near perfect android is disturbing.

Where you have seen it

Animated films that made audiences uneasy, waxworks, and the specific unease of AI generated faces where the eyes are almost right.

In films, books and games

The animated film The Polar Express (2004) is the usual example: reviewers said its lifelike characters looked eerie. The Sonic the Hedgehog film (2020) redesigned its title character after fans said the first trailer's human-like teeth were unsettling.

The uncomfortable bit

Mori proposed it as an essay with a hand drawn graph, not data. Fifty years of experiments still argue about whether the valley is a real curve or several unrelated effects wearing one name.

Source: Mori, Energy, 1970; English translation Mori, MacDorman and Kageki, IEEE Robotics and Automation Magazine, 2012

SOCIETY · Proven · Flood and Dresher, 1950

Prisoner's Dilemma

Two people acting perfectly rationally both end up worse off.

In plain words

Two friends are caught and questioned in separate rooms. Each can stay quiet or blame the other. If both stay quiet, both get a small punishment. If one blames and the other stays quiet, the blamer goes free and the quiet one gets the big punishment. If both blame, both get a medium punishment. Look at it from one side. Whatever your friend does, you do better by blaming. Your friend thinks the same. So both blame, and both end up with the medium punishment, when staying quiet together would have been better. Sensible choices add up to a bad result.

An everyday example

Two shops sell samosas on the same street. If both keep prices normal, both earn well. If one cuts its price, it steals the customers. So both cut prices, and both earn less than before. Try it with a friend and ten sweets each: each round, secretly choose share or grab, and see how fast trust falls apart.

The short version

Stay silent and you both get a light sentence. Betray while the other stays silent and you walk free. Whatever the other does, betraying is better for you individually, so both betray and both lose.

How it actually works

In Axelrod's version mutual cooperation pays each player 3, mutual defection pays each 1, and if one defects while the other cooperates the defector gets 5 and the cooperator 0. Defection is dominant: it pays more whichever the other player does, so mutual defection is the only Nash equilibrium, yet both would do better at 3 each. Any payoffs in that order give the same dilemma. Repetition without a known end changes it. If the same players will meet again with probability w, a defection today can be punished tomorrow, and Axelrod showed that cooperation can sustain itself once w is large enough, with these payoffs above one half. He called this the shadow of the future.

For his first tournament in 1979 he invited game theorists to submit programs; 14 came in and each played every other, itself and a random player for 200 moves. Tit for tat, submitted by Anatol Rapoport of the University of Toronto, cooperates on the first move and then copies its opponent's previous move. It was the shortest program entered and won with an average of 504 points per game against 600 for constant mutual cooperation. In the second round, 62 entries from six countries arrived, every author knowing what had won the first, and tit for tat won again without ever outscoring a single opponent.

Merrill Flood and Melvin Dresher devised the game at RAND in January 1950 while testing John Nash's equilibrium idea; two colleagues, Armen Alchian and John Williams, played 100 rounds and cooperated in 68 and 78 of them. Albert Tucker, Nash's supervisor at Princeton, wrapped the payoffs in a story about two arrested confederates for a talk to Stanford psychologists in May 1950, naming the game. Robert Axelrod reported the tournaments in the Journal of Conflict Resolution in 1980 and published The Evolution of Cooperation in 1984. Its fourth chapter used Tony Ashworth's 1980 study of the Western Front, where units facing each other for months shelled the same spots at the same times, the live and let live system, until commanders broke it up by ordering raids.

Where you have seen it

Arms races, price wars, overfishing, two shops undercutting each other into ruin, countries stalling on climate agreements.

In films, books and games

The British game show Golden Balls (2007 to 2009) ended every episode with it: two players secretly chose Split or Steal for the jackpot. The ferry scene in The Dark Knight (2008) is the same trap with two detonators.

The uncomfortable bit

Play it repeatedly and cooperation wins. In Axelrod's tournaments the champion strategy was brutally simple. Start friendly, copy whatever the other side did last, forgive quickly.

Source: Flood and Dresher, RAND, 1950; Axelrod, The Evolution of Cooperation, 1984 (the tournaments)

This entry has a playable demo when scripts are on.

SOCIETY · Proven · Dietrich Braess, 1968

Braess's Paradox

Add a new road and traffic gets worse. Close one and it improves.

In plain words

A city builds a new road to cut traffic. Traffic gets worse. This sounds impossible, but here is how it happens. Every driver picks the route that looks fastest for them. The new road looks like a shortcut, so lots of drivers switch to it. Now the roads that join the shortcut are jammed, and the old routes are slower too. Nobody wants to go back, because going back alone would be slower still. So everyone sits in a jam that did not exist before. The surprise is that removing a road can sometimes speed everyone up.

An everyday example

At school, one long queue for the canteen moves slowly. A second counter opens next to it. Everyone rushes across, the gap between the two counters gets blocked, and lunch takes longer than before. The same thing happens when a maps app sends every driver down the same quiet side street and it jams within minutes.

The short version

Every driver picks the route that is best for them alone. A new shortcut pulls everyone onto it, congests the shared segments, and the equilibrium settles at a slower time for everybody.

How it actually works

A road network is in equilibrium when no driver can shorten their own journey by changing route. Each driver counts only their own time and ignores the delay their car adds for everyone else on the link. A new link with a short free-flow time attracts drivers, but the links it joins now carry more traffic and slow down. The network settles into a new equilibrium that is stable, since no single driver gains by leaving it, and slower for everyone. Roughgarden and Tardos proved in 2002 that where delay grows linearly with flow, equilibrium travel time can exceed the best coordinated routing by up to four thirds.

Braess's example sends 4,000 drivers from start to end by one of two routes. Each route has one link taking, in minutes, the number of cars on it divided by 100, and one fixed 45-minute link. Split 2,000 and 2,000, the variable links take 20 minutes and everyone arrives in 65. Now add a link of zero travel time between the two midpoints. A variable link takes at most 40 minutes against a certain 45, so all 4,000 crowd onto both variable links and everyone arrives in 80. Youn, Gastner and Jeong found in Physical Review Letters in 2008 that in Boston, London and New York selfish routing cost up to about 30 per cent more than coordinated routing, and that each city had streets whose closure would shorten journeys.

Dietrich Braess of the University of Münster published the example in 1968 as 'Über ein Paradoxon aus der Verkehrsplanung' in Unternehmensforschung. On Earth Day, 22 April 1990, New York closed 42nd Street for the day; the gridlock forecast by traffic engineers did not happen, and Gina Kolata reported the outcome in The New York Times that December. In Seoul, the elevated Cheonggye expressway carried more than 160,000 vehicles a day. Demolition began in 2003 and the restored Cheonggyecheon stream opened in its place in October 2005. Traffic in the district did not seize up as forecast: car volumes fell and journeys shifted to other roads and public transport.

Where you have seen it

Seoul demolished an elevated highway in 2003 and traffic flow improved. New York closed 42nd Street for a day in 1990 and the predicted gridlock never came.

The uncomfortable bit

Individually rational choices producing a collectively worse outcome is not a traffic quirk. It is the default behaviour of shared networks, including electricity grids and the internet.

Source: Braess, Unternehmensforschung, 1968; Youn, Gastner and Jeong, Physical Review Letters, 2008

This entry has a playable demo when scripts are on.

SOCIETY · Mostly myth · Horst Siebert, 2001

Cobra Effect

Pay people for dead cobras and you get cobra farms.

In plain words

A government wants fewer snakes, so it pays money for every dead cobra. Clever people start breeding cobras at home to collect the money. When the government finds out, it stops paying. The breeders now have useless snakes, so they let them go. The city ends up with more cobras than it started with. This is a story about rewards that backfire. When you pay for a number, people find the easiest way to make that number, which may not be the thing you wanted. The surprise is that the famous Delhi version has never been found in any old record.

An everyday example

A teacher gives a sweet for every spelling mistake a pupil finds in a classmate's work. Within a week, pupils are sneaking mistakes into each other's work so they can find them. In a cricket club that pays bowlers per wicket, bowlers start begging to bowl at the weakest batters. Whatever you reward, you get more of, in whatever form is easiest.

The short version

The story says colonial officials in Delhi offered a bounty for cobra heads, people bred cobras to claim it, the scheme was scrapped and the breeders released their stock, leaving more snakes than before.

How it actually works

The story as usually told: the British administration in Delhi, worried about cobras, paid a bounty for every dead one. Enterprising residents began breeding cobras to collect the money. When the government noticed and scrapped the bounty, the breeders released their now worthless snakes, and Delhi ended up with more cobras than it started with. The German economist Horst Siebert made the tale famous with his 2001 book Der Kobra-Effekt. Nobody has found a record of the bounty, the breeders or the release in colonial archives. Historians treat the Delhi version as a parable rather than an event, which is why this entry is filed as mostly myth.

The documented case is rats, not snakes, and Hanoi, not Delhi. In 1902 the French administration, fighting rats in the new sewers, paid a bounty per rat tail. The historian Michael Vann traced the records in a 2003 paper. Daily tallies passed twenty thousand tails in June of that year, and then the city began noticing healthy rats running about with no tails. Catchers had been snipping the tail and releasing the rat to breed. Rat farms were found on the outskirts. The bounty was withdrawn, and Hanoi had a plague outbreak in 1906 regardless.

The mechanism is general. Pay for a proxy, tails, carcasses, a target number, and the proxy becomes the product, while the thing you actually wanted is untouched or worse. It is Goodhart's law with animals. Mexico City's 1989 rule banning each car from the road one weekday a week, meant to cut smog, led households to buy a second, older car with a different plate. A 1997 World Bank study by Eskeland and Feyzioglu found total driving did not fall. The lesson survives even though the cobras probably never existed: reward the outcome you want, not the evidence of it.

Where you have seen it

Quoted constantly to explain why targets backfire, from crime statistics to sales quotas.

The uncomfortable bit

No contemporary record of the Delhi cobra scheme has ever been produced. The well documented version happened in Hanoi in 1902, where a rat tail bounty produced tail-less rats and rat farms. The lesson survives. The anecdote does not.

Source: Siebert, Der Kobra-Effekt, 2001; Vann, 'Of Rats, Rice, and Race', French Colonial History, 2003 (the documented Hanoi bounty)

SOCIETY · Proven · Charles Goodhart, 1975

Goodhart's Law

The moment a measure becomes a target, it stops measuring anything.

In plain words

A measure is a number that tells you how well something is going, like a test score or how many customers a shop serves. A target is when someone says, get that number up. As soon as people are judged by the number, they start working on the number instead of the real thing. Test scores rise because pupils learn to pass tests, not because they learn more. The number that used to tell you something now tells you nothing. The surprise is that nobody is cheating. People simply do what they were asked to do.

An everyday example

A shop manager tells staff they must serve fifty customers each per hour. Staff begin rushing people, skipping questions and pushing anyone with a problem to the next counter. The number goes up. Customers leave unhappy. At home, if pocket money depends on how many minutes of homework you log, the minutes get longer and the homework does not.

The short version

People optimise for the number, not the thing the number stood for. The correlation that made the metric useful is exactly what the pressure destroys.

How it actually works

A measure is useful because it is correlated with something harder to observe. The correlation holds only while nobody has a reason to act on the number itself. Once pay, funding or promotion depends on it, the cheapest way to move the number is rarely to improve the underlying quality. It is to change how the number is produced: reclassify cases, pick easier work, record events differently, deliver the countable output and drop the uncountable one. The pressure selects the behaviours that break the link. Goodhart saw it in money: once the Bank of England targeted a monetary aggregate, banks and borrowers adjusted and the aggregate stopped tracking the economy.

English ambulance trusts were set a target of reaching 75 per cent of life-threatening calls within eight minutes. Gwyn Bevan and Christopher Hood showed in Public Administration in 2006 that recorded response times bunched just under eight minutes and fell away just over it, consistent with adjusted records rather than faster ambulances. At Wells Fargo, staff were pushed to sell eight products per customer. In September 2016 the bank paid 185 million dollars in penalties after regulators found about 1.5 million deposit accounts and 565,000 credit card accounts opened without customers' consent; around 5,300 employees were dismissed. The Soviet nail factory, making a few giant nails when paid by weight and thousands of tiny ones when paid by count, is an anecdote and a Krokodil cartoon, not a documented record.

Charles Goodhart, then at the Bank of England, wrote the original sentence for a 1975 Reserve Bank of Australia conference, in 'Problems of Monetary Management: The UK Experience': any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes. Donald Campbell stated the same rule for social indicators in 'Assessing the Impact of Planned Social Change', a 1976 Dartmouth College paper, now called Campbell's law. The wording usually quoted belongs to the anthropologist Marilyn Strathern, who wrote in a 1997 article on university audit in European Review that when a measure becomes a target, it ceases to be a good measure.

Where you have seen it

Schools coaching for exams instead of teaching. Support teams closing tickets fast instead of solving problems. Hospitals reclassifying patients to protect waiting time figures.

In films, books and games

The TV series The Wire (2002 to 2008) is built on it: police officers cooking the crime figures, and in its fourth season a school that teaches only what the test will ask.

The uncomfortable bit

It is not cheating, it is obedience. You asked for the number and you got the number. Design the target expecting people to be smarter than it.

Source: Goodhart, 'Problems of Monetary Management', 1975; Strathern's now-standard wording, 1997

SOCIETY · Proven · Mike Masnick, 2005

Streisand Effect

Try to bury something online and you personally hand it an audience.

In plain words

Someone finds a photo or a story about them online that they do not like. They try to get it removed, perhaps with a lawyer. Before that, hardly anyone had seen it. Now the news is that somebody is trying to hide something, and everybody wants to look. The thing they wanted buried gets copied, shared and saved thousands of times. The attempt to hide it is what made it famous. The surprise is that silence would have worked better. Most people never notice a quiet thing until someone makes a fuss about it.

An everyday example

A boy posts a slightly embarrassing photo of himself in the class group chat. Two people see it. He panics, deletes it, and sends three messages begging everyone not to screenshot. Now the whole class wants to know what it was, and someone did screenshot it. If he had left it, it would have scrolled away in an hour.

The short version

Named after a 2003 lawsuit by an American singer over an aerial photograph of her house that almost nobody had looked at. The legal action turned an obscure image into international news.

Where you have seen it

Every takedown notice, gag order and deleted post that ends up screenshotted by ten thousand people.

The uncomfortable bit

The suppression attempt is itself the story, and it is a far better story than the thing being suppressed. Legal teams still learn this the expensive way.

Source: Masnick, Techdirt, 2005; Streisand v. Adelman, California Superior Court, 2003

SOCIETY · Proven · Scott Feld, 1991

Friendship Paradox

Your friends have more friends than you. This is true for almost everyone.

In plain words

On average, your friends have more friends than you do. This sounds like an insult, but it is true for nearly everyone, and it is just maths. Think about a very popular person with a hundred friends. They show up on a hundred people's friend lists. A shy person with two friends shows up on only two lists. So when you look at your own friend list, the popular people are there much more often than the shy ones. Your friends are a sample that leans towards popular people. The surprise is that the same trick makes buses feel crowded and gyms look full.

An everyday example

Ask five classmates to write down how many close friends they have, then ask each of them to write the number for their best friend. Add up each column. The best-friend column almost always wins. On Instagram the accounts you follow look busier than yours because you naturally follow the busy accounts.

The short version

Popular people appear in far more friend lists, so they are massively overrepresented when you sample the friends of a random person. Most people are genuinely below the average of the group they observe.

Where you have seen it

Your feed looks like everyone is travelling, getting promoted and getting married, because you are sampling the most active people constantly.

The uncomfortable bit

The same shape explains why your bus always feels crowded and your gym always looks full. You are far more likely to be present during the busy times. It is called the inspection paradox.

Source: Feld, 'Why Your Friends Have More Friends Than You Do', American Journal of Sociology, 1991

SOCIETY · Contested · Robin Dunbar, 1992

Dunbar's Number

You can maintain about 150 real relationships. After that the brain runs out.

In plain words

The idea is that a person can only keep up about 150 real relationships, the sort where you know who someone is and how they connect to you. Beyond that, names start to slip. Inside the 150 are smaller rings: a handful of very close people, then about fifteen, then about fifty. The number came from comparing monkeys and apes: bigger brains went with bigger groups, and humans were placed on that line. The surprise is how often 150 pops up in villages, army units and companies. Later checks found the real range could be much lower or much higher.

An everyday example

Think about your phone. You might have five hundred contacts, but count how many you would happily call at midnight. Then count how many you would stop to chat with in a market. Then count how many names you would recognise at a wedding. The rings get bigger and looser, roughly five, fifteen, fifty, and then the crowd.

The short version

Dunbar found a correlation between primate brain size and group size, and extrapolated a human limit near 150. Inside it sit smaller circles, roughly 5 close friends, 15 confidants, 50 good friends.

Where you have seen it

Hunter gatherer bands, army companies, village sizes in old census records and the point where a growing company suddenly needs an org chart.

In films, books and games

Malcolm Gladwell's The Tipping Point (2000) turns it into the Rule of 150, with the story of Gore Associates, the Gore-Tex firm, which builds a new factory rather than let any one plant grow past about 150 staff.

The uncomfortable bit

Later researchers redid the analysis and found the error bars are enormous, from a handful to several hundred. It is a useful frame, not a measured constant, and it is quoted as though it were a law.

Source: Dunbar, Journal of Human Evolution, 1992; Lindenfors et al., Biology Letters, 2021 (the error-bar critique)

SOCIETY · Contested · Garrett Hardin, 1968

Tragedy of the Commons

Everyone taking their fair share of a shared resource destroys it.

In plain words

A field belongs to a whole village and anyone can graze cows on it. Each farmer thinks, one more cow will earn me more milk, and the field is big. But every farmer thinks the same. Soon there are too many cows, the grass is eaten to the roots, and everyone loses. Each person's choice made sense on its own. Together they wrecked the thing they all needed. This is used to explain overfishing and dirty air. The surprise is that real villages often avoid it, by making local rules and keeping an eye on each other.

An everyday example

The kitchen in a shared hostel has one bottle of dish soap for eight people. Everyone uses a bit extra because it is not theirs, and nobody buys a new one. It runs out in a week. The rooms that put up a rota and a small kitty never have this problem. Rules plus a bit of watching fix the field.

The short version

Each herder gains fully from one more animal on the common pasture while the cost of overgrazing is split across everyone. So everyone adds animals, and the pasture dies.

How it actually works

The mechanism is a payoff asymmetry. Under open access, a herder who adds an animal keeps the whole of the extra income, while the damage from one more mouth on the pasture is shared among every user. Private gain exceeds private cost at every step, so adding animals is the best move whatever anyone else does. The structure is a many-player prisoner's dilemma; the outcome depends on payoffs, not greed. What Hardin's model leaves out is communication, monitoring and enforceable rules. Ostrom's central distinction is between open-access resources and common-property regimes, where a defined group holds the resource, excludes outsiders and sanctions its own members. Only the first fits Hardin's model.

The northern cod off Newfoundland is the standard failure. Landings peaked at about 810,000 tonnes in 1968, and the stock did not recover after Canada extended its fishing zone to 200 nautical miles in 1977. On 2 July 1992 the fisheries minister John Crosbie declared a moratorium; the spawning stock had fallen to around one per cent of its earlier size and roughly 30,000 people lost their jobs. Ostrom's counter-cases have longer records. The Swiss village of Törbel has regulated its alpine grazing under written rules since 1517, including a rule that no household may summer more cattle than it can winter. The Tribunal de las Aguas in Valencia has settled irrigation disputes for about a thousand years and still sits every Thursday.

William Forster Lloyd, an Oxford political economist, described the overgrazed common in 'Two Lectures on the Checks to Population' in 1833. Garrett Hardin revived the example in Science on 13 December 1968 in 'The Tragedy of the Commons', a paper whose real subject was population growth. In a 1998 note in the same journal Hardin conceded that the title should have been the tragedy of the unmanaged commons. Elinor Ostrom's Governing the Commons appeared in 1990, built on field studies of fisheries, forests and irrigation systems, and set out eight design principles for durable commons. She shared the 2009 Nobel Memorial Prize in Economic Sciences with Oliver Williamson, the first woman to receive it.

Where you have seen it

Groundwater depletion, overfishing, air pollution, and every shared office fridge in history.

The uncomfortable bit

Elinor Ostrom won a Nobel Prize for showing this is not inevitable. Communities from Swiss alpine villages to Rajasthani water systems have managed shared resources for centuries with local rules, and Hardin's model quietly assumed no one could talk to each other.

Source: Hardin, Science 162, 1968; Ostrom, Governing the Commons, Cambridge University Press, 1990

LIFE · Proven · Leigh Van Valen, 1973

Red Queen Hypothesis

You have to keep running just to stay in the same place.

In plain words

In a race, if you run faster and everyone else runs faster too, you finish in the same place. Living things are in that race. A cheetah gets quicker, so the gazelles that survive are the quicker ones, so the cheetah needs to be quicker still. Germs change to beat our medicines, so we need new medicines. Nobody ever gets to stop. The name comes from a queen in a children's book who tells Alice that here you must run just to stay where you are. The surprise is that all this effort produces no clear winner, only endless keeping up.

An everyday example

Two friends play a card game every evening. One finds a clever trick and wins for a week. The other works out how to block it. Then the first finds a new trick. After a year both are far better players, and the score is still about even. Flu vaccines are changed every year for exactly this reason.

The short version

Species do not evolve toward some finish line. They evolve to keep up with everything else that is evolving against them. Predators, prey and parasites are all improving at once, so relative fitness barely moves.

Where you have seen it

Antibiotic resistance, crop pests defeating each new pesticide, and the annual scramble to redesign flu vaccines.

In films, books and games

The name is from Lewis Carroll's Through the Looking-Glass (1871), where the Red Queen tells Alice that it takes all the running you can do to keep in the same place.

The uncomfortable bit

It is the best explanation for why sex exists at all. Cloning is twice as efficient, but shuffling genes every generation is the only way to stay ahead of parasites that adapt faster than you do.

Source: Van Valen, 'A New Evolutionary Law', Evolutionary Theory, 1973

LIFE · Open question · Richard Peto, 1977

Peto's Paradox

A whale has a thousand times more cells than you and gets less cancer.

In plain words

Cancer starts when one cell goes wrong and starts copying itself. So an animal with more cells should have more chances for a cell to go wrong. A blue whale has vastly more cells than a mouse, and lives far longer, so it should be full of cancer. It is not. Elephants get less cancer than people. Big animals must have extra ways to catch broken cells before they cause trouble, and scientists are still working out what they are. The surprise is that inside one species the rule works fine: bigger dogs, and taller people, do get a bit more cancer.

An everyday example

Think of a school with ten pupils and one with ten thousand. The big school has far more chances for a fight to break out. You would expect it to have many more fights. The paradox is that the big schools have about the same number. Something about being big makes them better at stopping trouble early.

The short version

Cancer starts when a cell mutates. More cells and more years should mean far more cancer, so blue whales and elephants ought to be riddled with it. Across species, body size barely predicts cancer rate at all.

How it actually works

Cancer begins with mutations accumulating in a single cell lineage, and the multistage model of Armitage and Doll, 1954, treats every cell as an independent lottery ticket. Expected cancers then scale with the number of cells at risk multiplied by the divisions each undergoes over a lifetime. An adult human has roughly 37 trillion cells; a blue whale has about a thousand times as many and lives longer. The model predicts that large, long-lived animals should rarely reach adulthood without cancer. The accepted resolution is that per-cell risk is a trait under selection. A lineage evolving towards larger bodies or longer lives is selected for whatever suppresses tumours.

Lisa Abegglen and colleagues reported in JAMA in 2015 that the African elephant genome carries at least 20 copies of TP53; humans have one. In zoo records, cancer accounted for about 4.8 per cent of elephant deaths, against an estimated 11 to 25 per cent in humans, and irradiated elephant lymphocytes died by programmed cell death at about twice the human rate. The bowhead whale, which lives beyond 200 years, was sequenced by Keane and colleagues in Cell Reports in 2015 and showed changes in DNA-repair and cell-cycle genes. Cagan and colleagues reported in Nature in 2022 that yearly somatic mutation rates in 16 mammals vary inversely with lifespan, so a mouse and a giraffe both end life with around 3,200 mutations per cell.

Richard Peto set out the puzzle in two steps. A 1975 paper with Roe, Lee, Levy and Clack in the British Journal of Cancer, 'Cancer and ageing in mice and men', showed in skin-painting experiments that tumour incidence tracked duration of carcinogen exposure, not age. In a 1977 essay in the Cold Spring Harbor volume Origins of Human Cancer he noted that a mouse has about a thousandth of a human's cells and a thirtieth of the lifespan, yet the two species have comparable lifetime cancer rates, so a human cell must be far more resistant. Aleah Caulin and Carlo Maley fixed the name in a 2011 review in Trends in Ecology and Evolution.

Where you have seen it

Elephants carry about 20 copies of the tumour suppressing gene TP53. Humans carry one.

The uncomfortable bit

Inside a single species the expected pattern holds. Taller humans do have slightly higher cancer rates. Evolution solved the problem between species and never bothered solving it within one.

Source: Peto et al., 1975; Peto, 'Epidemiology, multistage models and short-term mutagenicity tests', 1977; Abegglen et al., JAMA, 2015 (elephant TP53)

LIFE · Contested · Berdoy and Webster, 2000

Toxoplasma Manipulation

A parasite rewires a rat's brain so that it stops fearing cats.

