Design an Online Chess Platform: Elo Matchmaking, WebSocket Moves, Game Clocks

Design an online chess platform in the style of Chess.com. Players look for a game and are matched with an opponent of similar strength by Elo rating, play the game in real time with moves delivered over WebSocket connections, and each player has a game clock that must be enforced fairly. Focus the design on three areas: **rating-based matchmaking**, **real-time move delivery over WebSockets**, and **game clocks (timers)**. Cover the rest of the platform (accounts, game history) only as needed. ```hint Who is the source of truth Decide which component owns each game's state and clocks, and what every other component is allowed to do with a move before that owner has accepted it. ``` ```hint Matching is a trade-off over time Consider how the acceptable rating gap for a waiting player should change the longer that player waits. ``` ### Clarifying Questions - How many concurrent players and games at peak, and which time controls (bullet, blitz, rapid, daily)? - Must players be matched within a few seconds, or is waiting longer for a closer rating acceptable? - How should network latency be treated on the clock: should a player be charged for time their move spent in transit? - What happens when a player disconnects in the middle of a game? - Are spectators, chat and anti-cheat in scope? ### What a Strong Answer Covers - Capacity estimates for connections, games and moves per second - A matchmaking design that balances rating closeness against wait time, and scales across many waiting players - A real-time architecture: connection handling, routing a game's traffic to its owner, fan-out to both players and spectators, and reconnection - Server-authoritative clocks, fair treatment of latency, and correct handling of time forfeits and disconnections - Move validation, game state persistence, recovery when a game server fails, and rating updates after the game ### Follow-up Questions - A game server crashes in the middle of a bullet game. What do the players experience, and how does the game resume? - How would you support 100,000 spectators watching one top-level game? - How would you detect players who are using a chess engine? - How would you prevent rating manipulation, such as two accounts deliberately losing to each other?