Skip to main content
Formal Specification

Azerbaijani draughts — Formal Specification

Azerbaijan · board-abstract · Confidence: medium

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: directGap-fills: 1
# Azərbaycan daması — AI-optimized specification ## General idea An orthogonal-movement draughts variant, structurally distinct from diagonal draughts (international/English checkers) in that both movement and capture occur along ranks and files rather than diagonals, and captured pieces are removed the instant they are leapt rather than at turn end. Capture is strictly mandatory and, when multiple capture lines exist, the rules force the single line that captures the most pieces — removing capture-choice as a decision entirely whenever more than one option differs in yield. ## Formal model - **Abstraction taken:** None needed beyond standard board-game formalization — this is a fully specified, deterministic abstract strategy game already given in rule-complete form by the entry. - **Player/agent count:** 2 (White, Black), 16 pieces each. - **Turn model:** Strictly alternating, sequential; White moves first (entry §1.5). A turn is a single move or a single (possibly multi-jump) capture sequence. ## Formal rules R1. Board: an 8x8 grid of 64 squares (uncheckered — all squares are playable, entry §1.1's "uniform squares"). Ranks 1-8, files a-h (standard algebraic labeling for reference). R2. Setup (entry §1.2-1.4): White's 16 men occupy ranks 2-3; Black's 16 men occupy ranks 6-7; ranks 1, 4, 5, 8 start empty. R3. Man movement (entry §2): a man moves exactly one square to an orthogonally adjacent empty square in the direction "forward" (toward the opponent's back rank), "left," or "right" relative to its own facing — never backward, never diagonal. R4. Man capture (entry §3): a man captures an orthogonally-adjacent (forward/left/right only, never backward, R3.4) opposing piece by leaping over it to the empty square immediately beyond, removing the captured piece immediately upon the leap. If a further orthogonal capture is available from the landing square, the same piece must continue jumping within the same turn (mandatory multi-capture chain). R5. Promotion (entry §4): a man that reaches the opponent's back rank (rank 8 for White, rank 1 for Black) is crowned Dama. If it arrives there mid-multi-jump with further man-legal jumps available, it must continue jumping as a man and is promoted only once the jump sequence concludes with it resting on the back rank. R6. Dama movement (entry §5.1): a Dama moves any number of unoccupied squares along a single orthogonal line, in any of the four directions (forward, backward, left, right). R7. Dama capture (entry §5.2-5.3): a Dama captures an opposing piece at any distance along an orthogonal line (no intervening pieces) by leaping over it and landing on any empty square beyond it on that line; it may chain multiple captures in one turn, changing direction by 90 degrees after each jumped piece, provided each subsequent jump is itself legal. R8. Mandatory capture (entry §6.1): if the player to move has at least one legal capture (by any piece), a non-capturing move is illegal that turn. R9. Maximum-quantity rule (entry §6.2): if multiple distinct capture sequences are available (by any piece(s), including choices of which piece captures or which direction a multi-jump continues), the player must select a sequence that captures the strictly greatest total number of pieces among all available sequences; sequences capturing fewer pieces are illegal moves that turn, regardless of which pieces they would capture. R10. Win conditions (entry §7.1): a player wins the instant either (a) the opponent has zero pieces remaining, or (b) the opponent to move has zero legal moves (all remaining pieces immobilized, whether by blockage or by having no legal capture/move available under R8-R9). R11. Draw (entry §7.2): a draw occurs by mutual agreement, or is recognized in theoretically-drawn endgame balances (e.g., 1 Dama vs. 1 Dama) [gap-fill: the entry names this as a recognized draw class but gives no formal detection rule — no repetition count, no move-without-capture limit, and no specific enumerated set of drawn material balances beyond the one example, so this is not a mechanically checkable rule as given, only a human/engine judgment call]. ## Edge cases - Tie at the end: handled only by R11's non-mechanical draw-by-agreement/theoretical-balance clause; the entry provides no formal, checkable tie-detection rule. - A player with no legal move: explicitly a loss condition, not a stalemate-draw (R10b) — this differs from international chess conventions and must not be implemented as a draw. - Repetition / cycles: **not excluded by the entry's rules.