In plain words

There is a tiny parasite that can live in many animals but can only have babies inside a cat. It has a trick for getting into cats. When it infects a mouse or rat, the animal stops being scared of the smell of cats. Some even seem to like it. So the rodent wanders near a cat, gets eaten, and the parasite arrives where it wanted to be. A living thing changed another animal's mind for its own benefit. The surprise is that many people carry the same parasite too, though claims that it changes our behaviour are weak.

An everyday example

Think of a friend who is scared of dogs. One day, after a bout of illness, he starts walking up to every dog in the park. That is roughly what happens to an infected mouse with cats. You cannot try this at home, but you can avoid the parasite: cook meat properly and wash your hands after cleaning a cat's litter tray.

The short version

Toxoplasma gondii can only breed inside a cat. Infected rodents lose their instinctive fear of cat odour, and some become mildly attracted to it, which delivers the parasite exactly where it needs to go.

Where you have seen it

It is one of the cleanest demonstrated cases of a parasite hijacking behaviour rather than just the body.

The uncomfortable bit

Roughly a third of humans carry it. Studies linking it to human risk taking and personality exist but the effects are small, inconsistent and heavily confounded. Treat the scary headlines with suspicion.

Source: Berdoy, Webster and Macdonald, Proceedings of the Royal Society B, 2000; Sugden et al., PLOS ONE, 2016 (no human personality link found)

LIFE · Contested · Kristen Hawkes, 1997

Grandmother Hypothesis

Why do human women live for decades after they stop having children?

In plain words

Most animals keep having babies until they die. Human women stop having children in their late forties and then often live for another thirty or forty years. This is odd for evolution, which usually favours having more offspring. One answer: a grandmother who helps feed and look after her grandchildren means more of those children survive. Those grandchildren carry her genes. So a long life after the babies stop can be worth more than a few risky late pregnancies. The surprise is that killer whales seem to have found the same trick, with old females leading the pod.

An everyday example

In many Indian homes, the dadi or nani runs the kitchen, walks the children to school and knows every remedy for a fever, while the parents work. Ask around your class how many pupils were partly raised by a grandmother. The answer is usually most of them. That daily help is exactly what the idea says evolution rewarded.

The short version

Almost no other mammal does this. The proposal is that a grandmother who gathers food for grandchildren spreads more of her genes than one who keeps having risky late pregnancies, so long post reproductive life was selected for.

Where you have seen it

Field studies of Hadza foragers found grandmothers doing serious daily food collection that directly improved grandchildren's growth.

The uncomfortable bit

Killer whales do the same thing, and post menopausal matriarchs lead the pod to food during lean years. Two species separated by 90 million years arrived at the same answer.

Source: Hawkes et al., PNAS, 1998 (Hadza foragers); Croft et al., Current Biology, 2017 (killer whales)

LIFE · Proven · J. Bristol Foster, 1964

Island Dwarfism

Strand elephants on an island and they shrink. Strand rats and they grow.

In plain words

Put a group of big animals on a small island and, over many thousands of years, they get smaller. Put small animals there and they often get bigger. Islands have less food, so a smaller elephant that needs less grass does better than a huge one. Islands also have fewer big hunters, so a rat no longer needs to hide and can grow into the space a bigger animal would have filled. Fossils show elephants the size of ponies and rats the size of cats. The surprise is that a small ancient human species on Flores may have shrunk the same way.

An everyday example

Think of a family moving from a large house to a small flat. The big sofa gets swapped for a small one, but the tiny stool gets replaced by a proper chair because there is now room where the sofa was. Shetland ponies, from small Scottish islands, are far smaller than mainland horses.

The short version

Islands have limited food and usually no large predators. Big animals shrink toward efficiency, small animals grow into the empty niches. Both directions have been documented many times in the fossil record.

Where you have seen it

Dwarf elephants the height of a large dog once lived on Mediterranean islands. Flores in Indonesia had dwarf elephants and rats the size of cats at the same time.

The uncomfortable bit

Flores also produced Homo floresiensis, a human relative just over a metre tall. The rule that shrinks elephants appears to have applied to us as well.

Source: Foster, 'The Evolution of Mammals on Islands', Nature, 1964; Brown et al., Nature, 2004 (Homo floresiensis)

TIME · Thought experiment · Time travel fiction, 1940s

Predestination Paradox

You travel back to prevent the disaster, and your visit is what causes it.

In plain words

A time traveller goes back to stop something terrible. While there, they do things that turn out to be the reason it happened. The disaster they remember was caused by their own trip. Nothing is broken in the story: the past always included their visit, they just did not know it. This is different from the puzzle where you go back and stop your own grandparents meeting, which cannot work. Here everything fits together perfectly. The surprise is what it does to choice. Every step you take to escape the ending is one of the steps that leads to it.

An everyday example

You get a note in your own handwriting saying, do not go to the match on Sunday. So you stay home and, bored, write a note to warn yourself, and find a way to send it back. The note only exists because you got the note. Old stories about fortune tellers work the same way: the prince who flees the prophecy runs straight into it.

The short version

The traveller's actions in the past turn out to be the very events that created the future they came from. Unlike the grandfather paradox, nothing contradicts itself. The loop is perfectly consistent, which is what makes it disturbing.

Where you have seen it

The prophecy that only comes true because someone heard it and tried to dodge it. Oedipus is the two thousand year old version.

In films, books and games

The film Predestination (2014) is built entirely on one closed loop. The Terminator (1984) has the same shape: John Connor sends his own father back in time. In Harry Potter and the Prisoner of Azkaban (1999), Harry is saved by a spell he casts later that same night.

The uncomfortable bit

In these loops free will becomes the casualty. Every choice you make to escape the outcome is already part of how the outcome happened.

Source: Heinlein, 'By His Bootstraps', 1941 and 'All You Zombies', 1959

TIME · Open question · Igor Novikov, 1983

Novikov Principle

Physics may allow time travel, but only trips that change nothing.

In plain words

Suppose time travel to the past is possible. A Russian physicist, Igor Novikov, suggested a rule: you can visit, but you cannot change anything. Whatever you do there already happened. Try to stop an event and something small will get in your way: you will miss the bus, drop the note, or arrive too late. Not because anyone is stopping you, but because the only versions of history that can exist are the ones that fit together. There is only one past, and you were always in it. The surprise is that this makes time travel safe and rather dull.

An everyday example

Your friend says he will go back and warn you not to lose your phone. Under this rule, he does go back, but you are in the shower, the note falls behind the bed, and you lose the phone anyway. In fact, you now remember finding an odd note behind the bed last year. That was him.

The short version

If closed timelike curves exist, Novikov argued only self consistent histories have non zero probability. You can visit the past. You cannot alter it, because any attempted alteration was already part of history.

Where you have seen it

Billiard ball versions have been worked out on paper. A ball knocked into a wormhole to collide with its earlier self always finds a consistent trajectory instead.

In films, books and games

The film 12 Monkeys (1995) follows the rule: the traveller's attempt to change the past turns out to be the childhood memory that haunts him.

The uncomfortable bit

The universe would not stop you with force. The gun jams, the letter is lost, you arrive late. Probability itself bends so history holds.

Source: Friedman, Morris, Novikov et al., 'Cauchy Problem in Spacetimes with Closed Timelike Curves', Physical Review D, 1990

TIME · Thought experiment · Max Tegmark, 1997

Quantum Immortality

If all outcomes happen in branching worlds, some version of you never dies.

In plain words

One reading of the physics of tiny things says that whenever an event could go two ways, the universe splits, and both happen in separate copies. Now think about something that might kill you. In some copies you die, and in some you live. You cannot experience being dead. So, from your own point of view, you only ever find yourself in the copies where you survived. Follow this to the end and it seems you can never die, from the inside. It is a thought game, not advice. Most physicists think it is wrong, and real deaths are not clean coin flips.

An everyday example

A friend flips a coin: heads you win a sweet, tails you lose. Under this idea there are now two of you, one with a sweet and one without. Now think of a game where tails means you vanish. The only you who can still be thinking about the game is the one who kept getting heads. Never test this. It is not a plan.

The short version

Under the many worlds reading of quantum mechanics, every survivable event splits into branches where you live and branches where you do not. Your experience can only continue in the branches where you live, so from the inside, you always survive.

Where you have seen it

A staple of internet rabbit holes and late night hostel arguments, usually presented far more confidently than any physicist would.

The uncomfortable bit

Tegmark himself points out the flaw. Dying is rarely an instant clean branch, and most physicists reject the argument. Do not update any life decision on this one.

Source: Tegmark, 'The Interpretation of Quantum Mechanics: Many Worlds or Many Words?', Fortschritte der Physik, 1998

TIME · Open question · Arthur Eddington, 1927

Arrow of Time

Almost no law of physics cares which way time runs. So why can't you unscramble an egg?

In plain words

Play a video of a ball bouncing backwards and it still looks normal. Play a video of a glass smashing backwards and everyone can tell. Yet the basic rules of physics work the same in both directions. The reason we can tell is about tidiness. There are only a few ways for the pieces of a glass to be arranged as a glass, and countless ways for them to be a mess. So messes are much more likely than tidy things. That is the only reason time seems to point forwards. The surprise is that it means the universe began strangely tidy, and nobody knows why.

An everyday example

Tip a jigsaw box onto the table and the pieces land jumbled. You would never expect them to land as the finished picture. Both are just arrangements, but there is one finished picture and millions of jumbles. Stir milk into your chai and watch it mix. You could stir for a year and it would never unmix.

The short version

The fundamental equations work the same forwards and backwards. The one exception is entropy. Disorder overwhelmingly increases because there are astronomically more messy arrangements than tidy ones. Time's direction is statistics, not law.

Where you have seen it

Broken glass never reassembles, tea never unmixes, and you remember yesterday but not tomorrow. Memory itself may exist only because entropy climbs.

In films, books and games

Christopher Nolan's Tenet (2020) is about objects and people whose entropy runs backwards, so bullets fly back into guns.

The uncomfortable bit

This pushes the mystery to the beginning. The past had lower entropy only because the universe started in an absurdly ordered state, and nobody knows why it did.

Source: Eddington, The Nature of the Physical World, Cambridge University Press, 1928

CHANCE · Proven · Kahneman and Tversky, 1973

Base Rate Fallacy

A 99 percent accurate test says you are sick. You are probably fine.

In plain words

A test for a rare illness is right 99 times out of 100. Your test comes back positive. It feels like you must be ill. But think about how rare the illness is. If only one person in ten thousand has it, then in a town of a million people, about a hundred are ill. The test will also give a wrong answer to about one in every hundred healthy people, which is roughly ten thousand people. So most positive results belong to healthy people. The surprise is that a very good test can still be wrong most of the time when the thing it hunts for is rare.

An everyday example

A school has 1,000 pupils and one of them has been taking samosas from the canteen without paying. A sniffer dog is right 99 times out of 100. It barks at about eleven pupils. Ten of them are innocent, because one percent of 999 honest pupils is about ten. Before you accuse anyone, ask how many people there were to begin with.

The short version

If a disease affects 1 in 10,000 people, then in a million tests the true patients number 100 while the false alarms number about 10,000. A positive result is overwhelmingly more likely to be an error, yet our intuition throws away the base rate.

How it actually works

Bayes' rule weighs a test result against how common the condition was before the test: the probability of disease given a positive result equals the true positives divided by all positives, and all positives include the false alarms generated by the healthy majority. When the condition is rare that majority is enormous, and even a small false positive rate applied to it produces more false alarms than true cases. Kahneman and Tversky's explanation is representativeness: a positive result resembles being ill, and the prior frequency, which resembles nothing, is discarded.

Casscells asked 60 students and staff at Harvard Medical School: a disease has a prevalence of 1 in 1,000, the test has a false positive rate of 5 percent, and a person tests positive. If the test misses no true cases, 1,000 people contain one true case, and about 50 of the 999 healthy test positive, so a positive result is right about 1 time in 51, or 2 percent. Eleven of the 60 gave that answer; 27 said 95 percent. In the cab problem, 85 percent of cabs are Green and 15 percent Blue, and a witness who is right 80 percent of the time says the cab in an accident was Blue. Out of 100 cabs, the witness would call 12 of the 15 Blue cabs Blue and 17 of the 85 Green ones Blue, so the chance the cab was Blue is 12 in 29, or 41 percent. Most respondents said 80 percent.

Daniel Kahneman and Amos Tversky reported the neglect of prior probabilities in On the Psychology of Prediction in 1973, with sketches of people drawn from a pool of lawyers and engineers; switching the stated mix from 70 lawyers to 70 engineers barely moved judgements. The cab problem appeared in print in their 1980 chapter Causal Schemas in Judgments under Uncertainty. Casscells, Schoenberger and Graboys published the Harvard result in the New England Journal of Medicine in 1978. Gerd Gigerenzer and Ulrich Hoffrage showed in 1995 that posing such problems as counts rather than percentages raised correct answers from 16 to 46 percent.

Where you have seen it

Medical screening debates, spam filters, airport security, and every scary headline built on a rare event detector.

The uncomfortable bit

Doctors get this wrong too. In repeated studies, a majority of physicians badly overestimated what a positive result means. Ask "out of how many" before you panic.

Source: Kahneman and Tversky, 'On the Psychology of Prediction', Psychological Review, 1973; Casscells et al., New England Journal of Medicine, 1978 (physicians)

This entry has a playable demo when scripts are on.

CHANCE · Proven · Daniel Bernoulli, 1738

St. Petersburg Paradox

A coin game with infinite expected value that nobody would pay fifty rupees to play.

In plain words

You flip a coin. If it lands heads on the first go, you win 2 rupees. If it takes two flips, you win 4. Three flips, 8. Each extra flip doubles the prize. Now work out the average prize. Every possible ending adds 1 rupee to the average, and there are endless possible endings, so the average is endless too. By that sum you should pay any price to play. Yet almost nobody would hand over even a small note. The puzzle is that the maths says one thing and every sensible person says another, and the sensible people are not wrong.

An everyday example

Try it in the kitchen with a friend and a coin. Your friend offers you the doubling game for 20 rupees a go. Flip until heads. Most games end on the first or second flip and pay 2 or 4. You will lose 20 rupees over and over, and the one huge win that would balance the books may never come in your lifetime.

The short version

Toss a coin until heads appears. The pot doubles with each toss. The maths says the expected payout is infinite, so any entry price should be a bargain. Real people will not stake more than a small note, and they are right.

Where you have seen it

Every lottery, every lottery-like startup bet, every argument between expected value and common sense.

The uncomfortable bit

Bernoulli's fix invented modern economics. Money has diminishing utility. Your second crore changes your life less than your first, and once you account for that, the infinite value collapses.

Source: Bernoulli, 'Specimen Theoriae Novae de Mensura Sortis', Commentarii, 1738

CHANCE · Open question · Adam Elga, 2000

Sleeping Beauty Problem

A coin flip probability that professional philosophers cannot agree is a half or a third.

In plain words

A girl agrees to an odd experiment. On Sunday she goes to sleep and someone flips a coin. If it lands heads, she is woken on Monday, asked a question, and that is all. If it lands tails, she is woken on Monday, given a pill that makes her forget that waking, then woken again on Tuesday. Each time she wakes she is asked the same thing: how likely it is that the coin was heads. One camp says a half, because a fair coin is a fair coin. The other says a third, because two of her three possible wakings come after tails. Clever people are still arguing.

An everyday example

Two friends can play a version by text. Ravi flips a coin in secret. Heads, he sends Priya one message saying guess. Tails, he sends two, an hour apart, and Priya has agreed not to check whether a message is her first. A message arrives. Priya must say how likely heads is. Over many games, two thirds of all messages follow tails, but only half of all games do.

The short version

Beauty sleeps. A coin is flipped. Heads, she is woken once. Tails, twice, with her memory wiped between. Waking up, what should she believe the chance of heads is? "Halfers" say nothing new was learned. "Thirders" say two of three awakenings follow tails.

Where you have seen it

Twenty five years of journal papers, and the split has not closed.

The uncomfortable bit

This is not a trick question with a hidden answer. It exposes that "probability" quietly mixes two different ideas, chance of the event and expectation over your own experiences, and they can disagree.

Source: Elga, Analysis, 2000 (the thirder case); Lewis, Analysis, 2001 (the halfer reply)

CHANCE · Proven · Diaconis and Mosteller, 1989

Law of Truly Large Numbers

Miracles should happen regularly. One in a million events hit constantly in a country of 1.4 billion.

In plain words

Something with a one in a million chance sounds impossible. But India has more than a billion people, and each one lives through thousands of small moments every day. Give a very rare thing that many chances and it will happen to somebody, somewhere, almost every day. The person it happens to feels chosen. Everybody else hears the story and feels a shiver. What nobody counts is the enormous crowd it did not happen to. Rare things happening is not a sign of magic. Rare things never happening would be the real mystery.

An everyday example

You dream about an old school friend and the next morning she rings you. It feels like a sign. Now count. More than a billion Indians go to sleep every night, and most people know hundreds of others. Even if the match is a one in a million fluke, it should happen to over a thousand people in the country by breakfast. You just heard from one of them.

The short version

With enough opportunities, outrageously unlikely things become routine. Someone somewhere will dream of a relative the night they die, win two lotteries, or be struck by lightning twice. The surprise is manufactured by ignoring the sample size.

Where you have seen it

Every viral coincidence story. The reporter never interviews the billion people it did not happen to.

The uncomfortable bit

A world with no one in a million coincidences would itself be statistical proof that something was rigged.

Source: Diaconis and Mosteller, 'Methods for Studying Coincidences', Journal of the American Statistical Association, 1989

CHANCE · Proven · Francis Galton, 1886

Regression to the Mean

The rookie of the year slumps in season two. Nothing went wrong. Maths happened.

In plain words

Every result mixes how good you are with how lucky you were that day. When someone does amazingly well, part of that was skill and part was a good day. The good day does not come again on purpose. So the next result usually drops back towards their normal level. Nothing broke. The luck simply ran out, as it always does. The same is true after a terrible day: things tend to look better next time all on their own. People see this drop or rise and invent a reason for it, a curse, a swollen head, a magic pep talk, when it is just numbers settling down.

An everyday example

A batsman scores 120 in one match, the best of his life. His coach praises him. Next match he scores 30. The coach decides praise made him lazy. Another batsman gets a duck, the coach shouts, and next match he scores 40. The coach decides shouting works. Both scores were just returning to normal. Try it with dice: after a six, the next roll averages three and a half.

The short version

Any extreme result mixes skill and luck. The luck component does not repeat, so extraordinary performances are usually followed by ordinary ones. No curse, no complacency, just the noise washing out.

Where you have seen it

The second album problem, the magazine cover jinx, the topper who scores lower in the next exam, the fund that beat the market once.

The uncomfortable bit

It runs the other way too, and it fools teachers and managers. Punishment after a terrible performance looks like it worked, and praise after a great one looks like it backfired, when both were just scores drifting back to normal.

Source: Galton, 'Regression Towards Mediocrity in Hereditary Stature', Journal of the Anthropological Institute, 1886

CHANCE · Proven · Thompson and Schumann, 1987

Prosecutor's Fallacy

The chance of the evidence given innocence is not the chance of innocence given the evidence.

In plain words

Two questions sound alike but are very different. One asks how likely the evidence would be if the person were innocent. Two asks how likely innocence is now that the evidence matches. A lawyer might say a match is one in a million, so the accused must be guilty. But in a big city, a one in a million match fits several people, and only one of them did it. The rare sounding number is about the evidence, not about the person. Mixing the two up has sent innocent people to prison, because the swap is hard to spot and sounds like proof.

An everyday example

Your class has 40 pupils. Someone ate the teacher's lunch and left a crumb of a biscuit only sold at one shop. Three pupils bought that biscuit this week. Saying only one in thirteen of you had it sounds bad for Anil. But three people had it, so the crumb alone gives Anil about a one in three chance of being the culprit, nowhere near certainty.

The short version

"Only one in a million people match this DNA" sounds damning. But in a city of ten million, ten innocent people match. Swapping the two conditional probabilities is a subtle inversion with catastrophic consequences.

Where you have seen it

In a notorious British case, a mother was convicted after an expert wrongly squared the odds of two cot deaths, treating them as independent. The conviction was later overturned.

The uncomfortable bit

Courts have jailed people on this arithmetic error. Statisticians now formally advise judges in several countries because the inversion fools juries almost every time.

Source: Thompson and Schumann, Law and Human Behavior, 1987; R v Sally Clark, Court of Appeal, 2003

CHANCE · Proven · Joseph Bertrand, 1889

Bertrand's Box

You drew a gold coin. The odds the next coin is gold are two thirds, not half.

In plain words

There are three boxes. One holds two gold coins. One holds two silver coins. One holds one of each. You pick a box without looking and pull out one coin. It is gold. You want the chance that the other coin in that box is also gold. Most people say a half, since the box must be the gold-gold one or the mixed one. But think about how you got here. The gold-gold box has two golds you could have drawn. The mixed box has only one. So you are twice as likely to be holding the gold-gold box. The answer is two thirds.

An everyday example

Try it with three matchboxes, three 5 rupee coins and three 1 rupee coins. Fill the boxes as two 5s, two 1s, and one of each. Shuffle the boxes and draw one coin blind. Whenever the first coin is a 5, note what the second coin is. After thirty such draws the second coin will also be a 5 about twenty times, not fifteen.

The short version

Three boxes: gold-gold, silver-silver, gold-silver. You pick a box, draw a coin, it is gold. Intuition says the box is now either gold-gold or gold-silver, so fifty fifty. But the gold-gold box gave you twice as many ways to draw gold, so it is twice as likely.

Where you have seen it

This 1889 puzzle is the direct ancestor of the Monty Hall problem, and it breaks brains the same way.

The uncomfortable bit

The lesson generalises. Evidence does not just narrow the options, it reweights them by how easily each option produces that evidence.

Source: Bertrand, Calcul des probabilites, 1889

LOGIC · Open question · Eubulides, fourth century BCE

Sorites Paradox

Remove one grain from a heap and it is still a heap. Repeat until one grain remains.

In plain words

Picture a big pile of sand. Take away one grain. It is still a pile. Take away another. Still a pile. No single grain can be the one that turns a pile into not a pile, that would be silly. But keep going and eventually you have one grain sitting on the table, and nobody calls that a pile. Nobody can say where the pile stopped being a pile. Words like pile, tall, old and rich have fuzzy edges. Our logic likes sharp yes or no answers, and it gets stuck when the world only gives it a smooth slope.

An everyday example

Ask at home how many hairs a man can lose before he is bald. Try it on a friend's height. At 150 cm he is not tall. Add one millimetre. Still not tall. Nobody agrees on the millimetre that makes him tall, yet everyone agrees that at 200 cm he is. Do the same with a cup of chai cooling on the table: name the second it stops being hot.

The short version

No single grain turns a heap into a non-heap, yet a million removals clearly do. Vague words like heap, tall, bald and rich have no sharp boundary, and classical logic chokes on them.

Where you have seen it

When exactly does a startup become a big company? At what rupee does someone become rich? Every legal age limit is a knife forced through a smooth slope.

The uncomfortable bit

The law's answer is to draw arbitrary lines and enforce them, which is why you can legally drive at 18 years and 0 days and not one day earlier. The line is fake and necessary at the same time.

Source: Attributed to Eubulides, 4th century BCE; Williamson, Vagueness, Routledge, 1994

LOGIC · Open question · Folk paradox, 1940s

Unexpected Hanging

The prisoner proves the surprise execution cannot happen. Then it happens, and he is surprised.

In plain words

A teacher tells the class there will be a surprise test next week, on a day nobody can guess in advance. A clever pupil thinks it through. It cannot be Friday, because by Thursday evening, with no test yet, everyone would know it was coming. So Friday is out. Then it cannot be Thursday either, for the same reason, now that Friday is gone. Day by day she rules out the whole week and relaxes. On Tuesday the test lands on her desk, and she is completely surprised. Her reasoning looked perfect. It was also wrong somewhere, and people still argue about where.

An everyday example

Try it on a younger brother. Tell him you will hide his cricket ball on one day this week and he will not see it coming. Let him work out that Sunday is impossible, then Saturday, and so on. When he proudly announces that you cannot do it at all, hide it on Wednesday. Watch his face.

The short version

A judge promises a hanging next week on a day the prisoner will not expect. The prisoner reasons it cannot be Friday, since by Thursday night he would expect it. Then not Thursday, by the same logic, and so on. He rules out every day. Wednesday arrives, and so does the hangman, unexpectedly.

Where you have seen it

Surprise fire drills, surprise tests, surprise audits. The announcement of a surprise creates the same knot.

The uncomfortable bit

Logicians still disagree about where the reasoning fails. The prisoner's argument looks airtight, and reality ignores it anyway.

Source: Quine, 'On a So-called Paradox', Mind, 1953; popularised by Gardner, Scientific American, 1963

LOGIC · Open question · William Newcomb, 1969

Newcomb's Paradox

Two boxes. A predictor who is almost never wrong. Taking both makes you poorer.

In plain words

A rich stranger puts two boxes in front of you. Box A is see-through and holds 1,000 rupees. Box B is closed. Yesterday, a machine that has guessed right about thousands of people decided what to put in B. If it guessed you would take both boxes, it left B empty. If it guessed you would take only B, it put a crore inside. The money is already there or not, so taking both can only add 1,000. Yet people who take both nearly always walk away with just 1,000, and people who take only B nearly always get the crore. Each choice has a solid argument, and they disagree.

