Hey everyone! Based on your feedback about AI opponents, I've made a major improvement to how the AI selects moves. Here's what changed:
AI Improvements
Previously, the AI had a "style overlay" system that would sometimes override the engine's calculated best move with a different move that matched the opponent's personality. The idea was to give each bot a distinct play style, but in practice it caused higher-level opponents to make moves that didn't match their rating — especially at Normal and Master level.
I've removed the style overlay and replaced it with a pool-based selection system.
How it works now: The engine calculates all moves, then builds a "pool" of acceptable options based on the opponent's Elo rating. Higher Elo = smaller pool = stronger play. The AI's personality picks between moves within that pool, so style only influences the choice when multiple moves are equally good.
What this means: A 2300-rated opponent now actually plays like a 2300. A 1000-rated opponent makes realistic inaccuracies, not random blunders. Every AI plays at their advertised strength.
Personality is preserved. Opponents still feel different from each other, aggressive bots pick sharp options, positional bots pick quiet ones, but only within moves the engine confirms are appropriate for their level.
Fixed a bug: where AI opening moves could be played "blindly" without checking if they still made sense after the opponent deviated from the expected line.
This affects all AI opponents from Normal through Legend. Let me know how the new AI feels!