** With no threefold-repetition or move-without-progress limit specified (R11's gap), two players can in principle repeat the same position indefinitely via non-capturing Dama maneuvering whenever neither side has a forced capture available — the formal ruleset as given does not prevent this. - Simultaneous claims on one position: not applicable — turns are strictly alternating and single-piece per turn; there is no shared-resource contest. - Elimination mid-round and turn-order repair: not applicable in the usual sense — captured pieces are simply removed (R4, R7); turn alternation (R2 model) continues unaffected by how many pieces either side has left, until R10 triggers. - Resource exhaustion: piece count is monotonically non-increasing (captures only remove pieces, R4/R7); reaching zero pieces is R10a, a defined win condition, not an anomaly. - Termination guarantee: **not guaranteed by the formal rules as specified.** Total piece count is monotonically non-increasing, which bounds how many *capturing* turns can occur (at most 30, since each removes exactly one piece and 32 pieces start on the board with 2 required to remain for the game to continue at all) — but the entry specifies no rule limiting *non-capturing* turns, so an arbitrarily long sequence of legal non-capturing Dama moves between two players with no available captures is not excluded. `termination_guaranteed: false` reflects this honestly, matching the entry's own reliance on draw-by-agreement (R11) rather than a formal stopping rule — this is analogous to chess needing an explicit 50-move/threefold-repetition rule that this entry's source material does not supply. ## State representation - Persistent state: `board` (64-square array, each empty/White-man/White-dama/Black-man/Black-dama), `side_to_move`. - Per-turn ephemeral state: for a multi-capture sequence, the currently-jumping piece's position and the set of already-captured pieces on this turn (needed to enforce R4/R7's continuation requirement and R9's maximum-quantity comparison, which must be computed over full sequences before the turn is committed). - Observable information: fully observable — the entire board is visible to both players at all times; no hidden information. ## Action space - `move_man(from, to)`: precondition: `to` is orthogonally adjacent to `from` in a legal direction (R3), `to` is empty, no capture is available anywhere on the board for the moving side (R8). - `move_dama(from, to)`: precondition: `to` is reachable along an unobstructed orthogonal line from `from` (R6), no capture is available anywhere on the board for the moving side (R8). - `capture_sequence(from, jump_list)`: precondition: the sequence is legal per R4/R7's leap-and-land mechanics, and among all legal capture sequences currently available to the moving side, this sequence captures the maximum total piece count (R9). ## Playability notes - Single-player-vs-AI: fully natural — this is a complete, deterministic two-player abstract strategy game with no adaptation needed beyond standard game-tree search (minimax/alpha-beta or MCTS), same as any draughts variant. - Reducing/scaling: not applicable — the game is inherently 2-player with no group-size adaptation question. - Physical-skill games: none — this is a pure abstract strategy game with zero physical component; no fidelity loss occurs in a digital implementation beyond the draw-detection gap noted above. - What is lost: essentially nothing mechanical; the tea-house social setting and the tactile experience of crowning a Dama with a second piece are the only non-formalizable elements, both purely cultural framing. ## Implementation hints Branching factor is moderate (bounded by up to 32 pieces × up to 4 directions, filtered hard by R8-R9's mandatory-maximum-capture rule, which sharply prunes the legal move set whenever any capture exists). The genuine implementation trap is R9: computing the legal move set requires enumerating *all* capture sequences (including multi-jumps with branching continuations) before filtering to only the maximum-length ones — a naive "first legal capture found" implementation will produce illegal games. First milestone: implement man movement/capture and R8's mandatory-capture check, verify against R9's maximum-quantity rule with hand-constructed multi-capture-choice positions, then add Dama promotion and long-range capture chains. Draw detection (R11) should be implemented via a conventional rule (e.g., move-count-without-capture or threefold repetition) chosen by the implementer, since the source specifies none.

Download as JSON

Download the full game corpus entry for this game as structured JSON.

Download JSON
Cite this page

Azerbaijani draughts (Azerbaijan). (n.d.). Games of Countries. https://gamesofcountries.com/games/az-azerbaijan/azerbaycan-damasi

@misc{gamesofcountries-az-azerbaijan-azerbaycan-damasi,
  title = {Azerbaijani draughts (Azerbaijan)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/az-azerbaijan/azerbaycan-damasi}
}