An everyday example

Your phone's shopping app already knows what you will buy before you open it. Now picture it offering a deal: a free gift sits in your cart only if it predicted you would not also grab the discount coupon. The coupon is right there. Grabbing it feels free. But the app has been right about you every single time.

The short version

Box A is transparent and holds a small sum. Box B holds either nothing or a fortune, decided in advance by a predictor who is nearly always right about what you will choose. Take both, and B was almost certainly left empty. Take only B, and it was almost certainly filled.

Where you have seen it

Recommendation engines already predict your choices well enough to make this less hypothetical every year.

The uncomfortable bit

When a philosophy magazine polled readers, the split was nearly even and both sides thought the other was obviously irrational. Two respectable decision theories give opposite answers, and picking one is picking a worldview.

Source: Nozick, 'Newcomb's Problem and Two Principles of Choice', 1969; PhilPapers survey of professional philosophers, 2020

LOGIC · Proven · Kurt Gödel, 1931

Gödel Incompleteness

Any system of maths rich enough to count contains true statements it can never prove.

In plain words

Maths is built from a set of starting rules, and from those rules you prove things step by step. In 1931 Kurt Gödel showed something shocking. He wrote a maths sentence that, in a sneaky way, says this sentence cannot be proved. If the rules could prove it, they would be proving something false, which is a disaster. So they cannot prove it, and that means the sentence is true. There is a true fact the rules cannot reach. This works for any set of rules big enough to do ordinary arithmetic. Add more rules and a new unprovable truth appears. Maths can never be finished.

An everyday example

Try it with a friend. Ask her to write a list of every sentence she can prove true. Then hand her a card that says: this card is not on your list. If she adds it, the card becomes false and her list now holds a falsehood. If she leaves it off, the card is true and her list has missed a truth. She loses either way.

The short version

Gödel built a sentence that effectively says "this statement is unprovable in this system". If the system proves it, the system proves a falsehood. If it cannot, the sentence is true and unprovable. Either way, no consistent system captures all mathematical truth.

How it actually works

The proof rests on three moves. The first, arithmetisation, assigns a number to every symbol, formula and finite sequence of formulas, so that 'the sequence numbered p is a proof of the formula numbered q' becomes a statement about whole numbers the system itself can express. The second is diagonalisation: for any expressible property of formulas, there is a sentence that says of itself that it has that property, and for 'is not provable' that sentence is G. A consistent system cannot prove G, since a proof of G would be a proof that G has no proof; so G is unprovable, and therefore true. The second theorem notes that this reasoning can be carried out inside the system, yielding a proof of 'if the system is consistent then G', so a proof of its own consistency would give a proof of G.

Gödel numbered the basic signs: 0 was 1, the successor sign 3, negation 5, or 7, the universal quantifier 9, and the brackets 11 and 13. A string of signs numbered a, b, c is coded as 2 to the power a times 3 to the power b times 5 to the power c, and unique prime factorisation reads it back: the string of opening bracket, 0, closing bracket becomes 2 to the eleventh times 3 times 5 to the thirteenth, 7,500,000,000,000. Later, natural unprovable statements were found: Jeff Paris and Leo Harrington showed in 1977 that a strengthened finite Ramsey theorem is unprovable in Peano arithmetic.

The paper, Über formal unentscheidbare Sätze der Principia Mathematica und verwandter Systeme I, appeared in the Monatshefte für Mathematik und Physik in 1931. It answered David Hilbert, who in 1900 had asked for a proof that the axioms of arithmetic are consistent and whose programme sought finite, checkable foundations for mathematics. Gödel first mentioned the result at a conference in Königsberg on 7 September 1930; John von Neumann questioned him and derived the second theorem himself within weeks. Gerhard Gentzen proved the consistency of arithmetic in 1936 using an induction principle from beyond arithmetic, as the second theorem requires.

Where you have seen it

It ended the era's grand project of putting all mathematics on one complete, self certifying foundation, and its echo appears in the halting problem.

In films, books and games

Douglas Hofstadter's Gödel, Escher, Bach (1979) is a whole book built round the theorem, and the graphic novel Logicomix (2009) tells the story of Russell's search for the foundations of mathematics up to the moment Gödel's result arrives.

The uncomfortable bit

A system also can never prove its own consistency. Mathematics can be trusted, but it cannot be the one to tell you so.

Source: Godel, 'Uber formal unentscheidbare Satze der Principia Mathematica und verwandter Systeme', Monatshefte fur Mathematik und Physik, 1931

LOGIC · Proven · Alan Turing, 1936

Halting Problem

No program can be written that tells you whether any program will finish or loop forever.

In plain words

Some computer programs finish their job and stop. Others get stuck and run forever. A very useful tool would look at any program and tell you in advance which kind it is. Alan Turing proved in 1936 that no such tool can ever exist. His trick: suppose the tool existed. Build a mischief program that asks the tool about itself, then does the opposite of whatever the tool says. If the tool says it stops, it loops forever. If the tool says it loops, it stops. The tool is wrong either way. So it cannot exist, and no amount of cleverness will ever build one.

An everyday example

A game with a friend. She claims she can always predict whether you will raise your hand. You say you will listen to her prediction first, then do the opposite. She can never be right. A program that waits for the halt checker's answer and then does the opposite plays exactly this trick, which is why your laptop sometimes has to say not responding instead of knowing.

The short version

Turing's proof is self reference again. Feed the supposed halt-checker a program built to do the opposite of whatever the checker predicts about it, and the checker fails. Not for lack of cleverness. In principle, forever.

How it actually works

The proof assumes the checker exists and shows that the assumption contradicts itself. Call the checker H: given any program's text and an input, it correctly reports whether the program stops. From H build a program D that takes a program P, asks H whether P halts when fed its own text, then does the opposite: if H says halts, D loops forever; if H says loops, D stops. Feed D to itself: whichever answer H gives, D does the reverse, so H was wrong. No such H can exist. The shape is Cantor's diagonal argument of 1891: lay out every program against every input, and D is built along the diagonal, disagreeing with each row where a program meets its own description.

The sharpest measure of the ceiling is the busy beaver function, defined by Tibor Radó in 1962. BB(n) is the largest number of steps a halting Turing machine with n states can take. No program can compute it, since knowing BB(n) would settle halting for every n-state machine. BB(1) is 1, BB(2) is 6, BB(3) is 21, BB(4) is 107, and BB(5) is 47,176,870, settled in 2024 by the Busy Beaver Challenge collaboration with a machine-checked proof. In 2016 Adam Yedidia and Scott Aaronson built a 4,888-state machine that halts only if the Goldbach conjecture is false, and later contributors cut it to 27 states. Whether that small machine ever stops is exactly as hard as the conjecture.

Turing's paper was received by the London Mathematical Society on 28 May 1936 and printed in two parts in the second series of its Proceedings, volume 42, with a correction in volume 43 in 1937. It answered the Entscheidungsproblem, posed by David Hilbert and Wilhelm Ackermann in 1928, a request for a procedure that decides any statement of first-order logic. Alonzo Church had reached the same negative answer a few months earlier in the American Journal of Mathematics using his lambda calculus. Turing wrote of machines being circle-free rather than halting; the phrase halting problem was popularised by Martin Davis in his 1958 book Computability and Unsolvability.

Where you have seen it

Why your antivirus cannot perfectly catch all malware, why compilers cannot find every infinite loop, and why some questions about software are answered only by running it.

The uncomfortable bit

This was proved before electronic computers existed. The hard ceiling on computing was mapped before the first machine was built.

Source: Turing, 'On Computable Numbers, with an Application to the Entscheidungsproblem', Proceedings of the London Mathematical Society, 1936

LOGIC · Resolved · Bertrand Russell, 1901

Russell's Barber

The barber shaves everyone who does not shave themselves. Who shaves the barber?

In plain words

A village has one barber. His rule: he shaves every man who does not shave himself, and nobody else. Now ask about the barber's own beard. If he shaves himself, he is a man who shaves himself, so by his rule he must not. If he does not shave himself, he is a man who does not, so by his rule he must. Both answers break the rule. The only way out is that such a barber cannot exist. Bertrand Russell used this story to explain a much bigger crack he found in maths in 1901, about lists that list themselves.

An everyday example

Try a school notice. Make a list titled every list in this school that does not mention itself. Now decide whether the list should mention itself. If it does, it no longer belongs on itself. If it does not, it should be on it. You can also try a library catalogue that lists every catalogue not listing itself. The same knot appears.

The short version

If he shaves himself he belongs to the group he must not shave. If he does not, he belongs to the group he must shave. The popular version of Russell's discovery that the set of all sets that do not contain themselves cannot exist.

Where you have seen it

The version with real casualties: Russell mailed the paradox to the logician Frege just as Frege's life's work on the foundations of maths went to print, and it broke the system at its base.

In films, books and games

Logicomix (2009), a graphic novel by Apostolos Doxiadis and Christos Papadimitriou, tells Russell's life story and puts this paradox, and the letter to Frege, at the centre of it.

The uncomfortable bit

Set theory had to be rebuilt with restrictions on which collections count as sets. The paradox was not a puzzle inside maths. It was a crack in the floor.

Source: Russell's 1902 letter to Frege; Russell, The Principles of Mathematics, 1903, appendix B

LOGIC · Proven · Peter Dirichlet, 1834

Pigeonhole Principle

Ten pigeons, nine holes, so two pigeons share. That triviality proves absurd things.

In plain words

If you have more things than places to put them, at least two things must share a place. Ten pigeons, nine holes: some hole gets two pigeons. It sounds too obvious to bother saying. The surprise is how much it can prove. A human head has fewer than 200,000 hairs. Delhi has millions of people. So there must be two people in Delhi with exactly the same number of hairs, guaranteed, without counting a single hair. Mathematicians use this simple idea to prove things about codes, files and numbers that look impossible to know without checking.

An everyday example

In a class of 13 pupils, two must share a birthday month, because there are only 12 months. Try it in your own class. Pick any 5 cards from a pack: two must be the same suit, since there are only 4 suits. If your drawer holds socks of 4 colours, pulling out 5 socks in the dark guarantees a matching pair.

The short version

If you have more items than containers, some container holds at least two. It sounds beneath mention, and it is one of the sharpest tools in mathematics.

Where you have seen it

It guarantees, right now, at least two people in Bengaluru with exactly the same number of hairs on their head. Heads have at most a couple of lakh hairs, and the city has over a crore of people.

The uncomfortable bit

The same one liner underpins data compression limits and collision proofs in cryptography. There is no such thing as a compressor that shrinks every file.

Source: Dirichlet's Schubfachprinzip, 1834

LOGIC · Open question · Lothar Collatz, 1937

Collatz Conjecture

A rule a child can follow, unsolved for ninety years. Halve it if even, triple and add one if odd.

In plain words

Pick any whole number. If it is even, halve it. If it is odd, multiply by three and add one. Now repeat with the new number. Start with 6: you get 3, 10, 5, 16, 8, 4, 2, 1. Every number anyone has ever tried ends up at 1. The guess is that every number does. Nobody can prove it. Nobody can find a number that escapes either. It has been checked for numbers with twenty digits. Children can play with it. The best mathematicians in the world cannot finish it, and some think it may never be finished.

An everyday example

Try 7 on paper: 7, 22, 11, 34, 17, 52, 26, 13, 40, 20, 10, 5, 16, 8, 4, 2, 1. Now try 27. It takes 111 steps and climbs above 9,000 before falling. Race a friend: each of you picks a number under 50 and counts the steps to 1. Longest chain wins.

The short version

Start with any number and repeat. Every number ever tested, into the quintillions, crashes down to 1. Nobody can prove all of them do, and nobody has found one that escapes.

How it actually works

The rule has a built-in bias downward. Every odd step produces an even number, since three times an odd number plus one is even, so each tripling is followed by at least one halving. If the halvings after a tripling behaved like coin tosses, the average number of them would be two, and the net effect of a tripling and its halvings would be to multiply by about three quarters. That is a heuristic, not a proof, because the parities along a trajectory are not independent. A counterexample would have to be either a number whose trajectory grows without bound or one caught in a cycle other than 4, 2, 1, and neither has been ruled out.

Starting at 27, the sequence takes 111 steps to reach 1 and climbs to 9,232 on the way. Below one million, the longest trajectory belongs to 837,799 at 524 steps. David Bařina reported in 2020, in the Journal of Supercomputing, that every starting number below 2 to the power 68, roughly 2.95 times 10 to the power 20, reaches 1. Terence Tao's 2019 result, published in Forum of Mathematics, Pi, is narrower than it is often summarised. It says that for any function tending to infinity, however slowly, almost every starting number, in the sense of logarithmic density, has a trajectory that at some point falls below that function of the start. It does not say every number reaches 1, and it leaves room for rare exceptions.

Lothar Collatz said he found the problem in 1937 but did not publish it. It spread by word of mouth; Bryan Thwaites met it in 1952 and later offered 1,000 pounds for a proof. Paul Erdős offered 500 dollars. John Conway showed in 1972, in a paper titled Unpredictable Iterations, that a general family of similar rules is undecidable. Jeffrey Lagarias's 1985 survey in the American Mathematical Monthly and his 2010 collection The Ultimate Challenge gather the results. Tomás Oliveira e Silva had verified every start below 5 times 2 to the power 60 by 2009, the record before Bařina's.

Where you have seen it

Mathematicians warn students off it. Erdős said mathematics is simply not ready for such problems, and offered cash for a proof that remains unclaimed in spirit.

The uncomfortable bit

Its cousin problems are provably undecidable, so there is a real possibility this one is true but unprovable. A Gödel sentence you can explain to a ten year old.

Source: Attributed to Collatz, 1937; Lagarias, 'The 3x+1 Problem and Its Generalizations', American Mathematical Monthly, 1985; Tao, 2019 (almost all orbits)

This entry has a playable demo when scripts are on.

LOGIC · Open question · Stephen Cook, 1971

P versus NP

Is checking an answer fundamentally easier than finding one? A million dollars says nobody knows.

In plain words

Some puzzles are quick to check but slow to solve. Hand someone a finished sudoku and they can confirm it is right in a minute. Hand them a blank one and it may take hours. Computer scientists call the quick to solve problems P and the quick to check problems NP. The big question is whether every quick to check problem is secretly also quick to solve, with a trick nobody has found yet. Nearly everyone believes the answer is no. Nobody can prove it. There is a one million dollar prize for a proof either way, unclaimed since 2000.

An everyday example

A jigsaw puzzle. Finished, you can see at a glance that it is right. Unfinished, a 1,000 piece box takes days. Or a school timetable: checking that no teacher is in two rooms at once is easy, but building one with no clashes is a nightmare. Your bank password is safe only because guessing is slow and checking is fast.

The short version

Sudoku solutions are quick to verify and slow to find. P vs NP asks whether every problem with quickly checkable answers also has a quick solving method. Almost everyone believes not. Nobody can prove it.

Where you have seen it

One of the seven Millennium Prize Problems. Delivery routing, chip design, protein folding and exam timetabling all sit inside it.

In films, books and games

The Simpsons episode Treehouse of Horror VI (1995) hides the equation P equals NP in its three dimensional world. The film Travelling Salesman (2012) is about mathematicians who prove it. The Elementary episode Solve for X (2013) has a murder over a claimed proof.

The uncomfortable bit

If P equalled NP, most encryption would collapse and creativity itself would partially mechanise, since recognising a good answer would be enough to generate one. Civilisation is quietly betting it is false.

Source: Cook, 'The Complexity of Theorem-Proving Procedures', STOC, 1971; Clay Mathematics Institute Millennium Prize, 2000

LOGIC · Open question · George Zipf, 1935

Zipf's Law

The second most common word appears half as often as the first. Nobody planned this.

In plain words

Take any long book and count how often each word appears. The most common word, usually the, comes up a lot. The second most common comes up about half as often. The third, about a third as often. The tenth, about a tenth. This neat staircase shows up in every language and in things that have nothing to do with words: how big cities are, how many visitors websites get, how much money people have. Nobody sets it up and nobody fully understands why. Even a monkey hitting random keys produces something like it, which makes it more puzzling, not less.

An everyday example

Try it on a page of your English textbook. Tally every word. The will win, of or and will sit close to half of it, and a long tail of words will appear only once. Then look at cities: Mumbai and Delhi at the top, Bengaluru, Chennai and Kolkata smaller, and hundreds of small towns trailing below.

The short version

In any large text, word frequency follows a strict pattern: rank two has half the frequency of rank one, rank three a third, and so on. The same curve appears in city sizes, website traffic and income distributions.

Where you have seen it

Mumbai, Delhi, Bengaluru and the long tail of Indian cities fit the shape. So does this paragraph.

The uncomfortable bit

A law this universal should have one clean cause, and it does not. Dozens of competing explanations exist, and random typing models reproduce it too, which makes it stranger, not simpler.

Source: Zipf, The Psycho-Biology of Language, 1935; Piantadosi, Psychonomic Bulletin and Review, 2014 (critical review of explanations)

COSMOS · Proven · EPR paper, 1935

Quantum Entanglement

Measure one particle and its partner's state is fixed instantly, across any distance.

In plain words

Two tiny particles can be made in a way that links them. Separate them, one in Delhi and one in Chennai, then measure the first. The moment you do, the second one's result is fixed to match, with no time for a signal to travel between them. Einstein hated this and called it spooky. Experiments have now shown it is real. The catch is that you cannot use it to send a message, because each result on its own looks like a coin toss. Only when the two lists are brought together and compared does the perfect matching show up.

An everyday example

Two friends each get a sealed envelope from the same magician. They travel to different cities, open them, and find the cards always show the same colour, red or blue, chosen at random. Ordinary envelopes would simply have been filled in advance. Real entangled particles pass tests that prove the colour was not decided until the first envelope was opened.

The short version

Two entangled particles behave as one system no matter how far apart. Einstein called it spooky action at a distance and designed the EPR argument to expose it as evidence quantum theory was incomplete. Bell's theorem and decades of experiments answered: the spookiness is real.

How it actually works

Two particles are entangled when their joint state cannot be written as one state for each. In the singlet state of two spin-half particles, a measurement along any axis gives opposite results at the two ends, while each end alone gives an even coin toss. A theory in which both results were fixed at the source must obey John Bell's 1964 inequality. In the 1969 form of Clauser, Horne, Shimony and Holt, a combination S of four correlations at different angles cannot exceed 2; quantum theory predicts up to 2 times the square root of 2, about 2.83. No signal travels because each observer's own results are random whatever the partner does; the correlation is visible only when the two records are brought together, which needs a channel no faster than light.

Alain Aspect, Philippe Grangier and Gérard Roger measured S at 2.697 with an uncertainty of 0.015 in 1982 using photon pairs from a calcium cascade. Those tests left loopholes: the detectors caught only a fraction of pairs, and the settings were not chosen fast enough to rule out communication. In 2015 Bas Hensen and colleagues at Delft entangled electron spins in two diamonds 1.3 kilometres apart, closed both loopholes at once, and over 245 trials found S at 2.42 with an uncertainty of 0.20, published in Nature. A 2008 Geneva experiment by Daniel Salart and colleagues set a lower bound on the speed of any hidden influence of ten thousand times the speed of light.

Einstein, Podolsky and Rosen published in Physical Review in May 1935; Niels Bohr replied in the same journal that October. Bell's paper, On the Einstein Podolsky Rosen Paradox, appeared in the short-lived journal Physics in 1964. Stuart Freedman and John Clauser ran the first experimental test in 1972. Anton Zeilinger's group reported teleportation of a photon state in 1997 and a test with strict spacelike separation in 1998. The 2022 Nobel Prize in Physics went to Aspect, Clauser and Zeilinger for experiments with entangled photons, the violation of Bell inequalities and quantum information science.

Where you have seen it

The 2022 Nobel Prize went to the experimenters who closed the loopholes. Quantum secure communication networks, including satellite links, already run on it.

In films, books and games

Ant-Man and the Wasp (2018) uses it as a plot device: after his trip to the quantum realm Scott Lang is said to be quantum entangled with Janet van Dyne, so that she can speak and act through him.

The uncomfortable bit

It cannot send messages. The results look random at each end and only reveal their correlation when compared later, so relativity survives on a technicality that feels like the universe filing paperwork.

Source: Einstein, Podolsky and Rosen, Physical Review, 1935; Bell, Physics, 1964; Nobel Prize in Physics, 2022

COSMOS · Proven · Werner Heisenberg, 1927

Heisenberg Uncertainty

Position and momentum cannot both be sharp. Not from clumsiness. From the shape of reality.

In plain words

For a tiny particle like an electron, you cannot know both exactly where it is and exactly how fast it is going. Pin down the place sharply and the speed goes blurry. Pin down the speed and the place goes blurry. People often think this is because looking at something so small knocks it about. That is not the reason. A particle is more like a ripple than a marble, and a ripple simply does not have one exact spot and one exact wavelength at the same time. The blur is part of what the particle is, not a problem with our tools.

An everyday example

Tap a guitar string very briefly. You hear a click with no clear note. Let it ring longer and the note becomes clear, but now the sound is spread over time. A sharp moment and a sharp note cannot both happen. Electrons face the same trade, with position in place of the moment and speed in place of the note.

The short version

The uncertainty principle is not about disturbing what you measure. A particle is a wave of possibility, and mathematically a wave cannot have both an exact location and an exact wavelength. The blur is built into the object, not the instrument.

Where you have seen it

Misquoted daily to mean observing people changes their behaviour. That is a real thing, but it is the Hawthorne effect, not Heisenberg.

In films, books and games

Breaking Bad (2008 to 2013) has Walter White choose Heisenberg as his drug name, a nod to the physicist rather than the principle.

The uncomfortable bit

The vacuum itself inherits the blur. Empty space seethes with short lived fluctuations, and their measurable push between metal plates has been detected in the lab.

Source: Heisenberg, Zeitschrift fur Physik, 1927

COSMOS · Open question · Hugh Everett, 1957

Many Worlds

The universe never chooses an outcome. It takes all of them, and splits.

In plain words

In the quantum world, a particle can be in several states at once until it is measured, and then one result shows up. The usual story says the other results just vanish. Hugh Everett said in 1957 that nothing vanishes. Every possible result happens, each in its own copy of the world, and you split along with it. There is a version of you that saw heads and one that saw tails. The copies never meet. It sounds like a comic book, but it is a serious idea that many physicists accept, and no experiment has yet been able to rule it in or out.

An everyday example

Roll a die at the kitchen table. On this view, six kitchens now exist, one for each face, each with a you who saw a different number. None of them can phone the others. What you call luck is just which branch you happen to be reading this in. Nothing at home can test this, which is exactly the complaint against it.

The short version

Everett removed the special rule where measurement collapses possibilities into one result. Take the equations literally and every quantum event branches the universe, observers included. No collapse, no dice, just an endlessly dividing wavefunction.

How it actually works

Quantum mechanics has two rules for how a state changes. Between measurements it evolves under the Schrödinger equation, which is linear: if two states are possible, their sum is too. At a measurement a second rule replaces the sum with one term, chosen at random. Everett's step was to keep only the first rule and apply it to everything, apparatus and observer included. A detector meeting a superposition then ends in a superposition of having registered each result. Branching is what happens next: the detector interacts with air, light and observer, and the components lose the ability to interfere. That process is decoherence, described by Dieter Zeh in 1970. The branches are not places but terms in one state that can no longer affect each other. Being standard quantum mechanics minus the collapse rule, it predicts identical outcomes and frequencies for every experiment.

Measure along z an electron prepared with spin along x: up and down, each with probability one half. After the measurement the state is up with the apparatus reading up, plus down with the apparatus reading down, and soon each term also contains a physicist who remembers one result. Repeat the measurement 1,000 times and there is a branch recording exactly 500 ups and a branch in which all 1,000 came up, with a squared amplitude of about one in 10 to the 301. What probability means when every run happens is the main open problem, addressed through decision theory by David Deutsch in 1999 and David Wallace in 2012, and not settled.

Hugh Everett wrote the theory as a Princeton doctoral thesis under John Wheeler. The long version, The Theory of the Universal Wave Function, was finished in 1956; a cut-down version appeared as 'Relative State' Formulation of Quantum Mechanics in Reviews of Modern Physics in 1957. Bryce DeWitt revived it, used the phrase many worlds in Quantum Mechanics and Reality in Physics Today in September 1970, and edited the 1973 volume that printed the full thesis. Deutsch's The Fabric of Reality in 1997 and Wallace's The Emergent Multiverse in 2012 made it respectable.

Where you have seen it

The multiverse of fiction borrows the name but not the idea. The branches share no portals, and you cannot visit the world where you took the other job.

In films, books and games

Everything Everywhere All at Once (2022) and Rick and Morty (from 2013) run on a multiverse of branching lives, though they add the travel between branches that the real theory forbids.

The uncomfortable bit

Everett's theory was ridiculed and he left physics for defence work. Today it is one of the leading interpretations among working physicists, and there is still no experiment that can tell it apart from its rivals.

Source: Everett, 'Relative State Formulation of Quantum Mechanics', Reviews of Modern Physics, 1957

COSMOS · Open question · Lord Kelvin, 1852

Heat Death

The universe does not end with a bang. It ends with everything the same temperature.

In plain words

Everything that happens needs a difference. Hot tea cools because the room is colder. A river flows because one end is higher. Engines, plants and people all run on differences like these. But every time something happens, a little difference is used up and cannot be got back. Over an unimaginably long time, stars will burn out, the last warmth will spread thin, and the whole universe will end up one even, cold, dark soup. Nothing will be hotter or denser than anything else, so nothing more can happen. Not an explosion. Just a very long quiet.

