Skip to content
One Quick Try

How Random Are You?

Press left or right a hundred times as randomly as you can, while a predictor tries to guess each press first.

Type
Judgment
Takes
about 2 min
Result in
% predicted

How Random

~2 min

  1. Press Left or Right 100 times, as randomly as you can.
  2. Before each press, a predictor makes its guess from the presses you have made so far.
  3. You see whether it guessed right after you press. Your result is how often it did.

Keyboard: F or ← for Left, J or → for Right.

Your result: — % predicted

How it's measured
  • Overconfidence

    Give ranges you're 90% sure of. How many hold?~3 min
  • Delay Discounting

    $50 now or $100 in a year? What is waiting worth to you?~2 min
  • Higher or Lower

    Is the hidden number higher or lower? Keep your streak going.~2 min
  • Reversal Learning

    One card wins more often, until the two swap. Can you keep up?~2 min
  • Time Perception

    Stop a hidden timer at the target time.~1 min
  • 1 Second

    Start the stopwatch. Stop it at exactly 1.000.~30 sec

How does the predictor guess your next press?

It looks at your last 5 presses and counts what you pressed next every earlier time those 5 came up. Whichever side followed more often is its guess. If that pattern is new, it tries your last four, then three, two and one; with nothing to go on, it picks the side you have pressed more, or flips a coin.

It guesses from your presses before the one it is guessing, never from that press, and you only see whether it was right after you press. Nothing is kept between games: it learns from you, this game, and no one else.

Why is it so hard to press at random?

Because real randomness looks streaky, and people avoid streaks. A fair coin switches sides about half the time and often lands the same way several times in a row; people asked to be random tend to switch too often and break up long runs, a pattern Willem Wagenaar described in his 1972 review of studies on making random sequences. A predictor that remembers your habits can use exactly that.

Here is how long the longest run of one side gets in 100 fair coin flips, worked out exactly by the same code that scores your result. On average the longest run is 7.0, so a run of five or six in your presses is not a sign of a pattern; never having one is.

A run of at least…Chance in 100 fair flips
4 in a row100%
5 in a row97%
6 in a row81%
7 in a row54%
8 in a row31%
10 in a row9%

How is your randomness result calculated?

Your result is the share of your 100 presses the predictor got right. Pure guessing gets about half, and a fair coin, which no predictor can read, is guessed right between 40% and 60% of the time in 95 games out of 100 (the binomial distribution, computed). Above 60% means the predictor found a pattern in you; below 40% usually means you were outguessing it on purpose, which isn't random either.

Beside the score are your switches from one side to the other (a coin averages 49.5 in 100 presses), your longest run (a coin averages 7.0) and how many times you pressed Left.

How can you be less predictable?

  • Let runs happen. Three or four of the same side in a row is ordinary for a coin. Breaking every run is the commonest giveaway.
  • Don't keep score in your head. Trying to press each side equally often makes the next press easier to guess.
  • Borrow randomness from outside. People can't generate randomness on demand, but they can read it: the last digit of something changing, odd for Left and even for Right, beats intuition.

For how well you judge what you can't count, try the Estimation Game; for how sure you are of what you know, the Overconfidence Test.

Questions

What is a good score?
Close to 50%. A fair coin is guessed right 40% to 60% of the time in 95 games out of 100, so anything in that range means the predictor found nothing to use.
Can I beat the predictor?
Yes, if you work out its guess and press the other side, but that is a pattern too: it pushes your score well below 50%, and a coin wouldn't do that.
Does the predictor learn from other players?
No. It only uses your presses in the game you are playing, and they are gone when you leave.
Why a hundred presses?
With fewer, luck swings the score a long way. At 100, a coin stays between 40% and 60% in 95 games out of 100, narrow enough to tell a pattern from chance.
Can I use the keyboard?
Yes. F or the left arrow presses Left, and J or the right arrow presses Right.