Skip to content
Lizely

Rock Paper Scissors Online

Challenge a transparent local opponent that changes its next counter from your disclosed move history.

Privacy: your files never leave your device. All processing happens locally in your browser.

How to use

  1. 1.Read the visible prediction source, predicted move, and queued counter; they are fixed before your next legal choice is recorded.
  2. 2.Choose Rock, Paper, or Scissors with the buttons or 1/R, 2/P, and 3/S. Ties still enter local history, while three wins complete and two losses close the run.
  3. 3.Play Scissors, Paper, then Rock to beat the queued Paper, Rock, and Scissors counters and finish with exactly 1,000 points.

About Rock Paper Scissors Online

Rock Paper Scissors is a deterministic three-choice game against a transparent local prediction rule. The opponent is not a language model, remote service, hidden randomizer, or person. It reads only the moves recorded in the current browser run, shows the source of its next prediction before you choose, and displays the counter it has already queued. That makes every round reproducible and lets you inspect how the opponent responds to history.

On the opening round, the fixed baseline predicts Rock and queues Paper. After at least one move, the opponent first checks whether your latest move appeared earlier with a following move. If those transition samples exist, it predicts their most frequent follower. If no usable transition exists, it predicts the most frequent move in your complete history. Equal counts always resolve in the disclosed order Rock, then Paper, then Scissors. The opponent then plays the move that beats its prediction.

A legal repeated choice is a new learning sample rather than an ignored duplicate. Ties are also recorded in history, because a tie can affect a later frequency or transition prediction, but ties change neither wins, losses, nor score. An invalid key is an atomic no-op. Each win awards 300 points, each loss subtracts 100 without taking the score below zero, and reaching three wins adds a 100-point completion bonus. Two losses close the run.

The exact perfect route is Scissors, Paper, Rock. The opponent plays Paper, Rock, Scissors, so all three rounds are wins and the final score is 300 plus 300 plus 300 plus the 100-point completion bonus: exactly 1,000. The disclosed failure route is Rock, Paper, Paper, Paper. Its outcomes are loss, tie, tie, loss, demonstrating both the baseline and a learned transition-pattern counter before the second loss ends the run.

Choose a move with the three large buttons, or use 1 or R for Rock, 2 or P for Paper, and 3 or S for Scissors. The screen shows the current prediction source, predicted player move, queued opponent move, wins, losses, ties, recent rounds, and exact rule explanation. Restart clears only this game's local run state and returns to the fixed opening. Completion and defeat freeze later choices.

The prediction table is verified independently. Tests restate the three moves, counter cycle, frequency counting, transition lookup, and Rock-first tie priority without importing the production planning function. Ten literal histories are checked against their predicted move, opponent counter, and strategy source. Separate stepwise replays compare the three-round winning route and four-round losing route, including proof that recorded history changes the opponent across Paper, Rock, and Scissors.

The word learning here has a narrow product meaning: deterministic counting within the current run. Nothing is trained, persisted to a server, inferred about the player, or personalized beyond the displayed move list. The opponent does not claim optimal play and the score is not a psychological, cognitive, educational, or employment assessment. Familiarity with the fixed route naturally makes replay easier.

All gameplay runs in the browser without an account, upload, API, model, or new dependency. The visible plan deliberately removes the secrecy common to game opponents and turns the match into a small prediction puzzle. Use Restart to explore how a different history changes the disclosed source and counter, or follow the exact route when you want a verifiable 1,000-point completion.

Methodology & sources

Use three legal moves and a deterministic local prediction rule. Open by predicting Rock. Thereafter, collect moves that historically followed the player's latest move; predict their most frequent value when samples exist, otherwise use full-history frequency, with Rock-Paper-Scissors tie priority. Queue the counter to that prediction before the player chooses. Record every legal choice and tie, award 300 per win, subtract 100 per loss with a zero floor, add 100 at three wins, and deadlock at two losses. Independently implement the history oracle, check ten literal plans, replay Scissors-Paper-Rock for exactly 1,000, replay Rock-Paper-Paper-Paper for two losses, and freeze terminal states.

Frequently asked questions

Does the opponent use AI, an API, or randomness?
No. It is a deterministic local counting rule that uses only the current run's visible move history and calls no model, API, random generator, or storage service.
How does the opponent learn my pattern?
It prefers the most frequent move that historically followed your latest choice. With no transition sample, it uses overall move frequency; equal counts resolve Rock, Paper, then Scissors.
What is the exact winning route?
Choose Scissors, Paper, and Rock. Those moves beat the disclosed Paper, Rock, and Scissors counters for three wins and exactly 1,000 points.

Related tools

Mini Games guides

View all