An everyday example

Pour hot milk into cold coffee and watch it. At first there are swirls and streaks. After a minute it is one even lukewarm cup, and no amount of waiting will unmix it or make one side hot again. The whole universe is that cup, with the stars as the last hot streaks slowly fading into the rest.

The short version

Every process that does anything runs on differences: hot to cold, dense to sparse. Entropy erases differences. Given enough time, stars burn out, black holes evaporate, and the cosmos settles into a uniform lukewarm nothing where no event can occur.

How it actually works

The second law of thermodynamics says that the entropy of an isolated system never falls. Work can only be extracted where there is a difference to exploit: a hot body beside a cold one. Each flow across such a gap shrinks it and raises the entropy, and once every gap has closed no engine, reaction or cell can run, although the total energy is unchanged. Heat death is the loss of usable energy. Gravity alters the detail: matter that clumps under its own weight raises its entropy, so the end state is a dilute scatter of black holes, dead stars and radiation close to absolute zero.

Fred Adams and Gregory Laughlin set out the sequence in a 1997 review. Star formation runs down as gas is locked into remnants, and the smallest red dwarfs burn out after roughly 10 to the 14 years. Whether the remnants then decay depends on proton decay, which has never been observed: Super-Kamiokande in Japan has pushed the lower limit on the proton lifetime beyond 10 to the 34 years. If they do, the remnants dissolve by about 10 to the 40 years and only black holes remain, losing mass through Hawking radiation. One of a solar mass takes about 10 to the 67 years to evaporate, the largest about 10 to the 100.

William Thomson, later Lord Kelvin, published On a Universal Tendency in Nature to the Dissipation of Mechanical Energy in 1852. Hermann von Helmholtz extended the argument to the universe in a Königsberg lecture in 1854. Rudolf Clausius coined the word entropy in an 1865 paper and closed it with two sentences: the energy of the universe is constant, and the entropy of the universe tends to a maximum. The discovery of accelerating expansion in 1998 changed the ending. Lawrence Krauss and Robert Scherrer calculated in 2007 that with dark energy every galaxy outside our Local Group leaves the observable horizon within about a hundred billion years, and the end state is a nearly empty de Sitter space with a horizon temperature near 10 to the minus 30 kelvin.

Where you have seen it

The timescales dwarf comprehension. Star formation ends in trillions of years, and the last black holes evaporate on timescales with hundreds of digits.

In films, books and games

Isaac Asimov's short story The Last Question (1956) follows humanity asking a computer, across trillions of years, whether the running down of the universe can be reversed. The Doctor Who episode Utopia (2007) visits the last humans at the end of the universe.

The uncomfortable bit

Time itself arguably stops meaning anything, because with no change, nothing distinguishes one moment from the next. The universe would not be dead so much as finished.

Source: Thomson (Kelvin), 'On a Universal Tendency in Nature to the Dissipation of Mechanical Energy', 1852; Adams and Laughlin, Reviews of Modern Physics, 1997

COSMOS · Open question · Stephen Hawking, 1976

Black Hole Information

Black holes evaporate. Quantum law says information cannot be destroyed. Pick one.

In plain words

A rule at the heart of physics says information can never be truly destroyed. Burn a letter and, in principle, the ashes and smoke still carry everything needed to rebuild it. Black holes seem to break this. Throw the letter in and it is gone behind a wall nothing can escape. Stephen Hawking then showed that black holes slowly leak a plain glow and finally disappear. If the glow carries no memory of the letter, the rule is broken. If it does, nobody can say how. Physicists have argued about this for fifty years and still have no agreed answer.

An everyday example

Drop your phone in a bonfire. Awful, but a perfect scientist could, in theory, read every photo back from the pattern of the smoke. Now drop it in a black hole. The black hole shrinks away over ages into a featureless glow. Ask the glow where the photos went and, so far, physics has to shrug.

The short version

Hawking showed black holes leak radiation and eventually vanish. If everything they swallowed disappears with them, quantum mechanics loses its most sacred rule. If information escapes, it must somehow ride out in featureless radiation.

Where you have seen it

Hawking bet against information surviving, conceded in 2004, and paid with a baseball encyclopedia, chosen because information can be recovered from it.

In films, books and games

Interstellar (2014) ends with Cooper sending data out from inside a black hole, a plot built on the hope that what falls in is not lost.

The uncomfortable bit

Attempted fixes keep breaking something worse. One serious proposal puts a wall of fire at the horizon, violating relativity's promise that falling in should feel like nothing. Fifty years on, the crisis is still open.

Source: Hawking, 'Breakdown of Predictability in Gravitational Collapse', Physical Review D, 1976; Almheiri et al., JHEP, 2013 (the firewall proposal)

COSMOS · Resolved · James Clerk Maxwell, 1867

Maxwell's Demon

A tiny gatekeeper sorting fast molecules from slow ones could break thermodynamics.

In plain words

A hot gas is just molecules moving fast. A cold one has them moving slowly. In 1867 James Clerk Maxwell pictured a tiny creature at a door between two rooms of gas. It lets fast molecules go one way and slow ones the other. Soon one room is hot and the other cold, for free, and you could run an engine on that. This seems to break the rule that heat never sorts itself out on its own. The answer took about a century. The creature has to remember what it saw, and wiping its memory costs energy. That cost is exactly enough to save the rule.

An everyday example

Think of a shopkeeper sorting mixed coins into a 1 rupee tray and a 5 rupee tray. The sorting looks free, but he has to look at each coin, remember it and move his hand. Even if the moving were free, his head fills up, and clearing it to carry on is where the bill arrives.

The short version

The demon watches a gas and opens a door only for fast molecules, heating one side for free. Entropy falls, the second law dies. It took over a century to find the flaw: the demon's memory fills, and erasing information has an unavoidable energy cost.

Where you have seen it

Landauer's principle, the resolution, now sets the theoretical minimum energy for computing and has been verified in single molecule experiments.

In films, books and games

Thomas Pynchon's novel The Crying of Lot 49 (1966) features the Nefastis Machine, a box said to hold a real Maxwell's demon that only a sensitive person can operate.

The uncomfortable bit

The demon's defeat proved something profound. Information is physical. Every deleted bit heats the universe, including the ones your phone is deleting right now.

Source: Maxwell, Theory of Heat, 1871; Landauer, IBM Journal of Research and Development, 1961; Berut et al., Nature, 2012 (measured erasure cost)

COSMOS · Contested · Brandon Carter, 1973

Anthropic Principle

Why is the universe fine tuned for life? Because if it were not, nobody would be asking.

In plain words

The universe has a few settings, like how strong gravity is, that seem tuned just right for stars, planets and life. Change them a little and nothing interesting could exist. That looks like a miracle. The anthropic answer is a sort of shrug: of course we find the settings friendly, because in an unfriendly universe there would be nobody around to notice. If there are many universes with many settings, only the friendly ones have observers, and we are naturally in one of those. Some people find this deep. Others say it explains nothing and just stops the question.

An everyday example

A fish in a pond is amazed that water is exactly the right wetness for fish. Every lottery winner, asked about the odds, is speaking from the one seat where the odds already paid out. Or think of a class survey asking how many pupils turned up today: the answer is always at least one, because somebody had to be there to be asked.

The short version

Several physical constants sit in narrow ranges where atoms, stars and chemistry are possible. The anthropic answer: observers can only ever find themselves in observer friendly universes, so the tuning is a selection effect, not a miracle.

Where you have seen it

Deployed in every debate about multiverses, design and why the vacuum energy is absurdly small.

In films, books and games

Douglas Adams's puddle, in The Salmon of Doubt (2002), wakes up in a hole that fits it so neatly it concludes the hole must have been made for it, and goes on thinking so even as the sun comes up and it shrinks.

The uncomfortable bit

Critics call it unfalsifiable and closer to a stop sign than an explanation. Even its defenders admit the weak version is nearly a tautology. The argument is about whether tautologies can still be informative.

Source: Carter, IAU Symposium 63, 1974; Barrow and Tipler, The Anthropic Cosmological Principle, Oxford, 1986

COSMOS · Proven · Henri Poincaré, 1889

Three Body Problem

Two orbiting bodies are solvable forever. Add a third and prediction dies.

In plain words

Two things pulling on each other with gravity, like the Earth and the Moon, follow paths you can work out exactly, forever. Add just one more, say a second moon, and the neat formula disappears. Henri Poincaré showed in the 1880s that no general formula exists. Worse, the three paths are so touchy that the tiniest error in where they start grows into a completely different future. This was the first glimpse of what is now called chaos. Computers can predict the paths step by step for a while, but never perfectly, and never forever.

An everyday example

Hang a small magnet on a string above two or three magnets laid on a table and let it swing. With one magnet below, it settles quickly and predictably. With three, it lurches about, and two swings started almost identically end up over different magnets. That is the same chaos on your desk, and a good rainy day experiment.

The short version

Newton solved two bodies completely. For three, Poincaré proved no general closed solution exists, and in the attempt discovered chaos: trajectories so sensitive that any measurement error snowballs into a different fate.

Where you have seen it

Space missions handle it numerically, computing orbits step by step, which is why trajectory corrections mid flight are routine rather than optional.

In films, books and games

Liu Cixin's novel The Three-Body Problem (2008 in Chinese, 2014 in English) takes its name and its alien world from it, and the Netflix series 3 Body Problem (2024) adapts the book.

The uncomfortable bit

The solar system itself is chaotic on long timescales. Planetary positions cannot be reliably predicted beyond a hundred million years, and a small chance exists that Mercury's orbit eventually destabilises.

Source: Poincare, Acta Mathematica, 1890; Laskar and Gastineau, Nature, 2009 (Mercury instability)

COSMOS · Contested · Erasto Mpemba, 1963

Mpemba Effect

A Tanzanian schoolboy insisted hot liquid froze faster than cold. His teacher laughed. He was onto something.

In plain words

Take two cups of water, one hot and one cold, and put both in the freezer. Common sense says the cold one freezes first, because it has less cooling to do. Yet many people have found the hot cup turning to ice first. This is named after Erasto Mpemba, a student in Tanzania who noticed it while making ice cream in the 1960s. Scientists still argue about why it happens, and even whether it always happens. Hot water loses some of itself as steam, swirls about more, and holds less dissolved air. Each of those may help. Nobody has settled the matter.

An everyday example

Try it at home with two identical ice trays. Fill one from the kettle after it has boiled and cooled for a minute, and the other from the tap. Put both on the same freezer shelf and check every twenty minutes. Write down which one has solid cubes first. Repeat it three times, because the result changes from night to night, which is exactly why scientists still argue.

The short version

Mpemba noticed his hot ice cream mix froze before cooler ones and refused to drop it, eventually publishing with physicist Denis Osborne. Under some conditions, hot water genuinely freezes first, though replication is maddeningly inconsistent.

Where you have seen it

Aristotle mentioned the same claim two thousand years earlier. Kitchen experiments worldwide keep the argument alive.

The uncomfortable bit

There may be no single Mpemba effect. Evaporation, dissolved gases, convection and supercooling all plausibly contribute, and a 2016 study argued the effect vanishes with strict controls. The debate itself is a masterclass in how hard "simple" experiments are.

Source: Mpemba and Osborne, Physics Education, 1969; Burridge and Linden, Scientific Reports, 2016 (effect vanishes under strict controls)

COSMOS · Proven · Johann Leidenfrost, 1756

Leidenfrost Effect

On a very hot pan, water drops survive longer than on a merely hot one.

In plain words

Flick a drop of water onto a warm pan and it spreads and hisses away in a few seconds. Flick one onto a much hotter pan and something odd happens. The drop rolls around like a little glass bead and lasts far longer. The bottom of the drop turns to steam the instant it touches the metal. That thin layer of steam lifts the drop and keeps it from touching the pan. Steam carries heat badly, so the drop is protected from the very heat that should destroy it. It is named after Johann Leidenfrost, a German doctor who described it in 1756.

An everyday example

Ask an adult to heat a frying pan on the stove. Sprinkle a few drops of water on it every half minute. At first the drops sizzle and vanish. When the pan is properly hot, they start to skate across it in tiny balls. That skating moment is the effect. Watch from a safe distance and never touch the pan to check.

The short version

Past a threshold temperature, the bottom of the drop vaporises instantly and the droplet levitates on its own cushion of steam, which insulates it. It skitters around for a minute instead of hissing away in seconds.

Where you have seen it

The tava test every Indian kitchen knows: the drop that dances means the pan is properly hot for dosa.

The uncomfortable bit

The same vapour cushion is why a briefly wetted hand can pass through molten metal in industrial demonstrations. Do not test this. The margin between physics trick and injury is thin and unforgiving.

Source: Leidenfrost, De Aquae Communis Nonnullis Qualitatibus Tractatus, 1756

COSMOS · Proven · Gaspard Coriolis, 1835

Coriolis Effect

Cyclones spin opposite ways in each hemisphere. Your bathroom sink does not care.

In plain words

The Earth is spinning. Anything that travels a long way across it, such as wind or ocean water, looks as if it is bending to one side. North of the equator it bends to the right. South of it, to the left. Nothing is pushing the air sideways. The ground underneath is simply turning while the air moves, so the path looks curved to anyone standing on the ground. This bending is what makes big storms spin. Many people believe it also decides which way water swirls down a plughole. It does not. A sink is far too small for the effect to matter.

An everyday example

Sit on a roundabout in a park while a friend stands in the middle and rolls a ball straight towards you. To you the ball appears to curve away, even though your friend rolled it dead straight. Nothing bent the ball. You moved while it travelled. The Earth is the roundabout, and the wind is the ball.

The short version

On a rotating planet, moving air curves: rightward in the north, leftward in the south. Over hundreds of kilometres this organises cyclones and trade winds. Over the thirty centimetres of a sink, it is millions of times weaker than the basin's shape and how you pulled the plug.

Where you have seen it

Cyclones approaching India's east coast spin anticlockwise, as all northern hemisphere storms do. Tourist shows at the equator demonstrating draining direction are theatre.

In films, books and games

The Simpsons episode Bart vs. Australia (1995) has Bart ringing strangers in the southern hemisphere to check which way their toilets swirl, which is the myth, not the science.

The uncomfortable bit

Snipers and long range artillery genuinely correct for it. At those distances the Earth measurably rotates out from under the shot.

Source: Coriolis, Journal de l'Ecole Polytechnique, 1835

COSMOS · Open question · Jerry Ehman, 1977

Wow! Signal

A 72 second radio burst from Sagittarius, exactly where alien broadcasts should be. Never heard again.

In plain words

In 1977 a radio telescope in Ohio was listening to the sky for anything that did not sound like nature. One night it picked up a strong, narrow burst that lasted just over a minute. The astronomer who checked the printout a few days later was so struck that he wrote the word Wow! in the margin. Nobody has heard it since. Telescopes have pointed at the same patch of sky again and again and found silence. It does not look like a star, a satellite or a fault. It might have been a message. It might have been a fluke. Nobody knows.

An everyday example

Think of hearing a single loud knock on your door at night. You open it and no one is there. You wait at the window every night for a year, and the knock never comes back. You cannot prove it was a visitor. You cannot prove it was the wind. That is the position astronomers are in with this one recording.

The short version

The signal hit the frequency of hydrogen, the natural dial-tone of the universe, and was thirty times louder than the background. The astronomer circled it on the printout and wrote "Wow!". Decades of re-listening have found nothing.

Where you have seen it

It remains the strongest candidate signal SETI has ever recorded, and the file is still open.

The uncomfortable bit

A 2017 comet explanation was widely reported and then widely rejected by radio astronomers. The honest status after nearly fifty years is two words: origin unknown.

Source: Ehman, Big Ear Radio Observatory, Ohio State University, 1977; Ehman, 'The Big Ear Wow! Signal: 30th Anniversary Report', 2007

COSMOS · Open question · Coleman and De Luccia, 1980

Vacuum Decay

Empty space might be metastable. If it snaps, a bubble of new physics spreads at light speed.

In plain words

Empty space is not really empty. It has a kind of energy level, like water resting in a bowl. Some physicists think our space is not resting at the very bottom of the bowl, but on a small ledge partway down. If it ever slipped off that ledge, a bubble of a new kind of space would form and spread outwards at the speed of light. Inside it the rules of physics would change, so atoms could not hold together. Nothing could warn us, because no warning can travel faster than light. The good news is that the expected wait is far longer than the age of the universe.

An everyday example

Balance a marble in a small dent near the top of a slope. It sits still and looks safe. One nudge and it rolls down to the real bottom, and it never comes back. Our universe may be the marble in the dent. The question physicists argue about is how deep the dent is, and whether anything could ever nudge it.

The short version

If our vacuum is a false minimum, a quantum event anywhere could nucleate a bubble of true vacuum, expanding at essentially light speed, rewriting chemistry and dissolving everything it touches. Current measurements of the Higgs sit uncomfortably close to the edge.

Where you have seen it

A favourite of physics nightmares because there is no warning. The bubble arrives with the news of its own existence.

In films, books and games

Greg Egan's novel Schild's Ladder (2002) opens with a physics experiment that seeds a new kind of vacuum, which then spreads outward at half the speed of light and swallows one star system after another.

The uncomfortable bit

The calculated timescales exceed the universe's age by absurd margins, and unknown physics likely changes the answer entirely. It is the safest apocalypse ever proposed, and also the only one you would never see coming.

Source: Coleman and De Luccia, 'Gravitational Effects on and of Vacuum Decay', Physical Review D, 1980; Degrassi et al., JHEP, 2012 (Higgs metastability)

MIND · Proven · Henry Beecher, 1955

Placebo Effect

A sugar pill relieves pain. The relief is real, measurable, and partly chemical.

In plain words

A placebo is a pretend medicine, such as a tablet made of sugar, that contains nothing that should work. Yet when people take one and believe it is real, they often feel better. Pain eases, sickness settles, tiredness lifts. This is not people pretending. The brain releases its own natural painkillers when it expects relief, and scientists can measure them. The effect works best on things the brain itself creates, like the feeling of pain. It cannot cure an infection or mend a broken bone. Every new drug has to be tested against a placebo to prove it does more than hope alone.

An everyday example

A mother kisses a child's scraped knee and says it will feel better now. Often it does. The kiss did nothing to the skin, but the child expected comfort, and the brain delivered it. Doctors see the grown-up version every day, when a patient starts feeling better on the drive home from the clinic, before the tablets have even been bought.

The short version

Expectation of treatment triggers genuine biology. Placebo painkillers release the brain's own opioids, and blocking those receptors with drugs blocks the placebo relief too. It is strongest for pain, nausea and fatigue, the symptoms the brain itself constructs.

How it actually works

Two things need separating. The placebo response is everything that improves in the placebo arm of a trial, and much of it owes nothing to the placebo: illnesses run their course, symptoms fluctuate, and patients who enrol at their worst measure better next time regardless. The placebo effect proper is the part caused by the treatment ritual, through expectation and through conditioning, where a body given a real drug repeatedly in one setting comes to respond to the setting alone. Levine, Gordon and Fields showed in the Lancet in 1978 that naloxone removed placebo relief after wisdom tooth extraction, and de la Fuente-Fernández and colleagues showed in Science in 2001 that placebo eased Parkinson's symptoms by releasing dopamine in the striatum.

Beecher's 1955 paper pooled 15 studies with 1,082 patients and reported that 35.2 per cent were satisfactorily relieved by placebo. In 1997 Gunver Kienle and Helmut Kiene reread all 15 and found none had separated a placebo effect from natural recovery, fluctuating symptoms and other treatments. Asbjørn Hróbjartsson and Peter Gøtzsche compared placebo arms with no-treatment arms in 114 trials in the New England Journal of Medicine in 2001. Placebo made no difference to binary outcomes; for continuous outcomes the effect was small and concentrated in pain and other patient-reported measures, about 6.5 mm on a 100 mm pain scale. Their Cochrane review, updated in 2010 to 202 trials, agreed. Ted Kaptchuk's 2010 trial gave 80 irritable bowel patients either no treatment or pills in a bottle labelled placebo. After three weeks 59 per cent of the placebo group reported adequate relief against 35 per cent of the controls.

Henry Beecher, an anaesthetist at Massachusetts General Hospital, published The Powerful Placebo in the Journal of the American Medical Association in December 1955, arguing that every new drug had to be tested against a placebo. The trials he helped make standard later cut his number down. Kaptchuk, at Harvard, has run open-label trials since, in chronic low back pain in 2016 and cancer-related fatigue in 2018, though such trials cannot be blinded and rest on patients' own reports.

Where you have seen it

Bigger pills outperform smaller ones, injections beat tablets, branded beats plain, and red stimulant placebos beat blue ones. The ritual is part of the dose.

In films, books and games

Dumbo (1941) has the magic feather: the crows tell Dumbo it will let him fly, he flies while holding it, and only in the circus finale does he learn that he never needed it.

The uncomfortable bit

Placebos can work even when patients are told they are placebos. And no, it does not shrink tumours. It edits the experience of illness, not the disease underneath, and that boundary is exactly where quackery sets up shop.

Source: Beecher, 'The Powerful Placebo', JAMA, 1955; Kaptchuk et al., PLOS ONE, 2010 (open-label placebo)

MIND · Proven · Colin Cherry, 1953

Cocktail Party Effect

In a roaring room you follow one voice, until someone across the hall says your name.

In plain words

At a noisy party you can pick out the one person talking to you and ignore everyone else. The brain turns the rest of the room into a blur. Yet if someone on the far side says your name, you hear it straight away. This shows the brain never truly switches the other voices off. It keeps listening to all of them quietly, checking for anything important, and lets a few special words through. Your own name is the strongest. Scientists have studied this since the 1950s. Picking one voice out of many turns out to be one of the cleverest things the brain does.

An everyday example

Sit in a busy school canteen and talk with a friend across the table. You can follow every word while forty other conversations go on around you. Now get someone at the next table to say your name in a normal voice. Your head turns before you even decide to look. Try saying a different name and notice that it does not work.

The short version

Attention filters the chaos into a single stream. But the filter is not a wall. Unattended sound is still processed for significance, which is why your own name, and only a few things like it, can break through from a conversation you were not following.

Where you have seen it

Hearing your name across a wedding hall. Sleeping parents surfacing at their baby's smallest sound while ignoring traffic.

The uncomfortable bit

Cherry's puzzle became urgent again in the age of voice assistants. Machines find separating one voice from a crowd monstrously hard, and the wake word is essentially an artificial version of your name breaking through the filter.

Source: Cherry, Journal of the Acoustical Society of America, 1953

MIND · Proven · Gilovich and Savitsky, 2000

Spotlight Effect

Nobody noticed. In the famous study, students wore an embarrassing t-shirt and doubled the real count.

In plain words

Most people feel that everyone around them is watching closely and noticing every small mistake. In fact, other people notice far less than you think. In a well known experiment, students had to walk into a room wearing a T-shirt with an embarrassing face printed on it. Afterwards they guessed how many people had noticed. Their guesses were about twice the real number. This happens because you are always at the centre of your own attention, so you assume you are at the centre of everyone else's too. You are not. Each person is busy being the centre of their own.

An everyday example

You arrive at school with a small ink stain on your shirt and spend the whole day sure that everybody has seen it. At home time, ask three classmates what colour your shirt was. Most will not even remember that. Do the same after you stumble over a word in a class reading, and notice how quickly the moment vanishes for everyone but you.

The short version

We dramatically overestimate how much others notice our appearance, mistakes and awkward moments, because we are the centre of our own attention and assume we occupy the same position in everyone else's.

Where you have seen it

The haircut you agonised over, the stumble in the presentation, the typo you spotted after sending. Ask people a day later. They do not remember.

The uncomfortable bit

The comforting corollary is universal. Everyone else is also too busy starring in their own spotlight to audit yours. Anxiety assumes an audience that never showed up.

Source: Gilovich, Medvec and Savitsky, Journal of Personality and Social Psychology, 2000

MIND · Contested · Bluma Zeigarnik, 1927

Zeigarnik Effect

Waiters remembered every unpaid order perfectly, and forgot them the moment the bill was settled.

In plain words

Jobs you have started but not finished stay stuck in your head. Jobs you have finished seem to fade away almost at once. Bluma Zeigarnik, a psychologist working in the 1920s, noticed that cafe waiters could hold long orders in their minds while the customers were still eating, then lose them once the tables had paid. Her experiments found people remembered interrupted tasks far better than completed ones. The mind seems to keep a little reminder running for anything left open. Not every later test has found the same result, but the idea shaped how television shows, games and phone apps are built.

An everyday example

Stop a video game right before the boss fight, or close a book in the middle of a chase, and notice how often you think about it during dinner. Finish it instead and it hardly crosses your mind. Students use this on purpose: leave a maths problem half done at bedtime and your brain keeps chewing on it while you sleep.

The short version

Zeigarnik observed that interrupted and incomplete tasks stay active in memory while finished ones are released. The open loop keeps a background process running that nags for closure.

Where you have seen it

Cliffhangers, "next episode starts in 5 seconds", unread message badges, and the homework you cannot stop thinking about precisely because you stopped midway.

The uncomfortable bit

Replications have been mixed for a century, but the design world never waited for the verdict. Half your apps are built on keeping loops deliberately open in your head.

Source: Zeigarnik, Psychologische Forschung, 1927; van Bergen, Task Interruption, North-Holland, 1968 (critical review of the replications)

MIND · Contested · Rosenthal and Jacobson, 1968

Pygmalion Effect

Tell teachers that randomly chosen students are about to bloom, and those students' scores rise.

