Rock, Paper, Scissors, explained.
Rock, Paper, Scissors is a game of cyclic dominance: each action beats one alternative and loses to another, so predictable pure choices can be exploited.
Why it happens
A mixed strategy specifies probabilities over actions before the outcome is known. With symmetric payoffs, a uniform mix protects against exploitation in expectation. Against a known fixed bias, a particular pure action may instead offer a positive expected score.
Rock, Paper, Scissors has cyclic dominance: rock beats scissors, scissors beats paper, and paper beats rock. No single pure action is an equilibrium of the symmetric game.
Read the result
Increase the bot's rock probability, then compare paper's expected score with actual rounds. The bot divides its remaining probability equally between paper and scissors. A winning expectation does not guarantee winning the next round.
A worked example
A bot favors rock
The bot chooses rock 80% of the time, and paper or scissors 10% each.
Always choosing paper wins 80%, draws 10%, and loses 10%, giving expected score +0.7 per round.
A uniform human mix gives zero expected score. It protects against an unknown opponent, but does not exploit this known fixed bias.
OPTIONAL DEEPER DETAILGo deeper: inside the model
Inside this model
The bot independently chooses rock with the selected probability and divides the remainder equally between paper and scissors. Wins score +1, draws 0, and losses −1. The summary shows expected scores for each pure action and the one-third uniform mix. The bot never observes the current human move before sampling.
Where this idea is useful
A practical use
When an opponent can exploit repeated choices, deliberately randomizing may protect against prediction. A biased opponent can also create an exploitable pattern.
A common misconception
“One move is intrinsically strongest.”
Every move has a counter. Its expected value depends on the opponent's probabilities and the payoff rules.
What this explanation leaves out
- This bot is fixed rather than adaptive. Real people have sequence habits, and unequal payoffs would change the appropriate probabilities. The game is traditional; the source explains its mathematical structure.
Does randomizing mean choosing whatever feels random?
Not necessarily. People can produce predictable sequences while feeling spontaneous. A mixed strategy is a defined probability rule, not just a lack of conscious planning.
Are you protecting against prediction or exploiting a known pattern?
Further reading
Explore the original research or the teaching reference behind this experiment.