How the Three AI Styles Work
The AI does not use minimax. For every legal move it calculates a heuristic score and chooses the highest-scoring move. The selected style changes only these heuristic weights. Therefore students can observe how the same legal-move generator produces different behavior when the evaluation function changes.
| Parameter | What it rewards / penalizes | Aggressive | Balanced | Defensive |
|---|---|---|---|---|
| BEAR_OFF | Removing a checker from the board | 1800 | 2000 | 1900 |
| HIT_OPPONENT | Hitting an opponent blot | 1900 | 1200 | 700 |
| MAKE_POINT | Turning one own checker into a made point | 300 | 400 | 850 |
| COVER_BLOT | Moving a lone checker away / improving safety | 180 | 300 | 700 |
| LEAVE_BLOT | Penalty for exposing a single checker | -180 | -400 | -900 |
| HOME_BOARD | Moving a checker into the home board | 40 | 60 | 120 |
| STACKING_PENALTY | Penalty for over-stacking a point | -20 | -40 | -70 |
| RANDOMNESS | Small noise added to break near-ties | 30 | 10 | 4 |
Aggressive
Strongly prefers hitting exposed opponent checkers and accepts more blot risk. It uses a larger random tie-breaker, so its play is less predictable.
Balanced
Closely matches the original AI. It balances hitting, making points, protecting blots, and bearing off without strongly favoring attack or safety.
Defensive
Strongly rewards making safe points and protecting exposed checkers. It heavily penalizes leaving blots and is less interested in hitting unless the tactical gain is clear.
Important: changing the selector does not alter a game already in progress. Press Restart with Selected Style to start a new game with the new parameter set. In Human vs AI mode the computer is Black. In AI vs AI mode both sides use the selected profile.