Find Nearby High Scores in a Large Game Leaderboard

Design a video-game score leaderboard with players, games, friendship relationships, and one stored high score per player-game pair. After a player finishes a game, the service should efficiently show nearby scores in a large dataset. ### Constraints The meaning of nearby, the number of neighbors, ranking ties, score direction, and whether results are global or friend-filtered are unspecified. Clarify these policies. Preserve a single high-score record per player-game pair rather than treating every play as an independent leaderboard entry. ### Clarifying Questions - Does nearby mean closest numeric scores, immediately higher/lower ranked players, or a range around the new score? - Is a higher score always better, and how should equal scores be ordered? - Are results global, friends-only, or both? - Is the lookup centered on the latest play's score or the player's stored high score? ```hint Separate high-score updates from neighbor queries An atomic best-score update and a range query over an ordered index solve different parts of the request. ``` ### What a Strong Answer Covers - Player-game score identity, atomic high-score changes, and a clear tie policy. - An ordered index supporting range or predecessor/successor queries. - Friend filtering, partitioning, hot games, and consistency between updates and displayed neighbors. ### Follow-up Questions - What if a player's latest score is below their personal best? - How would a friends-only leaderboard change the indexing or query strategy?