In plain words

What a teacher expects from a child can change how well that child does. In a famous 1960s study, teachers were told that certain pupils were about to make big leaps in learning. The names had actually been picked at random. Months later, those pupils had improved more than their classmates. The teachers had not cheated. They had smiled more at those children, waited longer for their answers, and given them harder questions and more encouragement. The children responded. It is named after a Greek myth about a sculptor whose statue came to life. Later studies find the effect is real but smaller than the story suggests.

An everyday example

A cricket coach decides one new player is the natural talent of the group. That player gets more throws in the nets, more advice, and a longer look before being dropped. A year on, the coach says he was right all along. Some of the skill was there to begin with. Some of it was built by the extra attention he gave.

The short version

Expectations leak. Teachers gave their supposed bloomers more warmth, more feedback and more chances to answer, and the children grew into the fiction. Belief became curriculum.

Where you have seen it

Managers with high expectations getting better performance, and the dark mirror version where written-off students obligingly sink.

In films, books and games

The name is borrowed from George Bernard Shaw's play Pygmalion (1913), later the musical film My Fair Lady (1964), in which a flower seller becomes a lady partly because one man insists on treating her as one.

The uncomfortable bit

The original study has real statistical problems and replications show smaller effects, strongest for young children. The honest summary: expectations matter, less magically than the legend says, and most powerfully on those with the least power.

Source: Rosenthal and Jacobson, Pygmalion in the Classroom, 1968; Jussim and Harber, Personality and Social Psychology Review, 2005

MIND · Proven · Norton, Mochon and Ariely, 2011

IKEA Effect

People value a wobbly box they assembled themselves above a better one built by experts.

In plain words

People place a higher value on things they have put together with their own hands, even when the result is a bit wonky. In experiments, people who built a plain box from a flat pack offered to pay much more for it than others would pay for the same box already assembled. They also rated their own clumsy origami about as highly as an expert's. The reason is that effort feels like proof of worth. Having worked on something, you see it as partly yours. The effect is named after the Swedish furniture shop whose products arrive in flat boxes.

An everyday example

Make two paper boats with a friend, then swap so each of you holds the other's boat. Ask both of you to say which boat is better made. Each of you will usually pick your own, even if the folds are messier. The same thing happens with the slightly lumpy birthday cake you baked, which always tastes better than the shop one.

The short version

Labour breeds love. In experiments, people paid significantly more for their own amateur creations and rated them near expert work. Effort gets counted as evidence of worth, and the object becomes part of the self.

Where you have seen it

Instant cake mixes originally flopped until makers required adding a fresh egg. A little labour turned a shortcut into cooking.

The uncomfortable bit

It only works if the build succeeds. Unfinished or failed assembly produces no attachment at all, which is why products are engineered to make you sweat exactly enough to finish.

Source: Norton, Mochon and Ariely, Journal of Consumer Psychology, 2012. The cake-mix 'add an egg' story is marketing folklore and is not well documented

MIND · Proven · Daniel Kahneman, 1993

Peak End Rule

Adding extra minutes of mild discomfort made patients remember the procedure as better.

In plain words

When you look back on an experience, you do not remember the whole thing evenly. Your memory mostly keeps two moments: the strongest one, good or bad, and the very last one. How long the experience lasted hardly counts at all. So a long holiday with a rough last day feels like a rough holiday, and a boring day with a great finish feels like a good day. Daniel Kahneman and his colleagues showed this in the 1990s with painful medical tests. Patients whose tests were made longer, but ended more gently, remembered them as less unpleasant. The memory and the experience are two different things.

An everyday example

Two children spend the same afternoon at a fair. One has a great ride early and then a long queue, a small bump on the dodgems, and a tired walk home. The other has an ordinary day and a huge ice cream at the gate. Ask them next week which afternoon was better. The second child usually wins, even though the first had more fun in total.

The short version

Memory does not average an experience. It keeps two samples, the most intense moment and the final one, and discards duration almost entirely. A long pleasant trip with a lost-luggage ending is remembered as a bad trip.

Where you have seen it

Why concerts end on the biggest song, why meals end with dessert, and why a smooth checkout can redeem a slow restaurant.

The uncomfortable bit

In the colonoscopy study, extending the procedure with a gentler tail end left patients with better memories and more willingness to return. The remembering self happily overrules the self that actually lived it.

Source: Kahneman, Fredrickson, Schreiber and Redelmeier, Psychological Science, 1993; Redelmeier and Kahneman, Pain, 1996 (the colonoscopy study)

MIND · Contested · Barry Schwartz, 2004

Paradox of Choice

A stall with 24 jams drew crowds. The stall with 6 sold ten times more.

In plain words

Having more options sounds better. Past a certain point, though, extra choices make people less likely to pick anything and less happy with what they picked. In a well known study, shoppers at a supermarket saw a table with a huge range of jam or a small one. The big table drew more people, but far fewer of them bought a jar. Too many options make the decision feel like hard work. They also raise the worry that one of the other options was better. Later research shows this only happens in some situations, so the effect is real but not as sweeping as its name suggests.

An everyday example

Take a friend to a sweet shop with two hundred jars and watch how long they take to pick one. Then offer them a choice of three sweets from your pocket. The second choice takes seconds and they seem happier with it. The same is true of choosing a film with the whole family, which is why so many evenings end with the same old favourite.

The short version

Beyond a point, more options raise the cost of deciding, the fear of choosing wrong and the regret afterwards. Abundance becomes a tax paid in hesitation.

Where you have seen it

Thirty minutes scrolling a streaming platform and watching nothing. Restaurant menus with two hundred dishes and a customer ordering the usual.

In films, books and games

Barry Schwartz's book The Paradox of Choice (2004) is the idea by name, from the 285 kinds of biscuit in his local supermarket to the case that fewer options leave people happier.

The uncomfortable bit

A large meta analysis found the overload effect averages near zero and appears only under specific conditions, complex options, no clear favourite, time pressure. The famous jam study is a special case wearing a universal law's costume.

Source: Iyengar and Lepper, JPSP, 2000 (the jam study); Schwartz, The Paradox of Choice, 2004; Scheibehenne et al., Journal of Consumer Research, 2010 (meta-analysis)

MIND · Proven · Richard Thaler, 1980

Sunk Cost Fallacy

You finish the terrible film because you paid for the ticket. The money was gone either way.

In plain words

Money, time or effort that has already been spent cannot be got back, no matter what you do next. So it should not matter when deciding what to do next. Yet people find that almost impossible. They keep eating a meal they dislike because they paid for it, and keep watching a dull film because they have already sat through an hour. Businesses and governments do the same with projects that have gone wrong. The useful trick is to pretend the spending never happened and look only at what is best from here on. The money is gone whatever you do, and stopping is often the better choice.

An everyday example

You buy a video game with a lot of pocket money and hate it after two hours. Playing on will not bring the money back. It only costs you more evenings. A friend who was given the same game for free would put it down at once. The only difference between you is a payment that is already gone.

The short version

Past investment is unrecoverable and therefore irrelevant to what you should do next. Humans cannot feel this. Money, time and effort already spent keep pulling decisions toward the failing option, throwing good resources after bad.

Where you have seen it

Governments completing doomed projects because so much was already spent. Careers, degrees and relationships extended for the same accounting error.

The uncomfortable bit

The famous supersonic jet gave the fallacy its nickname after two governments kept funding it long past commercial hope. Escape requires a brutal question: if I were starting fresh today, would I choose this?

Source: Thaler, Journal of Economic Behavior and Organization, 1980; Arkes and Blumer, Organizational Behavior and Human Decision Processes, 1985

MIND · Proven · Leon Festinger, 1957

Cognitive Dissonance

Paid one dollar to lie, people came to believe the lie. Paid twenty, they did not.

In plain words

It feels bad to believe one thing and do another. When that happens, the mind usually fixes the discomfort by changing the belief, not the action. In a classic 1950s experiment, people did an extremely boring task and were then paid to tell the next person it was fun. Those paid a small amount later said they had really enjoyed it. Those paid a lot did not. The well paid people had a good excuse for lying. The poorly paid ones had no excuse, so they quietly decided the task was fun after all. We are far better at explaining our actions than at choosing them.

An everyday example

A boy tells everyone he does not care about a match, then his team loses. The next day he says the match never mattered and the other side cheated. Or a girl saves for months to buy a phone that turns out to be slow. Instead of admitting the mistake, she starts listing reasons it is better than her friend's.

The short version

Holding contradictory beliefs, or acting against your beliefs, is mentally painful, so the mind quietly edits beliefs to fit behaviour. The small payment gave no excuse for lying, so participants resolved the tension by deciding the boring task was genuinely fun.

Where you have seen it

Festinger infiltrated a doomsday cult and watched the prophecy fail. The believers did not quit. Many doubled down, deciding their faith had saved the world.

In films, books and games

Aesop's fox, who cannot reach the grapes and decides they were sour anyway, is the oldest telling of it, and Leon Festinger's When Prophecy Fails (1956) the first modern one, following a flying-saucer group through the night the world failed to end.

The uncomfortable bit

The unsettling direction of causation: we do not act on our beliefs nearly as much as we believe in our actions. Behaviour comes first, and justification is manufactured to order.

Source: Festinger, A Theory of Cognitive Dissonance, 1957; Festinger and Carlsmith, Journal of Abnormal and Social Psychology, 1959; Festinger et al., When Prophecy Fails, 1956

MIND · Proven · Klaus Conrad, 1958

Pareidolia

A face on Mars, a god in a chapati, a shocked expression on a plug socket.

In plain words

Pareidolia is seeing faces, animals or shapes in things that are not really there. Clouds look like dragons. The front of a car looks cross or friendly. A burnt patch on a roti looks like a face. This happens because the brain has a fast face-spotting system that fires at anything that roughly looks like two eyes and a mouth. For our ancestors, missing a real face in the bushes was dangerous, while seeing a face that was not there cost nothing. So the system is set to fire too often rather than too rarely. It happens in a fraction of a second, before you can think about it.

An everyday example

Lie on the grass with a friend and take turns spotting animals in the clouds. Then look at the plug sockets, taps and car fronts around your street and decide which look happy, sad or surprised. Nearly everyone agrees on the expressions, which shows the face-spotting part of the brain works the same way in all of us.

The short version

The brain's face detector is tuned for speed over accuracy, firing on anything with two dots above a line. Missing a real face once cost more, evolutionarily, than seeing a thousand false ones, so the alarm is set hair trigger.

Where you have seen it

The 1976 Mars face that launched decades of conspiracy, religious images in toast and tree bark, and the front of every car you have ever assigned a personality.

In films, books and games

The Face on Mars, the most famous example, is at the centre of the film Mission to Mars (2000), where it turns out to be a real alien monument.

The uncomfortable bit

Brain imaging shows the face region responding to illusory faces within a fifth of a second, before conscious thought arrives. You cannot choose not to see them. The socket really is looking at you.

Source: Conrad, Die beginnende Schizophrenie, 1958; Hadjikhani et al., NeuroReport, 2009 (165 ms response to illusory faces)

MIND · Contested · Martin Seligman, 1967

Learned Helplessness

After enough uncontrollable failures, subjects stopped trying even when escape was easy.

In plain words

In the 1960s, psychologist Martin Seligman found that dogs given electric shocks they could not stop would later lie still even when a low barrier was all that stood between them and safety. They had learned that trying was pointless, and they carried that lesson into a situation where it was false. The same pattern shows up in people who fail at something over and over. They stop trying, even when the odds have changed. Seligman later changed his view. Giving up may be the brain's normal reaction to long stress, and what actually has to be learned is that you do have control.

An everyday example

A child fails three maths tests in a row after being taught badly. A new teacher arrives who explains things clearly, but the child no longer even tries, saying they are just bad at maths. The sums have changed. The belief has not. The way out is small wins, so the child sees that effort does make a difference again.

The short version

Seligman's animals, exposed to inescapable stress, later sat passively when a simple exit existed. The conclusion shaped fifty years of psychology: helplessness is learned from experiencing that nothing you do matters.

Where you have seen it

Students who stop attempting maths after early failures, employees gone silent under arbitrary management, citizens who stop reporting problems nobody fixes.

The uncomfortable bit

Seligman himself revised the theory in 2016. The neuroscience suggests passivity is the default response to prolonged stress, and what is actually learned is control. Hope, not helplessness, is the skill.

Source: Seligman and Maier, Journal of Experimental Psychology, 1967; Maier and Seligman, Psychological Review, 2016 (the revision)

MIND · Open question · John Searle, 1980

Chinese Room

A man in a room follows rules to answer Chinese notes perfectly. He understands nothing.

In plain words

Picture a man locked in a room who does not know any Chinese. Slips of paper with Chinese writing are pushed under the door. He has a huge book of rules that tells him which Chinese symbols to write back for each set of symbols he receives. He follows the rules, and the replies he sends out make perfect sense to the Chinese speaker outside. She thinks she is chatting with someone who understands her. He has not understood a single word. The philosopher John Searle used this in 1980 to argue that following rules with symbols, however well, is not the same as understanding what they mean.

An everyday example

A pupil memorises the whole answer key for a Hindi exam without learning a word of Hindi. Show him question 7 and he writes the matching answer perfectly. He passes. Ask him what any of it says and he has no idea. Whether he truly knows Hindi, and whether the pair of pupil plus answer key knows it, is the argument in a nutshell.

The short version

Searle's argument: a person with a rulebook could pass written Chinese conversation without knowing a word, so symbol manipulation alone, however fluent, is not understanding. Syntax is not semantics, and a program is all syntax.

Where you have seen it

Forty years later it is the sharpest question to ask about large language models, which is why the paper is cited more now than in any previous decade.

In films, books and games

In Peter Watts's novel Blindsight (2006), the narrator, a man who has lost the ability to feel, describes his own mind as a Chinese Room.

The uncomfortable bit

The strongest reply says understanding belongs to the whole system, man plus rules plus room, not the man. Searle called that desperate. The debate has not moved much, but the stakes have.

Source: Searle, 'Minds, Brains, and Programs', Behavioral and Brain Sciences, 1980

MIND · Thought experiment · David Chalmers, 1996

Philosophical Zombie

Imagine your exact physical duplicate, behaving identically, with nothing going on inside.

In plain words

A philosophical zombie is not the shuffling movie kind. It is a made-up person who looks, talks and acts exactly like you, atom for atom, but has no inner life at all. There is nobody home. It says ouch when pinched but feels nothing. The philosopher David Chalmers uses this idea to ask a hard question. If we can even picture such a being without contradiction, then feelings are something extra, beyond the working of the body. The idea bites hardest when you look at other people. Their smiles and complaints are all you can ever observe, and a zombie would produce exactly the same ones.

An everyday example

Two friends both eat a very hot chilli. Both go red, both gasp, both reach for water. You are sure your own mouth is burning. You only assume the same is true for your friend. Nothing you can see, not even a brain scan, shows you the burning itself. That gap between what you can see and what is felt is the whole puzzle.

The short version

If such a being is even conceivable, Chalmers argues, then consciousness is something extra beyond physical processes, since the physics could all be present with the experience absent. This is the sharpest framing of the hard problem of consciousness.

Where you have seen it

Every argument about whether an AI that reports feelings actually has them is this thought experiment wearing new clothes.

In films, books and games

Peter Watts's novel Blindsight (2006) is built round it: the alien scramblers are intelligent, capable and utterly without inner experience, and the crew has to decide whether consciousness is anything more than an expensive habit.

The uncomfortable bit

You cannot verify anyone else is not one. All the evidence you have for other minds is behaviour, which the zombie reproduces perfectly. Consciousness may be the one fact that only ever has a single witness.

Source: Chalmers, The Conscious Mind, Oxford University Press, 1996

MIND · Open question · Frank Jackson, 1982

Mary's Room

A scientist knows every physical fact about colour, from inside a black and white room. Then she steps outside.

In plain words

Mary is a scientist who has lived her whole life in a black and white room, seeing the world through a black and white screen. She has learned every fact about colour that science could ever discover: how light works, how eyes work, what happens in the brain when someone sees red. One day she walks outside and sees a red rose. The puzzle is whether she learns anything new in that moment. Most people feel she does. If so, then there are facts about experience that no amount of science could ever have told her. Frank Jackson set this puzzle in 1982.

An everyday example

Try describing the taste of a mango to someone who has never eaten one. You can say sweet, soft, a bit like peach, and give them the full chemistry. Then hand them a slice. The look on their face tells you that something arrived in that bite which all your words did not carry. Mary's puzzle is whether that something counts as a fact.

The short version

Mary knows everything science could ever say about red: wavelengths, retinal responses, brain states. When she finally sees red, does she learn something new? If yes, physical facts are not all the facts.

Where you have seen it

Anyone who has explained a taste, a colour or grief to someone who has not lived it, and watched the description bounce off.

In films, books and games

The film Ex Machina (2014) has Caleb explain Mary's black and white room to Nathan while they argue about whether Nathan's creation can truly feel anything.

The uncomfortable bit

Jackson, the argument's own author, later changed his mind and decided Mary learns no new fact, only a new ability. The room is so well built it eventually trapped its architect.

Source: Jackson, 'Epiphenomenal Qualia', Philosophical Quarterly, 1982; Jackson, 'Mind and Illusion', 2003 (his recantation)

MIND · Proven · Elizabeth Loftus, 1995

False Memory

A quarter of adults were talked into remembering a childhood event that never happened.

In plain words

A memory can feel completely real and still be false. Researchers led by Elizabeth Loftus showed this by telling adults about four childhood events, three true and one invented, in which they had supposedly got lost in a shopping centre. About one person in four came to remember the made-up event, and many added details that nobody had given them. Memory is not a recording. Each time you recall something, the brain rebuilds it, and it can build in things that never happened. This matters in court, where confident witnesses have sent innocent people to prison and been sincerely wrong.

An everyday example

Ask a sibling about a holiday you both took years ago. Retell it together, and slip in one small false detail, like a dog on the beach that was never there. Come back a month later and ask them to describe the day. A surprising number of people will now mention the dog, and will be sure they saw it themselves.

The short version

In the lost-in-the-mall studies, researchers planted a fictional memory using suggestion and family details. Participants did not just accept it. They elaborated, adding sensory detail and emotion to an event that never occurred. Memory is generative, and it does not label its inventions.

Where you have seen it

Overturned convictions built on confident eyewitnesses. Confidence, it turns out, barely correlates with accuracy.

In films, books and games

Blade Runner (1982) has Rachael, whose childhood memories were implanted and belong to someone else, and Total Recall (1990) sells Douglas Quaid a memory of a holiday on Mars that he never took, or perhaps did.

The uncomfortable bit

Loftus's work made her simultaneously one of psychology's most cited scientists and one of its most attacked, because the same findings that free the wrongly convicted complicate testimony from real victims. The science is solid. Its applications cut both ways.

Source: Loftus and Pickrell, 'The Formation of False Memories', Psychiatric Annals, 1995

MIND · Proven · Tversky and Kahneman, 1974

Anchoring

Spinning a rigged wheel changed people's estimates of African countries in the UN.

In plain words

The first number you see sticks in your mind and pulls every later guess towards it, even when that number has nothing to do with the question. In a famous experiment, people spun a wheel that landed on 10 or 65, then were asked what share of countries in the United Nations were African. Those who saw 65 guessed much higher than those who saw 10. Everyone knew the wheel was random. It made no difference. Shops, salespeople and negotiators use this all the time. The number that goes first sets the frame, and everything after is measured against it.

An everyday example

A shopkeeper asks 800 rupees for a bag you both know is worth 300. You bargain hard and feel clever getting it for 450. Try this with friends: ask one to guess how many sweets are in a jar after saying the number 20, and another after saying 200. The second guess will almost always be far higher.

The short version

The first number in view drags every subsequent judgement toward it, even when everyone knows the number is irrelevant. Estimates, negotiations and valuations all start from the anchor and adjust insufficiently.

How it actually works

Tversky and Kahneman's original account was anchoring and adjustment: people start from whatever number is available, move away from it and stop too soon. Later work found that self-generated anchors behave that way but external ones mostly do not. Thomas Mussweiler and Fritz Strack showed in 1999 that comparing a target with an anchor makes the mind search memory for evidence that the anchor could be right, and that evidence stays active when the estimate is made. Anchors bias what people know, not only where they start, which is why obviously random anchors work, why paying for accuracy does not remove the effect, and why warnings help only slightly.

In the 1974 experiment a wheel marked 0 to 100, rigged to stop at 10 or 65, was spun in front of each group. Participants first said whether the percentage of African countries in the United Nations was higher or lower than the number, then gave an estimate. The median was 25 for the group that saw 10 and 45 for the group that saw 65. Birte Englich, Mussweiler and Strack gave experienced German judges the file of a shoplifting case. Before sentencing they threw dice loaded to total 3 or 9, told that the total stood in for the prosecutor's demand in months. Those who threw 3 gave an average of five months; those who threw 9 gave eight.

Amos Tversky and Daniel Kahneman published Judgment under Uncertainty: Heuristics and Biases in Science on 27 September 1974. Gregory Northcraft and Margaret Neale showed in 1987 that estate agents touring a real house valued it according to the listing price they had been handed, while most insisted they had ignored it. Mussweiler and Strack published the selective accessibility model in the Journal of Experimental Social Psychology in 1999, and Englich's dice study appeared in Personality and Social Psychology Bulletin in 2006. Kahneman received the Nobel memorial prize in economics in 2002 and put anchoring at the centre of Thinking, Fast and Slow in 2011.

Where you have seen it

The crossed-out original price, the first figure quoted in salary talks, the "serves 2" on a packet, and every bargaining opener in every Indian market, where both sides know the game and the anchor works anyway.

The uncomfortable bit

Experts are not immune. Judges rolling loaded dice before sentencing in one study handed down sentences that tracked the dice. Knowing about anchoring barely protects you from it.

Source: Tversky and Kahneman, Science 185, 1974; Englich, Mussweiler and Strack, Personality and Social Psychology Bulletin, 2006 (judges and loaded dice)

SOCIETY · Proven · John Nash, 1950

Nash Equilibrium

The point where nobody can improve by changing strategy alone. It is often terrible for everyone.

In plain words

In any game with several players, a Nash equilibrium is a set of choices where no single player can do better by changing only their own move. Everyone is stuck, because moving alone would make things worse for them. That does not mean the result is good. Often everyone would be happier if they all changed together, but nobody dares go first. The mathematician John Nash proved in 1950 that every such game with a limited number of choices has at least one of these resting points. It explains traffic jams, price wars and why rival shops end up next to each other.

An everyday example

Two ice cream carts on a long beach. The best deal for swimmers is one cart at each end. But if one cart moves towards the middle, it steals customers, so the other moves too. They end up side by side in the middle, and neither can gain by moving away. That is why so many rivals cluster on the same corner.

The short version

Nash proved every finite game has at least one such resting point. It is where rational players end up, not where they would choose to be. The prisoner's dilemma's mutual betrayal is a Nash equilibrium. So is every arms race.

Where you have seen it

Petrol pumps clustered on the same crossing, identical airline pricing, two chai stalls side by side on one street while the other end of the market has none.

In films, books and games

The film A Beautiful Mind (2001) has Russell Crowe's Nash work out the idea in a bar while his friends compete for the same woman. The scene is famous, though the plan it shows is not actually a Nash equilibrium.

The uncomfortable bit

Nash wrote the proof at 21 in a doctoral thesis of about 27 pages, spent decades fighting schizophrenia, and won the Nobel Prize 44 years later. The equilibrium concept now underpins auction design, including how governments sell telecom spectrum.

Source: Nash, 'Equilibrium Points in n-Person Games', PNAS, 1950; Nash, Non-Cooperative Games, Princeton doctoral thesis, 1950

SOCIETY · Contested · Vilfredo Pareto, 1896

Pareto Principle

Twenty percent of causes produce eighty percent of effects. Pareto noticed it in his garden peas.

In plain words

In many situations a small share of causes produces most of the results. In the 1890s the Italian economist Vilfredo Pareto noticed that about a fifth of the people in Italy owned about four fifths of the land. The same lopsided pattern keeps turning up: a few customers bring most of the money, a few bugs cause most crashes, a few players score most of the runs. It is usually called the 80/20 rule, but the numbers are only rough. The real lesson is that spreading effort evenly is often a waste, and it pays to find the small share that matters most.

An everyday example

Count how many apps on your phone you actually opened this week. Out of forty, it is usually five or six. Ask a shopkeeper which of his hundreds of items sell the most, and he will name a handful. Then check which few friends you message most. The pattern is everywhere once you start looking, and it tells you where your time really goes.

The short version

Pareto found 80 percent of Italy's land held by 20 percent of families, then the same skew in his pea pods. The pattern recurs wherever outcomes compound: sales, software bugs, road accidents, citations.

Where you have seen it

A fifth of your customers generating most revenue, a fifth of your wardrobe worn most days, a handful of chapters carrying most exam marks.

In films, books and games

Richard Koch's The 80/20 Principle (1997) is a whole book on it, and Tim Ferriss's The 4-Hour Workweek (2007) builds its method on the same rule, cutting the 80 per cent of customers and tasks that bring in 20 per cent of the results.

The uncomfortable bit

The numbers 80 and 20 are not sacred and need not sum to 100. The real content is that effort and results are wildly unequal, so uniform effort is almost always misallocated. The abuse begins when the rough pattern is quoted as a precise law.

