AI Difficulty Levels
Choose your challenge: Easy (random), Medium (adjacent), or Hard (full strategy).
We’ve built three distinct AI behaviours across Units 9-12: random moves, adjacent priority, and full strategy with defense and position bonuses. But players can only face the hardest version. This unit lets them choose their challenge.
The title screen now offers four options. Two-player mode remains, but vs Computer splits into Easy, Medium, and Hard — each using a different AI algorithm.
Run It
pasmonext --sna inkwar.asm inkwar.sna
The new title screen shows all four options. Press 1-4 to select your game mode.
The Three Difficulties
Each difficulty level uses a different cell selection algorithm:
| Level | Algorithm | Behaviour |
|---|---|---|
| Easy | Random | Picks any empty cell at random. No strategy. |
| Medium | Adjacent priority | Expands from existing territory. No defense. |
| Hard | Full strategy | Offense + defense + position bonuses. |
Easy is beatable by anyone. Medium requires some thought. Hard plays optimally within its evaluation limits.
Difficulty Constants
We define the three levels as constants:
This code sample could not be loaded. The file may be missing or the path may be incorrect.
The title screen repositions to accommodate four menu options instead of two.
Title Screen Selection
The keyboard reading now checks four keys instead of two:
This code sample could not be loaded. The file may be missing or the path may be incorrect.
Keys 1-4 map to bits 0-3 of keyboard row $F7. Each AI option sets both game_mode (to GM_VS_AI) and ai_difficulty (to the appropriate level).
AI Dispatch
The core change is in ai_make_move, which now dispatches based on difficulty:
This code sample could not be loaded. The file may be missing or the path may be incorrect.
Three paths:
- Easy: Calls
find_random_empty_celldirectly - Medium: Calls new
find_adjacent_only(AI neighbours only) - Hard: Calls
find_best_adjacent_cell(full strategy)
Medium Difficulty Algorithm
Medium difficulty uses a simplified version of the scoring algorithm — only AI adjacency, no defense or position bonus:
find_adjacent_only:
; ... scan all cells ...
; For each empty cell:
call count_adjacent_ai ; Only this - no defense/position
; ... keep best score ...This creates an AI that expands its territory but doesn’t actively block you or prioritise strategic positions.
Why Three Distinct Algorithms?
We could have used probability-based mixing (e.g., Easy = 70% random, 30% strategic). But distinct algorithms are:
- Simpler to implement — No random number checks per move
- More predictable — Each level has consistent behaviour
- Easier to debug — Clear separation of concerns
- Better learning progression — Players see distinct AI personalities
The algorithms we built in Units 9-12 map naturally to difficulty levels.
The Complete Code
This code sample could not be loaded. The file may be missing or the path may be incorrect.
Try This: Add More Levels
You could add intermediate difficulties:
AI_VERY_EASY equ 0 ; Random
AI_EASY equ 1 ; Adjacent only
AI_MEDIUM equ 2 ; Adjacent + defense (no position)
AI_HARD equ 3 ; Full strategy
AI_EXPERT equ 4 ; Full strategy + look-ahead (future unit)More keys are available on the same keyboard row (key 5 = bit 4).
Try This: Display Difficulty
Show the current difficulty during gameplay:
; After drawing UI, show difficulty level
ld a, (ai_difficulty)
; ... display "EASY", "MEDIUM", or "HARD" ...This reminds players which AI they’re facing.
What You’ve Learnt
- Menu expansion — Adding options to existing UI
- Dispatch patterns — Branching based on configuration
- Code reuse — Leveraging existing algorithms for difficulty
- Player choice — Giving control over game challenge
What’s Next
In Unit 14, we’ll add sound effects — distinct sounds for claims, errors, and victory to give the game audio feedback.