✂️Can you beat a browser at rock paper scissors?

✓ Tested & verified Updated: How the randomness works

Rock · paper · scissors

Make your throw

The browser throws crypto-randomly — it cannot read your mind, only your luck

Score this session
You 0 · Browser 0 · Draws 0

Game night, drawn

Weekly

One game, one way to decide who starts, and one rule to change — drawn fresh each week. For the dice, cards and coin in the drawer.

Free, no account. One click unsubscribes and deletes your address.

Try also

Can you win at rock paper scissors against a computer?

You cannot reliably beat rock paper scissors here — and that is the interesting part. Against a human, the game is one of reading habits: people throw rock first more often than chance, repeat a winning throw, and switch after losing twice. Against this page there is nothing to read. Each browser throw is drawn fresh from crypto.getRandomValues, with no memory of what you played, so every round is an independent one-in-three.

The rules are the ones you already know: rock blunts scissors, scissors cut paper, paper wraps rock, and identical throws are a draw. Over many rounds that produces roughly a third wins, a third losses and a third draws for any strategy, including the same throw every time.

A running score under the board counts wins, losses and draws for the session. It resets when you close the tab, because nothing is stored — which is also why a streak of six losses is not the game learning your patterns. It is just six rounds.

Nine pairings, resolved the same way they have been resolved in playgrounds for a century.

You throwBrowser rockBrowser paperBrowser scissors
RockDrawLossWin
PaperWinDrawLoss
ScissorsLossWinDraw

Three of the nine cells are draws, three are wins and three are losses, which is where the flat 33.3% each comes from. It holds for any strategy: throw rock every round for an hour and your results will converge on the same thirds as throwing at random, because your opponent is not paying attention.

The one thing to expect is streaks. In a hundred rounds, a run of five losses is unremarkable and a run of seven is not surprising. Streaks are what randomness looks like from close up, which is the whole subject of how random is random.

Need a strictly two-way answer instead? Use the coin flip. Need to separate more than three people? The random name picker scales to any list.

Frequently asked questions

How do you win at rock paper scissors?

Against a person, by exploiting their patterns: opening throws skew towards rock, players tend to repeat what just won and to switch after a loss, and few people can produce a genuinely random sequence when asked. Against this page, none of that applies — the browser throw is drawn from a cryptographic generator each round, so no strategy beats any other over time.

Is the browser cheating by looking at my throw first?

No. The throw is drawn before the round is resolved, from crypto.getRandomValues, and it does not consult your input, your history or a win-rate target. That is worth stating plainly because most people assume any computer opponent is quietly rubber-banding — the reason we do not is that a rigged tiebreaker is worthless for settling anything.

What are the odds of a draw?

One in three, or 33.3%, whatever you throw. There are nine possible pairings of two throws, three of which match. Over a long session you should land near a third wins, a third losses and a third draws, with visible streaks along the way — random sequences produce runs far more often than people expect them to.

Is rock the best opening throw?

No throw is better than another against a random opponent. The folklore about rock exists because human players open with rock more often than a third of the time, which makes paper a slightly profitable first move against a stranger in a pub. Against this page that edge is exactly zero, and against a friend who knows the folklore it is a coin flip anyway.

Can two people use this to settle something?

Better to use it as a neutral third party than as an opponent: agree that heads-equivalent is a win for one of you, then have one person play a single round. For a genuine two-person contest the coin flip or the yes or no wheel is cleaner, because both players see the same single result rather than one of them making a choice.

Does the score save between visits?

No. The tally lives in the page for the current session and disappears when you close or reload the tab. Nothing is written to a server and no account exists, so there is no lifetime record to chase — which is either a limitation or a mercy, depending on how the last ten rounds went.

How the randomness works

Each round draws one value from crypto.getRandomValues and maps it to rock, paper or scissors using rejection sampling, so all three throws are equally likely with no modulo bias. The draw is independent of your selection, of previous rounds and of the running score — there is no adaptive model, no difficulty setting and no target win rate. The result is then resolved with the standard rules, and the session tally is incremented in memory only. Nothing about the game is stored or transmitted, and reloading the page starts a genuinely fresh sequence.

Sources:

Why beating a random opponent proves nothing

Rock paper scissors is a genuinely fair mechanism only when both players choose independently and simultaneously. Everything interesting about the game as humans play it comes from the fact that people are bad at independence: they anchor on their last throw, mirror an opponent under pressure, and produce sequences with far too few repeats to be random.

Playing against a crypto-random opponent strips all of that out, which makes the page useful in two ways. As a game it is a calibration exercise — a way of seeing how badly a losing streak distorts your sense of the odds. As a tool it is a one-in-three decider, handy when three people or three options need separating and a coin cannot do it.

What it will not give you is a story. There is no opponent to out-think, and a five-game winning run says nothing about you at all.