Are You Random?

weirdly accurate.

    Are You Random?

    How random are you?

    This randomness test lets you find out how random you really are: you make up twenty coin tosses in your head while a machine tries to call each one in advance. Psychologists have studied made-up random sequences for decades, including Falk and Konold in Psychological Review in 1997. Only after your twenty tosses do you see how many were called, which habit gave you away, and how a real coin compares.

    This test turns those habits into a game against a machine. You make 20 pretend coin tosses, and before each one a small predictor calls it from your earlier tosses alone: eight simple guessers, each betting on one human habit (switching, repeating, breaking a streak, following a recent pattern, evening out heads and tails), combined by a weighted majority vote that trusts whichever habit is working. You only see the calls afterwards. No predictor that sees only past tosses can beat chance on a real coin, so a real coin is missed about 10 times out of 20; your score is your misses as a share of that. Your first 12 tosses are also ranked against 3,449 online volunteers who each imagined 12 coin flips for Gauvrit and colleagues’ 2017 study, run through the same predictor. Then you watch twenty tosses of a real coin, and finally pick a number from 0 to 9 against a guess that was sealed before you chose.

    It is not a personality or intelligence test, and twenty tosses is a small sample: even a real coin gets called 13 or more times out of 20 about 13% of the time. The comparison group was people taking an online test, not people trying their hardest — 143 of them typed twelve of the same side — so read your rank as a reference point, not a verdict.

    What the test measures

    One thing, measured against a machine: how hard your made-up coin tosses are to predict.

    The task

    • 20 pretend coin tosses, one tap each on Heads or Tails, made up in your head
    • A predictor calls every toss in advance from the tosses before it: eight habit guessers combined by a weighted majority vote, so the same twenty tosses always get the same calls
    • Then twenty tosses of a real coin for comparison, drawn from your browser’s cryptographic random number generator
    • Last, one number from 0 to 9, picked after the machine’s guess is sealed

    Your result

    • How many of your 20 tosses were called in advance, and the one habit that showed most
    • “As random as a coin”: your misses as a share of a real coin’s expected 10, from 0 to 100%
    • Where your first 12 tosses rank among 3,449 online volunteers from Gauvrit et al. (2017): harder to call than what share of them
    • An analysis sheet: your switches against a real coin’s 50%, your longest run and how often a real coin runs that long, runs by length, and the habits that gave you away
    • A shareable card and a link that challenges a friend to be harder to call

    Before you start

    • About a minute
    • No headphones needed; sound effects are optional
    • Works on a phone or a computer (on a keyboard, the left and right arrows toss too)
    • Don’t use a real coin or a random number app: the point is what your own head produces

    Randomness and coin toss FAQ

    Is this test free? Do I need to sign up to see my result?

    Yes, it’s free, and there’s nothing to sign up for: no account, no email, no payment. Your result appears on this page the moment you finish, and you can share it or play again straight away.

    Can humans be truly random?

    That is the question this test puts to you, and the analysis sheet answers it after your twenty tosses, because knowing first would change how you toss. Researchers have measured the habits people show when they make up sequences, including how often they switch sides (Falk and Konold, 1997); the predictor here bets on eight such habits, and the sheet shows which of them, if any, it caught in yours.

    How does the randomness score work?

    The randomness score counts how often the predictor missed your 20 tosses, as a share of the 10 misses a real coin gets on average. Ten or more misses scores 100%, six misses scores 60%. The predictor only sees tosses you have already made, which is why it cannot do better than chance on a real coin.

    How many heads or tails in a row is normal in 20 coin tosses?

    The analysis sheet answers that once your twenty tosses are in, so toss by feel first. It sets your longest run beside the longest run a real coin most often makes in twenty tosses, gives the exact chance that a real coin runs at least as long as yours, and shows twenty real tosses straight after yours. These figures are exact counts over every possible sequence of twenty tosses.

    Why did I get a different result the second time?

    Twenty tosses is a small sample, so luck plays a real part. Even a real coin is called 13 or more times out of 20 by the predictor about 13% of the time. A few runs tell you more than one, and the habits listed on the analysis sheet are more stable than any single score.

    Is the real coin in the randomness test actually random?

    Yes. The real-coin comparison draws its twenty tosses from the browser’s cryptographic random number generator, not from a pattern. The real-coin figures on the analysis sheet are exact: every one of the 2²⁰ possible twenty-toss sequences was run through the same predictor.

    Does being random say anything about intelligence or personality?

    No. This is a one-minute game about a well-documented habit of human judgement, not a psychological assessment. The people it compares you with are 3,449 online volunteers from Gauvrit et al. (2017), many of whom answered casually, so a rank among them describes these twelve tosses, not you.

    Where the numbers come from

    Your rank compares the predictor’s misses on your first 12 tosses with the same predictor run on the 12 imagined flips of each of the 3,449 volunteers in Gauvrit et al.’s public data; only summary counts are kept. Real-coin figures are exact counts over every possible sequence, not simulations. The site does not publish its own players’ data.

    How it is scored

    Weighted majority vote

    Eight simple predictors each bet on one habit. Every time one of them is wrong, its say is cut to 55% of what it was, so the overall call leans on whichever habit you actually have.

    As random as a coin

    Your misses out of 20, divided by a real coin’s expected 10, capped at 100%. Ten misses scores 100%; six misses scores 60%.

    Rank among people

    Only your first 12 tosses go through the same predictor and are compared with the volunteers’ 12 imagined flips. The figure counts people who were called more often than you, plus half of those called exactly as often.

    Switch rate

    The share of tosses where the side changed from the one before. A fair coin’s expected switch rate is 50%.