Source: Pareto, Cours d'economie politique, 1896; named and generalised by Juran, Quality Control Handbook, 1951

SOCIETY · Proven · Cyril Parkinson, 1955

Parkinson's Law

Work expands to fill the time available. Written as satire, confirmed by every deadline since.

In plain words

A job takes as long as you give it. Give yourself a week to tidy your room and it will take a week. Give yourself an hour and it gets done in an hour. That is Parkinson's Law. It was first written as a joke about government offices that kept hiring more people even though there was less to do. It turned out to be true almost everywhere. People stretch a small job to fit the time, by fussing, checking and waiting. The surprise is that a shorter deadline usually gives the same result, just sooner.

An everyday example

Try it with homework. Tell yourself the maths sheet must be finished before the cricket match starts at five. It gets done by five. On a Sunday with nothing planned, the same sheet drags on all day, with snack breaks and phone checks in between. Same sheet, same you. Only the clock changed.

The short version

Parkinson watched the British colonial office grow its staff while the empire it administered shrank. His mock-scientific essay proposed that officials multiply subordinates and make work for each other regardless of workload.

Where you have seen it

The task that takes a week when given a week and an afternoon when given an afternoon. Committees, inboxes and meetings obeying the same physics.

In films, books and games

C. Northcote Parkinson's own book Parkinson's Law (1957), grown from a 1955 essay in The Economist, sets out the rule with the tale of the Colonial Office growing as the empire shrank. Tim Ferriss's The 4-Hour Workweek (2007) pairs it with the 80/20 rule.

The uncomfortable bit

His second law is deadlier: the time spent on an agenda item is inversely proportional to its cost. Boards nod through the crore-level project and argue forty minutes about the office tea budget, because the tea is the only item everyone understands.

Source: Parkinson, The Economist, 19 November 1955

SOCIETY · Proven · Laurence Peter, 1969

Peter Principle

People are promoted until they reach a job they are bad at. Then they stay there.

In plain words

In most companies, if you do your job well, you get moved up to a bigger job. Do that one well and you move up again. This goes on until you land in a job you are not good at. Then the moving up stops, because nobody promotes a person who is struggling. So you stay there. Over time, lots of jobs end up filled by people who were great at the job below. The surprise is that a system built to reward good work slowly fills the top with people who are stuck.

An everyday example

The best bowler in the school team is made captain. Bowling and setting a field are different skills. She is now a worse bowler, because she is busy, and an unhappy captain, because she never wanted to plan. The team has lost its best bowler and gained a captain who would rather be running in to bowl.

The short version

Promotion rewards performance in the current role, not aptitude for the next one. So everyone rises through jobs they are good at and comes to rest in the first one they are not, where the promotions stop.

Where you have seen it

The brilliant engineer made a miserable manager, the star salesperson now suffocating a sales team, hierarchies quietly crusting over with people at their level of incompetence.

In films, books and games

In the American sitcom The Office (2005 to 2013), Michael Scott was the branch's star salesman before he was promoted to run it, and much of the show is about how badly that job suits him.

The uncomfortable bit

Written as satire, then confirmed with data. A 2018 study of over 200 firms found companies really do promote their best salespeople into management, and those stars make measurably worse managers than weaker sellers who understood people.

Source: Peter and Hull, The Peter Principle, 1969; Benson, Li and Shue, 'Promotions and the Peter Principle', Quarterly Journal of Economics, 2019

SOCIETY · Contested · Joseph Overton, 1990s

Overton Window

Politicians cannot say what they think. They can say what the window currently allows.

In plain words

At any one time there is a set of ideas that people feel are normal to say out loud in politics. Ideas outside that set sound shocking or silly. That set is called the Overton window, after the man who described it. The window is not fixed. It slides. An idea that sounded mad thirty years ago can become an ordinary law, and an idea everyone once agreed with can become rude to mention. Politicians mostly stay inside the window because they want votes. The people who move the window are usually not politicians at all.

An everyday example

Think about smoking on a bus. Older relatives can remember when it was normal and nobody complained. Then some people started saying it should stop, and they were called fussy. Now a ban is normal and lighting up on a bus sounds outrageous. Nothing about smoke changed. What people felt allowed to say changed.

The short version

At any moment only a band of positions is considered acceptable public discourse, running from unthinkable through radical to policy. Politicians mostly operate inside it. Activists, provocateurs and shocks to the system move it.

Where you have seen it

Positions that ended careers twenty years ago now stated on prime time, and formerly mainstream views that have exited the window entirely. Both directions, depending on the topic.

In films, books and games

Glenn Beck's thriller novel The Overton Window (2010) took the idea for its title and its plot.

The uncomfortable bit

The strategic insight is uncomfortable: demanding the extreme makes the merely radical sound moderate. Some of the loudest voices in any debate are not trying to win the argument. They are moving the window for the argument behind them.

Source: Formulated by Joseph Overton at the Mackinac Center for Public Policy in the 1990s; written up by Lehman, Mackinac Center, 2010

SOCIETY · Proven · Jerry Harvey, 1974

Abilene Paradox

The whole family took a miserable trip that not one of them wanted.

In plain words

A group can end up doing something that nobody in the group actually wants. It happens when each person thinks the others want it, so they go along to be nice. Everyone says yes. Nobody says what they really think. Then they all sit through the thing, quietly miserable, each believing they are the only one who did not want to come. The name comes from a family that drove a long way to a town called Abilene for a bad lunch. Afterwards they found out not one of them had wanted to go.

An everyday example

Five friends are deciding where to eat. One suggests the new pizza place because he thinks the others like pizza. The rest nod, each thinking the others are keen. They go. The pizza is cold and everyone is quiet. On the walk home someone finally admits they wanted chaat. So did everybody else. Try a secret vote next time.

The short version

Harvey's family drove hours through Texas heat to a mediocre meal because each person agreed to what they believed the others wanted. Groups routinely act against every member's preference, because everyone mistakes silence for enthusiasm.

Where you have seen it

The project everyone privately doubts, the wedding function nobody enjoys but nobody cancels, the standing meeting that outlived its purpose years ago.

The uncomfortable bit

It is the opposite of groupthink, and more embarrassing. Groupthink is failing to disagree. Abilene is a group unanimously implementing a decision that a single honest sentence would have killed.

Source: Harvey, 'The Abilene Paradox: The Management of Agreement', Organizational Dynamics, 1974

SOCIETY · Contested · Wilson and Kelling, 1982

Broken Windows

One unrepaired window signals nobody cares, and the rest of the glass follows.

In plain words

If one window in a building is broken and nobody fixes it, more windows soon get broken. The first broken window sends a message: nobody here cares. People who would never throw the first stone feel fine throwing the second. The same thing works with litter, graffiti and rule breaking of all kinds. Small mess invites bigger mess. Scientists have tested this on real streets and it holds up. The argument is about what to do next. Fixing windows fast is one answer. Punishing tiny offences harshly is another, and that one has a much worse record.

An everyday example

Watch a school corridor. On a clean wall hardly anyone sticks a poster. Once one poster is stuck up and left, ten more appear within a week. Or look at a street corner. One dumped bag of rubbish becomes a pile in days, because each new person thinks the corner is already a dump. Clear it early and it stays clear.

The short version

Visible small disorder, the theory claims, licenses larger disorder by signalling that norms are not enforced. Dutch experiments confirmed the mechanism: litter and graffiti roughly doubled rates of other small violations by passersby.

Where you have seen it

The clean metro that stays clean, the wall that gathers posters the day the first one survives, the code base where one tolerated hack breeds fifty.

In films, books and games

Malcolm Gladwell's The Tipping Point (2000) tells the New York story, the graffiti scrubbed off every subway carriage and fare-dodgers arrested, as broken windows in action.

The uncomfortable bit

The mechanism is real. The policing built on it is another story. Aggressive enforcement of petty offences in 1990s New York claimed credit for a crime drop that occurred in cities that did nothing similar, while the human cost of the crackdowns landed on the poorest neighbourhoods.

Source: Wilson and Kelling, The Atlantic, 1982; Keizer, Lindenberg and Steg, Science, 2008 (the Dutch field experiments); Harcourt and Ludwig, University of Chicago Law Review, 2006 (critique of the policing)

SOCIETY · Proven · Nassim Taleb, 2007

Black Swan

The turkey is fed generously for a thousand days. Its confidence peaks the week before the feast.

In plain words

For hundreds of years, people in Europe had only ever seen white swans. So they were sure all swans were white. Then explorers reached Australia and found black ones. One sighting undid centuries of certainty. A black swan event is any huge surprise like that: nobody saw it coming, it changes everything, and afterwards everybody says it was obvious. The scary part is how we plan. We look at the calm past and assume tomorrow will match it. The past cannot warn you about the thing that has never happened yet.

An everyday example

A shopkeeper has sold about the same number of umbrellas every monsoon for twenty years and orders the same stock. One year the rain fails and he is stuck with boxes. Another year a cyclone sends everyone rushing for umbrellas and he sells out in a day. His twenty years of records never included either. Keep some money aside for the surprise.

The short version

A black swan is an event outside all experience, with massive impact, rationalised afterwards as predictable. Taleb's charge: our models are built from the thousand calm days and are structurally blind to the day that matters.

Where you have seen it

Europeans knew all swans were white until Australia. Financial models pricing risk from recent history, then a 2008. Pandemic plans gathering dust until 2020.

In films, books and games

Nassim Nicholas Taleb's The Black Swan (2007) is the idea by name, and made it a household phrase in the financial crisis that followed a year later.

The uncomfortable bit

The trap is that after each one, experts explain why it was obvious, which rebuilds the confidence that guarantees the next surprise. Taleb's advice is not prediction but posture: build things that survive being wrong.

Source: Taleb, The Black Swan, Random House, 2007

SOCIETY · Contested · Mandelbrot and Taleb, 2012

Lindy Effect

For ideas and books, age predicts longevity. What survived fifty years likely survives fifty more.

In plain words

For things that do not wear out, like ideas, books, games and recipes, being old is a good sign of lasting even longer. A book that has been read for two hundred years will probably be read for another two hundred. A book that came out last month may vanish by next year. This is the opposite of how living things work. An old dog has fewer years left than a puppy. But an old idea has proven it can survive fashions, wars and boredom, so each extra year makes it a safer bet. That is the Lindy effect.

An everyday example

Look at the games children play in a park. Tag, hide and seek and skipping are hundreds of years old and still played every evening. A phone game everyone had two years ago has already been deleted. If you want to guess which game will still be played when you are a grandparent, bet on the old one.

The short version

Perishable things die on schedule. Non perishable things, technologies, texts, institutions, show the reverse pattern: every year survived is evidence of robustness, so expected remaining life grows with age.

Where you have seen it

The wheel, the chair, arithmetic, the Panchatantra. The safest bet about what people use in 2100 is boringly ancient.

In films, books and games

Nassim Nicholas Taleb's Antifragile (2012) gave the effect its second life, using it to argue that a book still in print after fifty years will probably be in print for another fifty.

The uncomfortable bit

It is a heuristic, not a law, and it says nothing about quality, since bad ideas can be durable too. But as a filter it is brutal: most of what is urgent this year is, by Lindy, already dead.

Source: Goldman, The New Republic, 1964; Mandelbrot, The Fractal Geometry of Nature, 1982; Taleb, Antifragile, 2012

SOCIETY · Proven · G.K. Chesterton, 1929

Chesterton's Fence

Do not remove the fence until you know why someone built it.

In plain words

You are walking down a path and find a fence across it. There seems to be no reason for it. You want to pull it down. Chesterton's advice is simple. First find out why somebody built it. Fences do not appear by accident. Someone had a reason, and the reason might still matter. Maybe it keeps cattle out of a field. Maybe it stops a landslide. If you learn the reason and it no longer applies, remove the fence happily. The lesson is about asking before deleting, and it works for rules, old habits and settings on a computer.

An everyday example

A new student finds a rule that the school water tank is only filled after four o'clock. It sounds silly and she asks to change it. The caretaker explains that the pump trips the electricity if the science lab is running at the same time. Now she understands, and she can suggest a better fix. Ask why first, then change things.

The short version

Chesterton's rule for reformers: a rule whose purpose you cannot see was still put there by someone solving a problem you may not be able to see either. Understanding must precede demolition, or the old problem returns wearing the reformer's clothes.

Where you have seen it

The config line deleted as useless that takes production down. The ancient bureaucratic step that turns out to be the only fraud check. Traditional farming practices dismissed, then rediscovered by agronomists.

In films, books and games

G. K. Chesterton's The Thing (1929) is where the fence stands: the reformer who wants it cleared away is told to go and find out why it was put there first.

The uncomfortable bit

It is an argument for curiosity, not conservatism. Chesterton's reformer is told to go find the reason, and then, if the reason is dead, tear the fence down with confidence.

Source: Chesterton, The Thing, 1929, chapter 4

SOCIETY · Proven · Robert Hanlon, 1980

Hanlon's Razor

Never attribute to malice what is adequately explained by incompetence.

In plain words

When something goes wrong for you, it is tempting to think someone did it on purpose. Hanlon's razor says to stop and check a simpler explanation first: they made a mistake, forgot, or were careless. Mistakes happen all the time. Plots against you are rare, because plotting takes planning, teamwork and secrecy, and most people cannot manage all three. So the boring explanation is usually the right one. The rule does not say bad people never exist. It says do not assume the worst until the boring reasons have run out.

An everyday example

Your friend does not reply to your message all day. Your first thought is that he is angry with you. Before you sulk, run through the plain reasons. His phone died. He read it while busy and forgot. The message never sent. Nine times out of ten it is one of those, and a second message sorts it out.

The short version

Conspiracies require coordination, secrecy and follow through, which are rare. Error, laziness and miscommunication are abundant. When wronged, the odds heavily favour cock-up over plot.

Where you have seen it

The unanswered email that was buried, not hostile. The government form that hurts you through bad design, not persecution. Most office politics, on inspection.

In films, books and games

A near-identical line was already in Robert Heinlein's story Logic of Empire (1941), where a character is told he has attributed to villainy what is only the product of stupidity.

The uncomfortable bit

Its cousin, Occam's razor, does the general work: prefer the explanation requiring fewest assumptions. Neither is a law of nature, and both fail against actual bad actors, but as default settings they save relationships and careers.

Source: Hanlon, in Arthur Bloch, Murphy's Law Book Two, 1980

SOCIETY · Contested · Attributed to Ward Cunningham

Cunningham's Law

The fastest way to get a right answer online is to post the wrong one.

In plain words

If you ask a question on the internet, you might get no reply. If you post a confident wrong answer instead, people rush in to correct you. That is Cunningham's law. It works because putting someone right feels good, and lots of people cannot let a mistake stand. So the wrong statement pulls out experts who would have ignored a polite question. It is a bit cheeky, and it only works while people still care about being right. The joke is that Ward Cunningham, the man it is named after, says he never said it.

An everyday example

Post in a family group chat that the new bus to the station leaves at seven in the morning, when you really do not know. Within minutes an uncle replies with the exact timetable, the fare and a photo of the board. Ask politely instead and the message might sit unanswered all day.

The short version

Questions invite effort. Errors invite correction, and correction is irresistible. A confident wrong statement summons experts who would never have answered a polite request.

Where you have seen it

Forum threads where the real answer arrives as a rebuttal. Deliberately wrong posts engineered to bait experts into free consulting.

The uncomfortable bit

Cunningham, who invented the wiki, has disowned the law named after him and calls it wrong. Which means the law's own naming may be a demonstration of the law: someone posted it wrongly attributed, and the correction is still propagating.

Source: Coined by Steven McGeady, who attributed it to Ward Cunningham. Cunningham has publicly rejected both the attribution and the claim

SOCIETY · Proven · Kenneth Arrow, 1951

Arrow's Theorem

No voting system can be fair. Not "none found yet". Proven impossible.

In plain words

When people vote between two choices, the one with more votes wins and that seems fair. With three or more choices it gets strange. You can have a group where most people prefer A to B, most prefer B to C, and most prefer C to A. There is no clear winner, only a loop. A mathematician called Kenneth Arrow proved that no voting method using ranked choices can meet every simple fairness rule at once. Every method breaks at least one rule. Choosing the voting system is choosing which unfairness you can live with.

An everyday example

Three friends pick a film. Asha likes action, then comedy, then horror. Ben likes comedy, then horror, then action. Chitra likes horror, then action, then comedy. Vote action against comedy and action wins two to one. Comedy beats horror two to one. But horror beats action two to one. Whoever chooses the order of the votes chooses the film. Try it at home.

The short version

Arrow showed that with three or more options, no ranked voting method can simultaneously satisfy a short list of obviously reasonable fairness conditions. Every system must violate at least one, so every election method is a choice of which flaw to accept.

Where you have seen it

Condorcet spotted the seed in 1785: majorities can prefer A over B, B over C, and C over A, a loop with no legitimate winner.

The uncomfortable bit

This is not an argument against democracy. It is a proof that "the will of the people" is not a single mathematical object, and that whoever chooses the voting rules holds real, quantifiable power.

Source: Arrow, 'A Difficulty in the Concept of Social Welfare', Journal of Political Economy, 1950; Social Choice and Individual Values, 1951

SOCIETY · Open question · Philippa Foot, 1967

Trolley Problem

Pull the lever and one dies instead of five. Push the man off the bridge and everyone recoils.

In plain words

A runaway trolley is rolling down a track towards five people who cannot move. You are standing by a lever. If you pull it, the trolley goes onto another track, where there is one person. Do nothing and five die. Pull the lever and one dies, because of you. Most people say pull it. Now change one thing. Instead of a lever, you can stop the trolley only by pushing a large man off a bridge onto the track. Same numbers: one dies, five live. Most people now say no. The puzzle is why the same sum feels so different.

An everyday example

A school bus driver sees a ball roll out in front of the bus with a child chasing it. Braking hard will hurt the forty children on board. Swerving hits a parked car with someone inside. Every driver has a split second version of this, and the law and our gut do not always agree on the answer.

The short version

The two cases have identical arithmetic, one life for five, yet most people pull the lever and refuse the push. The pair exposes that our moral machinery runs on more than outcomes: using someone as an instrument feels categorically different from redirecting a harm.

Where you have seen it

Sixty years of philosophy seminars, and now engineering meetings, since driverless cars encode answers to versions of it whether anyone admits it or not.

In films, books and games

The Good Place devotes an episode, The Trolley Problem (2017), to it, with Chidi at the controls and the lever made real. The film Eye in the Sky (2015) is the same dilemma with a drone strike and one girl selling bread. Superhero films use it constantly: in The Dark Knight (2008) the Joker gives two ferries each other's detonator.

The uncomfortable bit

A global online experiment collected around 40 million judgements and found cultures disagree systematically, on sparing the young, the many, the lawful. Whichever ethics gets programmed, it will be somebody's ethics, shipped worldwide.

Source: Foot, Oxford Review, 1967; Thomson, The Monist, 1976 (the footbridge case); Awad et al., 'The Moral Machine Experiment', Nature, 2018

SOCIETY · Open question · Karl Popper, 1945

Paradox of Tolerance

Unlimited tolerance destroys tolerance. The open society must sometimes refuse.

In plain words

A tolerant society lets people hold all sorts of views. But some groups use that freedom to work towards ending freedom for everyone else. If the society tolerates them fully, they may win and end tolerance for good. So a tolerant society has to be able to say no to the people who would destroy it. That is the puzzle: to protect tolerance you sometimes have to refuse it. The philosopher Karl Popper put it this way and added a warning. Argue first. Only shut a group out when it answers arguments with violence.

An everyday example

A school debating club welcomes every opinion. One member keeps shouting over others and says the club should be closed and only his friends allowed to speak. If the club tolerates him every week, the club dies. Keeping the club open means telling him he cannot behave like that. The club is not being unfair by protecting the space where fairness happens.

The short version

Popper argued that if a tolerant society extends tolerance to movements dedicated to ending tolerance, the intolerant win by default. The society must therefore claim the right, in extremis, to not tolerate the intolerant.

Where you have seen it

Every debate about banning parties, deplatforming speakers and the limits of free speech eventually arrives at this passage, usually as a screenshot.

In films, books and games

Karl Popper's The Open Society and Its Enemies (1945) states it in a footnote to the first volume, written in wartime New Zealand.

The uncomfortable bit

The screenshot omits Popper's own restraint: suppress only movements that refuse rational debate and answer arguments with fists. He put counter-argument first and suppression last. Both sides of the modern debate quote half of him.

Source: Popper, The Open Society and Its Enemies, 1945, note 4 to chapter 7

SOCIETY · Contested · Richard Easterlin, 1974

Easterlin Paradox

Countries got dramatically richer. Reported happiness barely moved.

In plain words

Inside a country, rich people usually say they are happier than poor people. So you would expect a country that gets much richer over fifty years to become much happier. Mostly it does not. Reported happiness barely moves. The reason seems to be comparison. We measure our lives against neighbours and against last year. When everyone gets richer together, nobody feels ahead. And a new comfort quickly becomes normal. Economists still argue about how strong the effect is, but the treadmill feeling is familiar to almost everyone.

An everyday example

A family gets its first fridge and everyone is delighted for a month. A year later it is just the fridge. Then the neighbours buy a bigger one and suddenly yours feels small. Nothing about your fridge changed. Try noticing when a treat you once longed for has quietly turned into something you expect.

The short version

Within a country at one moment, richer people are happier. Yet as whole nations grow wealthier over decades, average happiness climbs weakly or not at all. Easterlin's explanation: we judge our lives against neighbours and against yesterday, and both benchmarks rise with us.

Where you have seen it

Rising incomes accompanied by stable complaint. The raise that thrilled for one month before becoming the new baseline.

The uncomfortable bit

The paradox is genuinely contested. Later economists with bigger datasets found happiness does track income, just logarithmically, meaning each doubling buys the same small bump. Either way the treadmill is real: absolute wealth satisfies less than relative position.

Source: Easterlin, in Nations and Households in Economic Growth, 1974; Stevenson and Wolfers, Brookings Papers on Economic Activity, 2008 (the rebuttal)

SOCIETY · Proven · William Jevons, 1865

Jevons Paradox

More efficient engines were supposed to reduce coal use. Britain burned more than ever.

In plain words

You would think a machine that uses less fuel would mean less fuel used overall. Often the opposite happens. When something becomes cheaper to run, people run it more, buy more of them, and find new uses for it. The savings get eaten by the extra use, and total use goes up. William Jevons noticed this in the 1860s, when better steam engines led Britain to burn more coal, not less. The surprise is that efficiency alone does not shrink use. It can make a thing so cheap that use explodes.

An everyday example

A family swaps its old air conditioner for one that uses half the power. The bill should halve. Instead they now run it all night, and put one in the second bedroom too, because it is cheap. The electricity bill ends up higher than before. Check your own bills after buying something efficient.

The short version

Efficiency lowers the effective price of using a resource, and cheaper use invites more use. When the rebound is large enough, total consumption rises. Jevons watched it happen with steam engines and coal.

Where you have seen it

Wider highways filling with traffic, LED lighting leading to more of the world lit at night, fuel efficient cars driven further, energy efficient ACs cooling more rooms.

The uncomfortable bit

It haunts climate policy. Efficiency alone, without caps or prices on the resource, can quietly finance the growth it was meant to prevent. The fix is not less efficiency. It is refusing to let the savings leak back into consumption.

Source: Jevons, The Coal Question, 1865, chapter 7

SOCIETY · Proven · Stanley Milgram, 1967

Six Degrees

Letters given to strangers in Nebraska reached a Boston target in about six hops.

In plain words

Pick any two people on Earth. It seems they should be miles apart, socially. Yet you can usually link them through a short chain of friends of friends, about six steps long. A scientist called Stanley Milgram tested this by giving letters to people in one part of America and asking them to pass them on, only through people they knew, towards a stranger far away. The letters that arrived took about six hands. The reason is that a few people have friends in far away places, and those long links shrink the whole world.

An everyday example

Try it at school. Pick a famous cricketer. Your cousin's friend plays club cricket. The club coach once played with a state player. That state player shares a dressing room with the famous one. That is four steps and you never left your own contacts. Draw the chain on paper and see how few steps it takes.

The short version

Milgram asked random people to forward a letter toward a distant stranger through personal acquaintances only. Chains that completed averaged around six links. Watts and Strogatz later showed why: a few long-range shortcuts make any clustered network astonishingly small.

Where you have seen it

A 2011 study of the largest social network measured the average separation between its hundreds of millions of users at under five hops, and shrinking.

In films, books and games

John Guare's play Six Degrees of Separation (1990), filmed in 1993 with Will Smith, gave the idea its popular name. The party game Six Degrees of Kevin Bacon, linking any actor to Kevin Bacon through shared films, started in 1994.

The uncomfortable bit

The catch in the original study: most letters never arrived. The world is small if people bother to pass the message. The six degrees exist. The willingness to be a link is the scarce part.

Source: Milgram, Psychology Today, 1967; Travers and Milgram, Sociometry, 1969; Backstrom et al., 'Four Degrees of Separation', 2012

LIFE · Contested · David Strachan, 1989

Hygiene Hypothesis

Children from larger families got less hay fever. Too clean might mean less trained.

In plain words

Children with lots of older brothers and sisters get fewer allergies. A doctor noticed this and wondered why. His guess: the body's defence system needs practice. Big families share more germs, so the young immune system learns early what is harmless and what is dangerous. In very clean homes it gets less practice and may start attacking harmless things like pollen or peanuts. That is the hygiene hypothesis. It does not mean stop washing hands. Washing stops real diseases. The missing thing seems to be contact with soil, animals and ordinary muck.

