Reconnaissance Blind Chess

( Not sure where to start? Try a game here! )

Play RBC

Game Modes

Ranked Matchmaking

Get paired against an automatically selected opponent. It may take a few seconds to find a matchup. The outcome of your game will affect the RBC leaderboard!

To play in a ranked game, you must create an account if you have not. See Ranking Protocol for more information about ranked matches.

Invite a Friend

Play RBC with a friend by emailing them a one-time-use link. You do not need an account for this, and the game will not affect the leaderboard.

Challenge a Bot

Play against the connected bot of your choice. If the list of bots is empty, none are available right now. You do not need an account for this, and the game will not affect the leaderboard.

Display Modes

Current Observation Only (Default)

Your opponent's pieces will only appear on your board within the most recent sense window.

Extra Piece Placement

You may place extra pieces on your board representing your opponent's pieces. This does not affect true game state but can aid memory or visualization. Displayed pieces will be automatically updated in response to sense observations.

Computer Assisstance

None (Default)

Play without any computer assistance.

Piece Tracking

The computer tracks all possible locations of opponent pieces and displays them in a semi-transparent overlay. This overrides the display mode as longs as tracking is active. If the number of possible boards exceeds 5,000 at the end of your turn, the assistant turns off.

Baseline Bots

Bot name Strategy description
random Senses and moves randomly.
attacker Senses randomly and tries to capture the opponent king with a classic 4-move "mate" or by attacking with a knight.
trout Keeps track of a naive single board state with basic, semi-random sensing of unoccupied squares and makes moves with Stockfish.
Oracle Keeps track of all the possible board states, sensing to minimize possible opponent states, and makes moves with Stockfish plus heuristics.
Marmot Uses a modified Monte Carlo counterfactual regret minimization algorithm for sensing and moving that leverages online outcome sampling and a heuristic state evaluation.