AI ENGINE
• Minimax search with alpha-beta pruning
• Custom evaluation: material + position + pawn structure
• Heavy center pawn bonus in first 12 moves
• Knights de-prioritized early game
• Two dials per level: depth (plies searched) and width (how many of the search's own top moves it then picks from, at random)
• width 1 = always the best move found; width 5 = one of its top 5
• Levels 1–12 climb: 2 plies wide → 4 plies at width 1, measured, not assumed
• There is no “aggression” dial any more. It used to negate a move's score, scale it by 1.6 and take the maximum — which is arithmetically the same move as taking the minimum of the unscaled score, tie-break included. It selected nothing, so it is gone rather than re-described
TIMERS (White vs Orange)
• White timer starts on your turn
• Orange (Computer) timer starts instantly after you move
• No duplicate intervals, no overlapping timers
• Format: MM:SS
THINKING TIME SYSTEM (v3.0)
• Base pause: depth × 600ms + 0.8s (a pause before the reply, not the search)
• Deep Search: forces depth 5 at width 1, still inside the active thinking limit
• Level 12: 3.5–8.5s (randomised pause, capped at active limit)
• Thinking 10s / Thinking 6s: caps the pause, and anything deeper than 3 plies searches one ply less — so Levels 10–12 and Deep Search change, Levels 1–9 do not
FEATURES (v3.0)
• 10 Piece Themes: 5 Outline + 5 SOLID filled pieces
• Live status + captured label updates on theme change
• Last Move Highlight with toggle
• AI Recommend Move button
• Every level states its own depth, width and timing on its own card
CURRENT LEVEL
Select a level to see exactly what it does.
Last updated: v2.9.5 — 10 Themes (5 Solid) + Live status + Captured color labels