An everyday example

Two children in the same city. One grows up on the edge of town with a dog, a garden and a grandmother who keeps goats. The other lives in a flat, plays indoors and has everything wiped. Studies suggest the first child is less likely to get hay fever. Let children play outside and get dirty, then wash before dinner.

The short version

Strachan noticed allergies fell with the number of older siblings. The refined version: immune systems calibrate against early microbial exposure, and environments scrubbed of old companions, soil microbes, farm animals, fermented foods, leave the system undertrained and trigger happy.

Where you have seen it

Allergy and autoimmune rates climbing fastest in the most sanitised societies, and farm-raised children showing measurably lower rates.

The uncomfortable bit

Scientists now prefer "old friends hypothesis", because the enemy is not soap and the answer is not skipping handwashing. Hygiene against pathogens saves lives. The missing ingredient is dirt-adjacent biodiversity, which is a different thing from disease.

Source: Strachan, 'Hay Fever, Hygiene, and Household Size', BMJ, 1989; Rook, the 'old friends' reframing, Clinical and Experimental Immunology, 2010

LIFE · Proven · Richard Owen, 1840s

Convergent Evolution

Evolution invented the eye about forty separate times. It keeps finding the same answers.

In plain words

Animals that are not related at all often end up looking alike. Sharks are fish. Dolphins are mammals. Both have the same smooth shape, because that shape is the best way to move through water. Birds, bats and insects all fly with wings that were invented separately. Eyes like ours have appeared in many different animal families. This is called convergent evolution. It happens because nature keeps setting the same tests, like water, air or catching food, and there are only a few good answers. Different animals keep landing on the same one.

An everyday example

Look at a pencil, a pen and a paintbrush. Three different tools, made by different people, but all long and thin because that is the easiest shape for a hand to hold. Nature does the same thing. Next time you are at a zoo, compare a sugar glider and a flying squirrel and see if you can tell them apart.

The short version

Unrelated lineages facing the same problem repeatedly arrive at the same solution. Camera eyes in humans and octopuses evolved independently. Powered flight arose in insects, pterosaurs, birds and bats, four separate times.

Where you have seen it

Dolphins and extinct marine reptiles sharing one torpedo shape. Australia's marsupials producing lookalikes of wolves, moles and flying squirrels with no close kinship.

The uncomfortable bit

The octopus eye is arguably better wired than ours, with no blind spot. And convergence carries a cosmic hint: if physics keeps funnelling life toward the same forms here, alien life may be less alien than fiction assumes.

Source: Owen's homology and analogy distinction, 1843; Land and Nilsson, Animal Eyes, Oxford, 2002 (independent eye origins); Conway Morris, Life's Solution, 2003

LIFE · Proven · Cann, Stoneking, Wilson, 1987

Mitochondrial Eve

Every human alive carries mitochondrial DNA from one woman in Africa roughly 150,000 years ago.

In plain words

Inside each of your cells are tiny parts called mitochondria, and they carry their own small bit of DNA. You get it only from your mother, who got it from hers, and so on back. Follow that mother-to-mother line for every person alive today and all the lines meet at one woman who lived in Africa long ago. She is called Mitochondrial Eve. She was not the first woman and she was not alone. Many other women lived then too. She is simply the one whose daughter line never broke. That is a matter of maths, not a lucky find of bones.

An everyday example

Think of a surname passed from father to son. In a village, some surnames die out when a family has only daughters. After centuries only a few names remain, and everyone with one of them traces back to one man. Mitochondrial DNA works the same way, only through mothers. Ask your grandmother about her mother's mother and see how far back you get.

The short version

Mitochondria pass only from mother to child, so their DNA traces one unbroken maternal thread. All living lineages converge on a single woman. She is a mathematical certainty, not a discovery of bones.

Where you have seen it

The 1987 paper's African root became a pillar of the out-of-Africa account of human origins.

In films, books and games

The novel Parasite Eve (1995) and the PlayStation game of the same name (1998) take their title from her, with the mitochondria in every human cell waking up as a single ancient organism that calls itself Eve.

The uncomfortable bit

She was not alone and not the first woman. Thousands of others lived alongside her and left descendants too, just not through unbroken daughter-to-daughter chains. Her title can even migrate to a different woman as lineages die out. There is a Y-chromosome Adam as well, and he lived in a different era. They never met.

Source: Cann, Stoneking and Wilson, Nature 325, 1987; Poznik et al., Science, 2013 (Y-chromosome dating). Current estimates place her nearer 150,000 to 200,000 years ago

LIFE · Contested · Amotz Zahavi, 1975

Handicap Principle

The peacock's tail is glorious precisely because it is a terrible idea.

In plain words

A peacock's tail is heavy, bright and easy for a tiger to spot. It makes the bird slower and easier to catch. So it is strange that peahens like it so much. The idea is that the tail is a truthful advert. Only a very healthy, well-fed bird can grow a giant tail and still survive. A weak bird could not afford one. The cost is what makes the message trustworthy. Scientists still argue about how often this works in nature, but the basic thought is that showing off is believable only when it is expensive.

An everyday example

A boy at school buys the most expensive cricket bat in the shop. Everyone knows he could only do that if his family has money to spare. A cheap bat says nothing. Or watch a friend who makes a hard sum look easy by chatting while solving it. The showing off is the message: I have so much spare that I can waste it.

The short version

Zahavi's claim: signals stay honest when they are costly. A huge tail handicaps survival, so only a genuinely fit male can afford one. The waste is the message.

Where you have seen it

Gazelles leaping vertically in front of predators, advertising fitness instead of fleeing. Luxury spending, from designer weddings to sports cars, running on the same logic of visible cost.

The uncomfortable bit

The maths went back and forth for decades, and modern models suggest costly signalling works under narrower conditions than Zahavi claimed. The tail, meanwhile, still slows the peacock down, and peahens still take notes.

Source: Zahavi, Journal of Theoretical Biology, 1975; Grafen, JTB, 1990 (the formal model); Penn and Szamado, Biological Reviews, 2020 (the critique)

LIFE · Contested · Lovelock and Margulis, 1972

Gaia Hypothesis

Life does not just live on Earth. It regulates Earth, like a body regulates temperature.

In plain words

Your body keeps its temperature steady. You sweat when hot and shiver when cold. James Lovelock suggested the whole Earth does something similar, and that living things are the reason. Plants, plankton and microbes have kept the air, the seas and the temperature in a range life can bear for billions of years, without anyone planning it. That is the Gaia hypothesis, named after the Greek earth goddess. The mystical version, Earth as a living being with a purpose, most scientists reject. The practical version, that life shapes its planet, is now simply how climate science works.

An everyday example

Keep a fish tank. Water plants make oxygen, the fish use it and give off waste, and bacteria turn the waste into food for the plants. Left alone, the tank settles into a balance that no single fish or plant is in charge of. The Gaia idea says Earth is the biggest tank, and nobody owns it.

The short version

Lovelock proposed that organisms and environment form one self-stabilising system, keeping oxygen, temperature and ocean chemistry within livable bands for billions of years. Life terraformed its own planet and maintains the settings.

Where you have seen it

Oxygen itself is biological output. Plankton influence clouds. The version stripped of mysticism, Earth system science, is now simply how climate is studied.

In films, books and games

In Isaac Asimov's novel Foundation's Edge (1982), the planet Gaia is a whole world, rocks and all, that shares one mind.

The uncomfortable bit

Evolutionary biologists' objection stands: the planet cannot be selected for, since there is no population of competing Earths. And life once nearly froze the planet by exhaling oxygen. Gaia, if she exists, has tried to kill herself before.

Source: Lovelock and Margulis, Tellus, 1974; Kirchner, Climatic Change, 2002 (the evolutionary objection)

LIFE · Proven · Lynn Margulis, 1967

Endosymbiosis

Every cell in your body contains the descendants of a captured bacterium.

In plain words

Long ago, one tiny cell swallowed a smaller bacterium. Instead of digesting it, the big cell let it live inside and share its energy. The two stayed together, and their descendants are in every one of your cells today. Those descendants are mitochondria, the parts that make your energy. They still have their own DNA, separate from yours, and still copy themselves on their own timetable. Plants did it twice, with a second bacterium that became the green part that catches sunlight. So you are, in a way, two very old living things stuck together.

An everyday example

Think of a small shop that sets up a stall inside a big shop. At first they are separate businesses. Over the years the big shop comes to depend on the stall for its best sales, and the stall cannot survive outside. Now they are one shop with two sets of books. Your cells are exactly that, with two sets of DNA.

The short version

Mitochondria were once free-living bacteria, swallowed by a host cell around two billion years ago and never digested. The arrangement became permanent. They still carry their own separate DNA and still divide on their own schedule inside you.

How it actually works

The evidence is written into the organelle itself. Each mitochondrion carries its own genome, in humans a circular DNA molecule of 16,569 base pairs encoding 37 genes: 13 proteins of the respiratory chain, 22 transfer RNAs and two ribosomal RNAs. It makes those proteins on ribosomes of the bacterial type, blocked by chloramphenicol, which leaves the cell's other ribosomes untouched. Mitochondria have two membranes and arise only by the splitting of existing mitochondria. Comparisons of ribosomal RNA sequences, beginning in Carl Woese's laboratory in 1985, place mitochondrial genes within the Alphaproteobacteria, the group that includes the rickettsias. Chloroplasts tell the same story: their genomes, ribosomes and photosynthetic machinery match cyanobacteria.

The genome of Rickettsia prowazekii, the typhus bacterium, sequenced by Siv Andersson and colleagues in Nature in 1998, was at 1.1 million base pairs and 834 protein-coding genes the closest bacterial relative of mitochondria then known. Molecular clocks and the fossil record place the origin of mitochondria between about 1.5 and 2 billion years ago; the oldest fossil confidently identified as a complex alga, Bangiomorpha pubescens from Arctic Canada, is dated to about 1.05 billion years, so the chloroplast merger was earlier. In 2010 Nick Lane and William Martin argued in Nature that the merger was the enabling step for complex life: a cell with thousands of internalised, stripped-down power units has orders of magnitude more energy per gene than any bacterium.

Konstantin Mereschkowsky, a Russian botanist, proposed in 1905 that chloroplasts descended from cyanobacteria living inside a colourless host. Ivan Wallin at the University of Colorado argued in his 1927 book Symbionticism and the Origin of Species that mitochondria were bacteria; the idea was dismissed. In 1963 Margit and Sylvan Nass showed by electron microscopy that mitochondria contain DNA. Lynn Margulis, then publishing as Lynn Sagan, assembled the modern theory in 'On the Origin of Mitosing Cells' in the Journal of Theoretical Biology in 1967, after about fifteen rejections. Robert Schwartz and Margaret Dayhoff supplied the first sequence-based support in Science in 1978, and the ribosomal RNA phylogenies of the 1980s settled the question.

Where you have seen it

Plants ran the acquisition twice: chloroplasts are a separately domesticated photosynthetic bacterium. Margulis's paper was rejected by many journals before becoming settled textbook fact.

In films, books and games

Parasite Eve (novel 1995, PlayStation game 1998) runs the story backwards: the mitochondria remember they were once free-living and rise against their hosts. Star Wars: The Phantom Menace (1999) gives the Jedi midi-chlorians, microscopic life forms inside every cell that Qui-Gon calls symbionts.

The uncomfortable bit

That single merger may be the reason complex life exists at all. It happened, as far as the evidence shows, exactly once. Some biologists suspect this step, not the origin of life, is the great filter that empties the galaxy.

Source: Margulis, writing as Sagan, 'On the Origin of Mitosing Cells', Journal of Theoretical Biology, 1967; Lane and Martin, Nature, 2010

LIFE · Proven · Ernst Mayr, 1942

Founder Effect

A whole population's genetics shaped by the random luggage of a few founders.

In plain words

When a few people leave a big group and start a new one somewhere else, they take only their own genes with them. If one of them happens to carry a rare trait, that trait becomes common in the new group, just by chance. If none of them carries something that was common back home, the new group loses it. The founder effect is this luck of the draw. It is why some islands and small communities have unusual amounts of certain conditions, and why cheetahs, which once shrank to a tiny number, are all so alike.

An everyday example

Take a big jar of mixed sweets and grab five without looking. Start a new jar with those five. If four happen to be orange, the new jar is mostly orange, though the original was mixed. Any flavour you did not pick is gone from the new jar for good. Small groups of people starting new villages work the same way.

The short version

When a small group starts a new population, whatever gene variants they happen to carry, common or rare, become the baseline for all descendants. Chance, not fitness, sets the frequencies.

Where you have seen it

Rare disorders concentrated in island and isolated communities worldwide. An entire island community once famous for hereditary colourblindness after a typhoon reduced the population to a few dozen survivors.

The uncomfortable bit

Humanity itself is one long founder story. Genetic diversity thins with distance from Africa, tracing each migration's shrinking sample of the original variety. Cheetahs show the extreme endpoint: a past bottleneck so tight that unrelated cheetahs can accept each other's skin grafts.

Source: Mayr, Systematics and the Origin of Species, 1942; Brody et al. on Pingelap achromatopsia, Lancet, 1970; O'Brien et al., Science, 1985 (cheetah skin grafts)

LIFE · Mostly myth · Folk science, 1800s

Boiling Frog

Heat the water slowly and the frog never jumps out. Except it does. Every time.

In plain words

There is a famous story. Drop a frog in hot water and it jumps out. Put it in cool water and heat it slowly, and it never notices, so it boils. The story is used to warn that slow changes sneak up on you. But it is wrong. A real frog gets restless as the water warms and climbs out long before it is in danger. The story came from old experiments on frogs whose brains had been damaged. Scientists have said so for over a hundred years. People kept repeating it anyway. The warning about slow change is still fair, for humans.

An everyday example

Sit in a bath and slowly add hot water. You notice each change and stop when it gets too warm. Frogs notice too. Now think of a phone screen that gets a little more cracked every week. You stop seeing the crack. That is the human version, and it is real, even though the frog version is not.

The short version

The claim comes from 1870s experiments on frogs with parts of the brain removed. An intact frog grows increasingly agitated as the water warms and exits well before danger. Biologists have been correcting this for over a century, and the metaphor outsells the correction.

Where you have seen it

Business books, climate speeches and self-help talks, all warning against gradual change using an experiment that shows the opposite.

The uncomfortable bit

The irony is complete: the myth about failing to notice gradual change spread gradually, unquestioned, through people who never checked. Humans, unlike frogs, genuinely do normalise slow decline. The metaphor is right about us and wrong about the frog.

Source: Goltz, 1869 and Sedgwick, 1888, the brain-lesioned frog experiments; Gibbons, 'The Legend of the Boiling Frog', Ecoviews, 2007

LIFE · Open question · G.E. Hutchinson, 1961

Paradox of the Plankton

One drop of seawater, a handful of resources, and somehow hundreds of coexisting species.

In plain words

Plankton are tiny living things floating in the sea. They all need the same few things: light, and a handful of nutrients dissolved in the water. Textbook rules say that when many species compete for the same few things, the best one should win and the rest should die out. But a single bucket of seawater can hold hundreds of plankton species living side by side. That should be impossible, and nobody fully agrees on why it happens. The best guess is that the sea never sits still long enough for any one species to win.

An everyday example

Picture a cricket league where every team plays for the same cup. If the pitch, the weather and the rules never changed, the same strong team would win every year and the rest would give up. Now change the pitch, the weather and the rules every week. Different teams win at different times and none ever drops out. The sea is that changing pitch.

The short version

Competition theory says species competing for the same few nutrients should collapse to a handful of winners. Plankton ignore this, maintaining spectacular diversity in a seemingly uniform ocean. Hutchinson named the contradiction and ecology has been chewing on it since.

Where you have seen it

The same puzzle scales up: rainforest plots holding hundreds of tree species all competing for the same light, water and soil.

The uncomfortable bit

The leading resolution is that equilibrium never arrives. Turbulence, seasons and predators keep shuffling the deck faster than any winner can consolidate. Diversity may persist not despite instability, but because of it.

Source: Hutchinson, 'The Paradox of the Plankton', The American Naturalist, 1961

LIFE · Open question · Documented from 2008

Kodinhi Twins

One Kerala village has several times the global twin rate. Nobody has explained it.

In plain words

Kodinhi is a small village in Kerala. In most places twins are fairly rare, and a whole school might have one or two pairs. In Kodinhi you keep meeting children with the same face, pair after pair, in nearly every street. Scientists have tested the villagers' blood, the water they drink and the food they eat. Nothing they found explains it. That is what makes this one special. It is not a trick of numbers or a story that grew in the telling. It is a real place, with a real puzzle, and the answer is still missing.

An everyday example

Count the twins in your own school. In a big school of a thousand children you might find a handful of pairs. Now picture a school where nearly every class has two or three sets, and the teachers keep mixing them up. That is roughly what a Kodinhi school looks like, and nobody can tell you the reason.

The short version

Kodinhi in Malappuram district records twin births at many multiples of the Indian average, hundreds of pairs in a village of a few thousand families. Researchers have sampled DNA, water and diet. No confirmed cause has been published.

Where you have seen it

The village has a twins welfare association and a signboard announcing itself as God's own twin village. Similar unexplained clusters exist in Nigeria and Brazil.

The uncomfortable bit

The honest scientific status is a shrug, which makes it precious. Genuine open mysteries you can visit by bus are rare, and every proposed explanation so far, genetics, water chemistry, diet, has failed to close the case.

Source: No peer-reviewed study of the cluster has been published. Rate as reported by the Kodinhi Twins and Kins Association; BBC News coverage, 2011

TIME · Proven · Wolfgang Rindler, 1961

Ladder Paradox

A ladder too long for a barn fits inside with both doors shut, and also does not.

In plain words

One of the odd rules of very fast travel is that moving things look shorter. A ladder flying past you at nearly the speed of light looks squashed. So a long ladder could seem to fit inside a short barn, with both doors shut for a moment. But from the ladder's point of view, the barn is the thing rushing past, so the barn looks squashed and the ladder sticks out at both ends. Both views are right. The trick is that the two doors do not shut at the same moment for the ladder. Same moment is not the same for everyone.

An everyday example

Two friends stand at opposite ends of a cricket pitch and clap when a ball passes them. To a third friend in the middle the claps sound together. To someone running alongside the ball, one clap comes first. Nobody is lying. Who hears what depends on how they are moving, and the ladder puzzle is that idea pushed to its limit.

The short version

Run a ladder at a barn at near light speed. From the barn, the ladder contracts and fits, so both doors can close at once. From the ladder, the barn contracts and the ladder never fits. Both accounts are correct. The doors do not close at the same moment in the ladder's frame.

Where you have seen it

Length contraction is measured every day in particle accelerators, where a proton's electric field flattens into a thin disc. Muons from the upper atmosphere reach the ground for the same reason.

The uncomfortable bit

Nothing in the story is an illusion. At once is not a fact about the world, it is a fact about who is asking. In the ladder's own bookkeeping the far door shuts and reopens before the near door ever closes, so the ladder was never trapped.

Source: Rindler, 'Length Contraction Paradox', American Journal of Physics 29, 1961; Einstein, Annalen der Physik, 1905

TIME · Thought experiment · Rietdijk and Putnam, 1966

Andromeda Paradox

Two people pass on a pavement and disagree about whether a distant invasion has been launched.

In plain words

We like to think there is one single now that the whole universe shares. Einstein's rules say there is not. Your now depends a little on how you are moving. On Earth the difference is far too small to notice. But stretch it across the huge distance to the Andromeda galaxy and it grows. Two people walking past each other in the street, one heading east and one west, have nows that differ by days out there. For one, an alien fleet has already launched. For the other, the aliens have not decided yet. Neither person can check, because light takes millions of years to arrive.

An everyday example

Two people pass each other on a busy Delhi road. One walks towards the setting sun, one away from it. Ask each of them what is happening right now in a galaxy far away. The maths gives them different answers, days apart, because of nothing more than the direction they are walking. With a phone app you can find the Andromeda galaxy in the night sky, a faint smudge, and think about that.

The short version

Relativity has no universal now. Each observer's present slices the universe at a slightly different angle, and the tilt grows with distance. Two walkers passing each other on Earth have presents that differ, at the Andromeda galaxy, by days. In one, the fleet has set off. In the other, the council is still voting.

Where you have seen it

Nowhere you can point to. Andromeda is two and a half million light years away, so nothing either walker learns can reach them for that long, and the paradox never bites in practice.

The uncomfortable bit

Rietdijk and Putnam read it as proof that the future is already fixed, since someone's present always contains your tomorrow. Penrose used it to argue that the block universe is unavoidable. Others reply that a present nobody can inspect is not a present at all.

Source: Rietdijk, Philosophy of Science 33, 1966; Putnam, 'Time and Physical Geometry', Journal of Philosophy 64, 1967; Penrose, The Emperor's New Mind, 1989

TIME · Open question · Josef Loschmidt, 1876

Loschmidt's Paradox

Every law of motion runs equally well backwards, yet the cup never unbreaks.

In plain words

Drop a cup and it smashes. Nobody has ever seen the bits jump back together into a cup. Yet the rules that govern how each tiny piece moves work just as well backwards as forwards. Reverse the motion of every atom and, by the same rules, the pieces would gather back into a whole cup. So the rules allow un-smashing. The puzzle is why we never see it. The usual answer is that there are far more messy arrangements than tidy ones, so mess is what you get by chance. That does not fully explain why the universe began so tidy in the first place.

An everyday example

Shuffle a new pack of cards. It starts sorted and ends jumbled. Keep shuffling and it stays jumbled. Every single shuffle is allowed to bring the pack back to sorted order, and nothing in the rules stops it. It just almost never happens, because there is one sorted order and billions of jumbled ones. Try it at home; you will be shuffling for a long time.

The short version

Boltzmann claimed to derive from Newton's laws that gas always drifts towards disorder. Loschmidt objected. Take any such process, reverse every molecule's velocity, and the same laws carry it back to order. For every entropy-raising history there is a mirror one that lowers it. A one-way arrow cannot come out of two-way rules.

Where you have seen it

Film a glass shattering and play it in reverse. Every frame obeys physics. Nobody has ever seen the reverse version, and no equation forbids it.

In films, books and games

Christopher Nolan's film Tenet (2020) is built on objects whose entropy runs backwards, so a bullet leaps back into the gun. Martin Amis's novel Time's Arrow (1991) tells a whole life in reverse, with meals coming out of mouths.

The uncomfortable bit

Boltzmann's reply was that disorder is simply far more common than order, so the odds are overwhelming, not the laws. That moves the question rather than answering it. The arrow now rests on the universe having started in an extraordinarily tidy state, and nobody knows why it did.

Source: Loschmidt, Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften, Wien, 1876; Boltzmann, 'Weitere Studien über das Wärmegleichgewicht unter Gasmolekülen', 1872 (the H-theorem)

TIME · Contested · Brandon Carter, 1983

Doomsday Argument

You are probably not among the first tiny fraction of all humans who will ever live.

In plain words

Think of everyone who will ever be born as a long line of people, from the very first to the very last. You are somewhere in that line. If you picked a spot at random, you would most likely land somewhere in the middle, not right at the start. Now count the people born before you. If you are near the middle, roughly that many more will come after you, and then humans end. The argument does not look at bombs or the weather. It works from one fact alone: that you were born when you were. Many people are sure it is wrong, but struggle to say where.

An everyday example

You join a queue at a railway ticket counter without knowing how long it is. Look at how many people are ahead of you. If there are ten, it is a fair bet the queue is not ten thousand long, because you would be very lucky to be that near the front. The doomsday sum applies that guess to the whole of human history.

The short version

Treat your birth as a random draw from every human who will ever exist. A random draw rarely lands in the first one percent. About 100 billion people have lived so far. If they are not a tiny fraction of the total, the total is not enormous, and humanity's run is shorter than we assume.

Where you have seen it

Gott visited the Berlin Wall in 1969 and used the same reasoning to predict it would stand for at least two and two thirds more years but fewer than 24. It fell twenty years later.

The uncomfortable bit

The argument uses no data about weapons, climate or asteroids, only the fact that you exist now. Critics reply that the same reasoning would have told a Roman that Rome was nearly finished, and that being unable to find the flaw is not the same as there being none.

Source: Carter, Philosophical Transactions of the Royal Society A 310, 1983; Gott, 'Implications of the Copernican principle for our future prospects', Nature 363, 1993; Leslie, The End of the World, 1996

CHANCE · Proven · Joseph Berkson, 1946

Berkson's Paradox

Two unrelated diseases look linked, just because both of them get you admitted.

In plain words

Sometimes two things look connected only because of how you chose who to look at. Picture a hospital. People get in if they have a broken leg, or if they have flu. Now look only at people inside the hospital. If someone has no broken leg, they must be there for the flu. So among patients, no broken leg seems to go with flu, as if the two were linked. In the town outside, they are not linked at all. The link was made by the front door. Any time a group is filtered, fake links like this can appear.

An everyday example

A cricket coach picks a team of eleven. To get in you need to bat well or bowl well. Look at the team. The weak batters must all be good bowlers, or they would not be there. It looks like batting and bowling skills trade off. Among all the children in the school they do not. The selection made the pattern.

The short version

Berkson worked the arithmetic for a hospital. Take two illnesses with no connection. Either one is enough to get a person admitted. Among patients, then, anyone without the first illness is more likely to have the second, because something put them in a bed. The records show a link that the population does not have.

