Skip to main content
Formal Specification

Turkmen Knucklebones — Formal Specification

Turkmenistan · dexterity-small · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 5
# Aşyk oýny — AI-optimized specification ## General idea A traditional Central Asian knucklebone game combining four-faced stochastic bone rolls with target-shooting trajectory physics. Players wager astragalus bones in a central line or pot, determine turn priority through a four-faced orientation toss (*alçy* > *towha* > *jikge* > *tükge*), and throw heavy lead-weighted shooter bones (*saka*) to knock target bones out of a boundary zone or match landed faces. ## Formal model - **Abstraction taken:** Hand wrist flicks, ground friction, and knucklebone bounce dynamics are modeled as a 4-state multinomial random variable for rolling/tossing, and a 2D line-collision resolver parameterized by throw power and aim accuracy for the *Bükdüm* target shooting variant. - **Player/agent count:** 2 to 6 individual players. - **Turn model:** Strictly sequential, ordered by opening toss hierarchy. ## Formal rules R1. **Astragalus Four-Face Probability Distribution:** - Each astragalus bone landing face has an intrinsic empirical probability: - Face 1: `Alçy` (ventral hollow) — [gap-fill] P(Alçy)=0.15 (highest rank). - Face 2: `Towha` (dorsal flat) — [gap-fill] P(Towha)=0.25. - Face 3: `Jikge` (lateral groove) — [gap-fill] P(Jikge)=0.30. - Face 4: `Tükge` (lateral convex) — [gap-fill] P(Tükge)=0.30. R2. **Setup & Pot Contribution:** Each of N players contributes k target *aşyks* (default k=2) placed upright in a straight line along the target mark (y=0). Total target pool =N⋅k bones. A shooting line is marked at distance y=−5 m. R3. **Turn Order Toss:** All players roll their *saka* simultaneously from the shooting line. Order is determined by landing face rank (`Alçy` > `Towha` > `Jikge` > `Tükge`); ties are broken by proximity to the target row. R4. **Shooting Turn (*Bükdüm*):** - The active player chooses aim target xaim∈[−X,X] and throw velocity v∈[vmin,vmax]. - A shot resolver simulates the *saka* trajectory with noise ϵ∼𝒩(0,σ2(skill)). - Any target *aşyk* knocked completely beyond the boundary line (|y|≥1.0 m) is captured by the shooter. - If the shooter captures at least 1 target bone AND the *saka* lands in `Alçy` or `Towha`, the shooter earns an immediate bonus shot; otherwise the turn passes to the next player in order. R5. **Victory Condition:** The round terminates when all target bones in the row have been captured. The player with the highest count of captured bones wins the stakes. ## Edge cases - Tie at the end condition: If two players capture an equal number of bones, tied players play a 1v1 shoot-off at a single centered target bone. - A player with no legal move: Not applicable — shooting a *saka* at the target line is always legal while target bones remain. - Repetition / cycles: Bounded because target bones are strictly removed upon capture; if all players miss across 3 full rounds of rotation, [gap-fill] the shooting line is moved 1 meter closer to ensure termination. - Simultaneous claims on the same resource or position: Turns are strictly sequential; target bones are captured by the active shooter on their individual turn. - Player elimination mid-round; turn order repair: If a player disconnects, their uncaptured contributed bones remain in the target row; turn order skips the disconnected player. - Resource exhaustion (pieces, tokens, space, time): Target bone pool is finite and strictly decreases upon successful hits. - Termination guarantee: why the game provably ends: Guaranteed by finite target pool size combined with the loop-breaking distance reduction rule. ## State representation - Persistent state: Player profiles, cumulative career bones won/lost, match turn order list. - Per-player state: `saka_weight: float`, `skill_accuracy: float`, `captured_bones_count: int`, `player_id: int`. - Per-turn ephemeral state: Positions of remaining target bones `List[Tuple[float, float, bool]]`, active shooter ID, sampled throw trajectory, landing face of *saka*, captured subset. - Observable information: Fully observable — all bone positions, orientations, and scores are visible to all players. ## Action space - `RollForOrder()`: Cast *saka* for turn order determination. - `ShootSaka(aim_x, velocity)`: Aim and hurl *saka* at target row. Preconditions: Player is active in turn order; remaining targets >0. ## Playability notes - Single-player-vs-AI: Natural — 1 human player aims and throws against AI knucklebone shooters in turn rotation. - Group games (n>4): Fully scalable to any group size (2 to 6) without structural loss. - Physical-skill games: Physical finger-flick, hand weight balance, and wrist snap are abstracted into the aim parameter, throw velocity, and Gaussian error function. - What is lost in the adaptation: The tactile feel of lead-filled bone density, the clattering acoustic resonance of colliding sheep bones on hard ground, and the traditional inspection of bone smoothness are replaced by numerical probabilities. ## Implementation hints Low complexity. The state is simply a 1D array of bone coordinates on a line plus integer bone tallies. Branching factor is continuous (aim coordinate) but easily discretized into discrete target bone choices. Suggested first milestone: implement the 4-face probability roll engine and turn-order hierarchy, then add the 2D line-shooting collision logic.

Download as JSON

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

Download JSON
Cite this page

Turkmen Knucklebones (Turkmenistan). (n.d.). Games of Countries. https://gamesofcountries.com/games/tm-turkmenistan/asyk-oyny

@misc{gamesofcountries-tm-turkmenistan-asyk-oyny,
  title = {Turkmen Knucklebones (Turkmenistan)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/tm-turkmenistan/asyk-oyny}
}