Where you have seen it

The belief that good-looking people are rude. You only date people who clear some overall bar, so among them, the more attractive ones needed less charm to get in.

The uncomfortable bit

The paper was widely disregarded for over thirty years because Berkson had only arithmetic and no real data. Every study built on volunteers, patients, customers or survivors is exposed to it, and the correlation it produces is not weak or noisy. It is clean, strong and entirely fake.

Source: Berkson, 'Limitation of the Application of Fourfold Table Analysis to Hospital Data', Biometrics Bulletin 2, 1946; Ellenberg, How Not to Be Wrong, 2014 (the dating version)

CHANCE · Open question · Maurice Kraitchik, 1953

Two Envelopes Problem

Whichever envelope you hold, the maths says the other one is worth more.

In plain words

You are handed two envelopes. One holds twice as much money as the other, but you do not know which. You pick one. Suppose it holds 100 rupees. The other must hold 200 or 50. Swapping could win you 100 more or cost you only 50, so swapping seems clever. Now put the envelope back and pick the other. The same thinking says swap again. And again. You could stand there swapping for ever. The sum seems to say each envelope is better than the other one, which cannot be true, and experts still argue about where the slip is.

An everyday example

Try it with a friend and two folded notes. Your friend writes an amount on one and double it on the other. You pick one, look, and decide whether to swap. Play many rounds and count your money. Always swapping wins nothing extra in the long run. Feel the pull of the argument, then watch the results refuse to back it up.

The short version

Two sealed envelopes, one holding twice the money of the other. You pick one and call its contents A. The other is either 2A or half of A, apparently with equal odds, so its average is one and a quarter A. Switch. Now repeat the reasoning from the new envelope, and switch back. Forever.

Where you have seen it

Kraitchik told it with two men comparing neckties, the dearer tie going to the other man. Each reasons that he risks his own tie but might win a better one, so both think the bet favours them.

The uncomfortable bit

Every step has been attacked and every attack has been answered. The cleanest objection is that no way of choosing the amounts makes every possible A equally likely to be the smaller one. Yet versions exist where the expected gain from switching is genuinely infinite, and mathematicians still argue about what that means.

Source: Kraitchik, Mathematical Recreations, 2nd edition, 1953 (the necktie form); Nalebuff, 'The Other Person's Envelope is Always Greener', Journal of Economic Perspectives 3, 1989

CHANCE · Proven · Classic puzzle, origin unrecorded

Potato Paradox

Potatoes go from 99 percent water to 98 percent and lose half their weight.

In plain words

A big sack of potatoes is almost all water, 99 parts water to 1 part potato. Leave it out and some water dries away, so now it is 98 parts water to 2 parts potato. That sounds like almost nothing changed. But look at the dry part. It has not changed at all, and it now makes up 2 out of every 100 instead of 1 out of every 100. For the dry part to double its share, the whole sack must have halved. Half the weight has gone. People's brains hear a one percent change and expect a tiny loss.

An everyday example

A jug of squash is 99 parts water and 1 part syrup. Now you want it 98 parts water and 2 parts syrup. You cannot add syrup, only pour water away. To make the syrup twice as strong you must pour out half the jug. Try it at the kitchen sink with a measuring jug and it stops feeling wrong.

The short version

One hundred kilograms of potatoes are 99 percent water, so they hold one kilogram of solid matter. Leave them out overnight until they are 98 percent water. The solid kilogram has not changed. It is now two percent of the total, so the total is fifty kilograms. Half the pile has evaporated.

Where you have seen it

Any figure quoted as a share of a total, where the total itself moves. Fat-free claims, water content of vegetables, a company's costs as a percentage of shrinking revenue.

The uncomfortable bit

Nothing about the puzzle is tricky except our habit of hearing one percent as a small change. Track the part that stays fixed, the solids, and the answer is obvious. Track the part that moves, and you will be sure the answer is about 99 kilograms.

Source: No known first printing; the arithmetic checks by hand, since one kilogram of solids at two percent of the whole is fifty kilograms

CHANCE · Proven · Alvan Feinstein, 1985

Will Rogers Phenomenon

Move a patient from one group to another and both groups' survival improves.

In plain words

Split a class into two groups by their marks: a top group and a bottom group. Take the weakest child from the top group and move them to the bottom group. The top group's average goes up, because its lowest mark has gone. The bottom group's average also goes up, because the new arrival scores higher than everyone already there. Both groups look better. Nobody learned anything new. This happens in hospitals when better scanners move some patients from an early stage of illness to a later stage. Every stage seems to do better, but no one is actually healthier.

An everyday example

Two cricket teams in a colony league. Team A's worst batter averages 20, which is still better than anyone in Team B. Move him to Team B. Team A's average rises, Team B's average rises, and the league table looks like everyone improved. Try it with any two lists of numbers on paper and you will see it every time.

The short version

Better scanners find small secondary tumours that older methods missed. Those patients move from the early-stage group to the late-stage group. They were the sickest of the early group, so its average survival rises. They are the healthiest of the late group, so its average rises too. Nobody has lived a day longer.

Where you have seen it

Feinstein compared lung cancer patients from the 1950s with patients from 1977. Survival within every stage had improved. Overall survival had not. Rogers had joked that Okies moving to California raised the average intelligence of both states.

The uncomfortable bit

Any hospital that buys a better scanner will report better stage-specific results the next year without treating anyone differently. Comparing today's stage 2 with the stage 2 of 1990 is comparing two different groups that happen to share a label.

Source: Feinstein, Sosin and Wells, 'The Will Rogers Phenomenon', New England Journal of Medicine 312, 1985

CHANCE · Proven · Maurice Allais, 1953

Allais Paradox

People choose the sure thing, then reverse themselves when the same odds are dressed differently.

In plain words

People say they choose money by weighing the odds. Then they break their own rule. Offered a sure prize or a gamble with a small chance of nothing, most take the sure prize, even when the gamble pays more on average. Fine. Now take the same two choices and remove the same big chunk of luck from both. The sure prize is no longer sure, and most people switch to the bigger gamble. By the rules of careful choice the two answers should match, because only a shared part was taken away. They do not match. Certainty feels special, and people pay to have it.

An everyday example

A shopkeeper offers you 500 rupees now, or a lucky dip where nearly every ticket pays 600 but one in a hundred pays nothing. Most of us take the 500. Now he offers two dips instead: one where 11 tickets in 100 pay 500, or one where 10 in 100 pay 600. Suddenly most people go for the 600. Same gap in risk, opposite choice.

The short version

A sure million, or 89 percent a million, 10 percent five million, 1 percent nothing. Most take the sure million. Then, 11 percent a million or 10 percent five million. Most take the five. Expected utility theory says the two answers must match, since the pairs differ only by a shared 89 percent.

Where you have seen it

Allais put the choices to the leading American theorists at a Paris conference in 1952. Savage, who had written the axioms, answered them the way ordinary people do, and then agreed that his own rules showed he had erred.

The uncomfortable bit

It has been replicated for seventy years and it does not fade with education or with real money. Certainty is not one more probability. It is treated as a different kind of thing, and Kahneman and Tversky built prospect theory on that gap.

Source: Allais, 'Le comportement de l'homme rationnel devant le risque', Econometrica 21, 1953; Kahneman and Tversky, 'Prospect Theory', Econometrica 47, 1979

LOGIC · Proven · Evangelista Torricelli, 1641

Gabriel's Horn

A horn you can fill with paint but never paint.

In plain words

Picture a trumpet that gets thinner and thinner and never ends. It is a mathematical shape, so it can go on for ever. Now the strange part. If you poured paint inside, it would fill up. The amount of paint needed is a fixed, finite number. But if you tried to paint its inside wall, you could never finish, because the wall goes on for ever and its area adds up without end. A pot of paint that fills it cannot coat it. The trick lives in the difference between a number and real stuff: mathematical paint has no thickness, real paint does.

An everyday example

Roll a sheet of paper into a cone and keep imagining it stretching out thinner and thinner behind your back, out of the room, out of the city, never stopping. You could fill the wide end with a cup of water, but if a spider tried to walk the inside wall to the tip, it would walk for ever. The volume ends; the surface does not.

The short version

Take the curve y equals 1 over x from x equals 1 outwards and spin it round the axis. The trumpet has a finite volume, exactly pi cubic units. Its inner surface is infinite. A finite tin of paint fills it to the brim, yet no amount of paint can coat the inside wall.

Where you have seen it

Torricelli found it a generation before calculus existed, using Cavalieri's method of slicing. The shape shocked his contemporaries, and several philosophers refused to believe it.

The uncomfortable bit

The paint story resolves once you notice that paint has thickness. Far down the horn the bore is thinner than any molecule, so real paint stops there. Mathematical paint, with no thickness, coats the wall with the finite pint that filled it. The paradox lives only in the gap between a number and a substance.

Source: Torricelli, 'De solido hyperbolico acuto', Opera Geometrica, 1644 (the result dates from 1641)

LOGIC · Open question · Carl Hempel, 1945

Raven Paradox

A green apple counts as evidence that all ravens are black.

In plain words

Suppose you want to check the claim that all ravens are black. Seeing a black raven helps. Now here is the odd part. The claim means exactly the same as this: anything that is not black is not a raven. A green apple, which is not black and is not a raven, fits that second claim perfectly. So by strict logic the apple supports the idea that ravens are black. That feels absurd. You cannot learn about birds by looking at fruit. The usual answer is that the apple does help, but by such a tiny amount that it hardly counts, because there are so many non-black things.

An everyday example

You want to know if every child in your school wears black shoes. You could check every child. Or you could check every pair of white shoes in the world and confirm none belong to your school. Both would prove the same thing. One takes an afternoon, the other takes for ever, and that difference is the whole puzzle.

The short version

The statement all ravens are black is logically identical to all non-black things are non-ravens. Anything that supports one supports the other. A green apple is a non-black non-raven, so by the plain rules of logic it confirms the claim about ravens.

Where you have seen it

Indoor ornithology. In principle you could gather evidence about ravens from an armchair by listing the colours of your furniture, and nobody who has ever run a survey believes that.

The uncomfortable bit

The standard escape is that the apple does confirm the claim, by an amount so small it rounds to nothing, because there are vastly more non-black things than ravens. That answer is still argued over, and it means the logic was right and your instinct was right, and they still disagree.

Source: Hempel, 'Studies in the Logic of Confirmation', Mind, 1945

LOGIC · Proven · Bertrand Russell, 1908

Berry Paradox

The smallest positive integer not definable in under sixty letters has just been defined in fifty-seven.

In plain words

Words can describe numbers. Ten, one hundred, the number of days in a year. Short phrases can only describe so many numbers, because there are only so many short phrases. So there must be numbers that no short phrase can describe. Pick the smallest one. Now look at what you just did. You described it, with a short phrase: the smallest number no short phrase can describe. So it can be described after all, which means it is not that number. Every candidate fails the same way. The mistake is that describe is a slippery word, and logic needs words that stay still.

An everyday example

Play it with a friend. Agree that a description may use at most ten words. Write down numbers and describe each one. Sooner or later you reach one where ten words are not enough. Now say: the first number we could not describe in ten words. That is nine words. You just described it. Watch your friend's face.

The short version

Only finitely many phrases have fewer than sixty letters, so some positive integers cannot be defined by any of them, and one of those must be smallest. But that very phrase picks the number out, using fifty-seven letters. The number is defined by a phrase that says it cannot be.

Where you have seen it

Russell credited it to G. G. Berry, a junior librarian at Oxford's Bodleian. Its modern descendant is Chaitin's proof that no program can reliably measure how compressible a string of data is.

The uncomfortable bit

The escape is that definable is not a well defined word. Once you pin it to a formal language the phrase stops being a definition in that language, and the paradox vanishes. Everyday English never gets pinned down, so in everyday English the paradox is still sitting there.

Source: Russell, 'Mathematical Logic as Based on the Theory of Types', American Journal of Mathematics, 1908 (crediting G. G. Berry)

COSMOS · Open question · Sagan and Mullen, 1972

Faint Young Sun Paradox

The young Sun was too dim to keep Earth's oceans liquid, and the oceans were liquid anyway.

In plain words

A star like the Sun slowly turns up its own brightness over billions of years. So when Earth was young, the Sun was a lot dimmer than it is today. With that little heat, the oceans should have frozen into a ball of ice. But old rocks show the opposite. There were rivers, seas and even early life while the Sun was weak. Something kept the planet warm. The best guess is a thick blanket of gases that trapped heat, far thicker than today's. Yet no single mix of gases fits everything the rocks say, so the case is still open.

An everyday example

Leave a bowl of water outside on a cold winter night in Shimla with only a weak lamp shining on it, and it freezes. Cover the bowl with a thick blanket and it might stay liquid. Early Earth had the weak lamp and stayed liquid, so scientists are hunting for the blanket. On a cloudy night you can feel the same thing: clouds hold the day's heat in.

The short version

Stars brighten slowly as they age. Standard models put the Sun at about 70 percent of its present output four billion years ago, and with today's atmosphere that would freeze the planet solid. Yet rocks from that era show flowing water, sediments and, soon after, life.

Where you have seen it

The leading fix is a much thicker greenhouse blanket of carbon dioxide and methane, which is why early Earth is studied as a test case for how much climate an atmosphere can hold.

The uncomfortable bit

Sagan's own answer was ammonia, which turned out to break down too fast in sunlight. Fifty years on, no single greenhouse recipe fits all the rock evidence, and Mars has the same problem with river valleys and an even fainter Sun.

Source: Sagan and Mullen, 'Earth and Mars: Evolution of Atmospheres and Surface Temperatures', Science 177, 1972

COSMOS · Resolved · Lord Rayleigh, 1900

Ultraviolet Catastrophe

By the physics of 1900, a warm oven should pour out infinite energy at the violet end.

In plain words

Heat something up and it glows. Physicists in 1900 tried to work out exactly how much light a hot oven should give off at every colour. Their formula worked for red light, but as they moved towards blue and violet the answer kept growing and never stopped. It said an ordinary oven should blast out an endless amount of energy. Real ovens do not. Max Planck fixed it by saying light comes in small packets, and a violet packet costs more energy than a red one, so the oven rarely affords them. That small fix became quantum physics, the science of the very small.

An everyday example

Watch a heating coil in the kitchen. At first it glows dull red. Turn it up and it goes orange, then yellow. It never goes blue, let alone violet, however hot it gets. The old formula said it should. Your stove disproves nineteenth century physics every time you make chai.

The short version

The Rayleigh-Jeans law shares heat evenly across every possible wavelength of light inside a hot cavity. There are ever more short wavelengths to fill, so the total climbs without limit. Measured ovens do nothing of the kind: their glow peaks and then falls away towards the ultraviolet.

Where you have seen it

Planck's fix let energy come only in whole packets, and short wavelengths need packets that heat rarely supplies. That move is the birth of quantum theory, and his curve still fits every furnace, star and light bulb.

The uncomfortable bit

Planck did not set out to overturn physics. He called the packet an act of desperation, a mathematical trick to make the formula fit, and expected someone to remove it later. Nobody could. The name arrived in 1911, after the catastrophe had already been averted.

Source: Rayleigh, Philosophical Magazine, 1900; Planck, Verhandlungen der Deutschen Physikalischen Gesellschaft, 1900; named by Ehrenfest, Annalen der Physik, 1911

MIND · Proven · Edward Thorndike, 1920

Halo Effect

A single impression quietly fills in every judgement you think you made separately.

In plain words

When we like one thing about a person, that liking spills over. If someone looks smart and neat, we assume they are also clever, kind and honest, without any proof. One good impression paints a glow over everything else. The same works in reverse: one bad first impression makes us judge everything harshly. The odd part is that we do not notice it happening. We are sure we judged each thing on its own. Tests show we did not. This is why looks, height, a good voice or a famous brand name quietly change how we rate things that have nothing to do with them.

An everyday example

A new teacher walks in wearing a smart shirt and smiles. By lunchtime the class has decided she is kind, fair and knows her subject, before she has taught a lesson. Try it yourself: pour the same cold drink into a branded bottle and a plain one and ask a friend which tastes better. Most people pick the brand.

The short version

Thorndike asked army officers to rate their men on physique, intelligence, leadership and character as separate traits. The scores correlated far too well. A man rated tall and well built was also rated smarter and more loyal. One overall glow was leaking into every box on the form.

Where you have seen it

Attractive defendants get lighter sentences, taller candidates get more votes, and the brand on the box changes how the biscuit inside tastes. Interviews are the halo effect with a handshake.

The uncomfortable bit

In 1977 Nisbett and Wilson showed students the same lecturer being warm or cold. The warm version was rated better looking and less annoying in his accent. Students insisted his manner had not touched those ratings. They were sure, and they were wrong.

Source: Thorndike, 'A Constant Error in Psychological Ratings', Journal of Applied Psychology, 1920; Nisbett and Wilson, Journal of Personality and Social Psychology, 1977

MIND · Proven · Camerer, Loewenstein and Weber, 1989

Curse of Knowledge

Once you know something, you cannot picture what it was like not to know it.

In plain words

Once you understand something, it becomes hard to remember what it felt like not to understand it. The knowledge sits in your head and you cannot switch it off. So when you explain the thing to someone else, you skip steps, use words they do not know and assume they can see what you see. Teachers do it. Parents do it. Anyone giving directions does it. It is called a curse because knowing more makes you worse at guessing what a beginner needs. The fix is to remember that clear to you is not clear to them.

An everyday example

Ask your grandmother to teach you a recipe she has cooked for fifty years. She says a pinch, a little, until it looks right. She is not being unhelpful. She cannot see the steps any more, because they are automatic. Now try to explain to her how to send a photo on WhatsApp and notice how many things you skip.

The short version

Better informed people fail to set their knowledge aside when guessing what less informed people will do. Camerer, Loewenstein and Weber found this in traders who had been told a company's real earnings. They kept predicting that the uninformed market would price the shares almost as if it knew too.

Where you have seen it

Every textbook written by an expert, every set of instructions written by the person who built the thing, every teacher who cannot see why the class is lost.

In films, books and games

Chip and Dan Heath's Made to Stick (2007) makes it the villain of the book, with the tappers-and-listeners experiment in which people tapping out a tune expect half of listeners to name it and about one in forty does.

The uncomfortable bit

In 1990 Elizabeth Newton had people tap out well known songs on a table while listeners tried to name them. Tappers predicted listeners would get about half right. Listeners got 2.5 percent. The tapper hears the whole tune in their head, and to the listener it is knocking.

Source: Camerer, Loewenstein and Weber, 'The Curse of Knowledge in Economic Settings', Journal of Political Economy, 1989; Newton, Stanford University PhD thesis, 1990 (tappers and listeners)

MIND · Contested · Henry Landsberger, 1958

Hawthorne Effect

Workers produced more when the lights went up, and more again when the lights went down.

In plain words

In a factory near Chicago, bosses tried to find out whether brighter lights would make workers faster. They turned the lights up and work sped up. Then they turned the lights down and work sped up again. Whatever they changed, the numbers got better. The explanation people settled on was that the workers were performing because they knew they were being watched. Being studied was the real change, not the lighting. Later, when the old records were checked, the story turned out to be muddier than the textbooks say. Still, the idea stuck: people act differently when they know someone is counting.

An everyday example

A teacher announces that she will be watching how tidy each desk is this week. Every desk is spotless by Tuesday. Next week, when nobody mentions it, the mess creeps back. You can test it at home: tell a younger brother you are timing how fast he brushes his teeth and watch the sudden effort.

The short version

Between 1924 and 1932 the Western Electric plant at Hawthorne, near Chicago, ran experiments on lighting, rest breaks and pay. Output seemed to rise whatever was changed. The conclusion drawn was that being watched and singled out for study was itself the treatment, and that people perform for an audience.

Where you have seen it

Every workplace pilot that succeeded and every rollout that then failed. Handwashing rates in hospitals jump when the auditor is visible and fall when the counting goes electronic.

The uncomfortable bit

The original illumination data, long thought lost, was found and re-analysed by Levitt and List in 2011. The clean story was not in it. Output rose on Mondays anyway, after the weekend, and the lighting was changed on Sundays. The most famous effect in management science leans on a scheduling artefact.

Source: Roethlisberger and Dickson, Management and the Worker, Harvard University Press, 1939; Landsberger, Hawthorne Revisited, 1958; Levitt and List, American Economic Journal: Applied Economics, 2011

LIFE · Resolved · C. A. Thomas, 1971

C-value Paradox

An onion carries about five times as much DNA as a human.

In plain words

You might expect that more complicated living things need more DNA. Humans should have more than onions, and onions more than tiny worms. It does not work like that. An onion has far more DNA than you do. Some fish and some flowers have dozens of times more. The amount of DNA has almost nothing to do with how complex the creature is. The answer, found later, is that most DNA in these bulky genomes does not carry instructions for building the body. It is filler, much of it old copies of stretches that copied themselves. Why some species keep so much filler is still unclear.

An everyday example

Compare two schoolbags. One belongs to a top student and is light, with just the books needed. The other is stuffed with old notebooks, ten years of photocopies and torn pages, but the owner learns no more. The onion is the stuffed bag. Next time you slice one in the kitchen, remember it is carrying more DNA than you are.

The short version

The C-value is the amount of DNA in a single set of chromosomes. Thomas noted in 1971 that it bears almost no relation to how complex an organism is. The marbled lungfish carries about forty times the human amount, and a Japanese woodland flower, Paris japonica, about fifty.

Where you have seen it

Genome size ran ahead of gene number as soon as the two could be measured apart. Humans have roughly 20,000 protein-coding genes, about the same as a nematode worm, in a genome about thirty times larger.

The uncomfortable bit

Most of the bulk is non-coding DNA, much of it copies of parasitic elements that spread through genomes for their own sake. That settles the paradox and opens the enigma: nobody can yet say why a lungfish tolerates hundreds of times more of it than a pufferfish does.

Source: Thomas, 'The Genetic Organization of Chromosomes', Annual Review of Genetics, 1971; Gregory, Biological Reviews, 2001 (the C-value enigma)

LIFE · Resolved · Cyrus Levinthal, 1969

Levinthal's Paradox

A protein would need longer than the age of the universe to try every shape, and folds in milliseconds.

In plain words

A protein is a long chain that must fold into one exact shape to work, a bit like a strip of paper folded into a particular paper crane. The chain can bend in a huge number of ways. If it tried each way one by one, even very fast, the universe would end before it finished. Yet in your body proteins fold correctly in a tiny fraction of a second. The way out is that the chain does not try shapes at random. Each small fold makes the next one easier, so it slides towards the right shape like a marble rolling to the bottom of a bowl.

An everyday example

Give a friend a shoelace and ask them to tie a bow by trying every possible tangle in turn. They would never finish. Nobody ties a lace that way. Your hands make one loop, then the next, and each step narrows what can come next. Proteins fold like your fingers tie laces, not like a computer trying every option.

The short version

A protein is a chain of a few hundred amino acids, and each link can bend several ways. The number of possible shapes is astronomical. If the chain tried them at random, even at a trillion per second, the search would never finish. Real proteins settle into their working shape in microseconds to milliseconds.

Where you have seen it

Every enzyme in your body found its shape within moments of being made. When folding goes wrong it does not fail quietly: misfolded proteins are behind Alzheimer's, Parkinson's and mad cow disease.

The uncomfortable bit

Levinthal posed it to make a point, not to stump anyone. The chain never searches at random. Each partial fold is a little more stable than the last, so the landscape of shapes is a funnel and the protein rolls down it. The paradox only exists if you assume nature guesses.

Source: Levinthal, 'How to Fold Graciously', in Mössbauer Spectroscopy in Biological Systems, University of Illinois Press, 1969; Bryngelson, Onuchic, Socci and Wolynes, Proteins, 1995 (the folding funnel)

SOCIETY · Proven · C. Northcote Parkinson, 1957

Bike-shedding

A committee waves through a nuclear reactor in minutes and then argues at length about a bicycle shed.

In plain words

Give a committee a huge decision and a tiny one, and it will spend more time on the tiny one. The expensive item slips through, because arguing about it would mean admitting you do not follow it. The cheap item gets picked apart, because anyone can have a view on a shed. The name comes from a made-up story about a nuclear power plant and a bicycle shed. It is not a scientific law, it is a joke that turned out to be true. People talk about what they feel able to talk about, and that is usually the small stuff.

An everyday example

A school committee meets to plan the annual day. The budget for hiring the hall and sound system is agreed in two minutes. Then the colour of the ribbons on the invitation cards takes forty minutes and three votes. Watch any family planning a wedding and you will see the same thing with the menu.

The short version

Parkinson's law of triviality says the time a committee spends on an item is inversely proportional to the money involved. His fictional finance committee nods through an atomic reactor in two and a half minutes because nobody understands it, then spends three quarters of an hour on a bicycle shed because everybody does.

Where you have seen it

Code reviews that ignore the architecture and fight over variable names. Housing society meetings where the water pump budget passes silently and the colour of the gate takes the evening.

In films, books and games

It comes from C. Northcote Parkinson's book Parkinson's Law (1957), where a committee nods through a nuclear reactor in minutes and then spends the afternoon on a bicycle shed.

The uncomfortable bit

It is not a measured law. Parkinson wrote satire, and it has never been tested the way a law should be. It survives because everyone who has sat on a committee recognises it. The people talking about the shed are not stupid. They are contributing where they can, which is the whole problem.

Source: Parkinson, Parkinson's Law, and Other Studies in Administration, John Murray, 1957 (the chapter 'High Finance, or the Point of Vanishing Interest')