Skip to main content
Formal Specification

Golden Gourd (Mounted Archery at Suspended Target) — Formal Specification

Turkmenistan · equestrian-team · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 2
# Altyn gabak — AI-optimized specification ## General idea A traditional mounted archery contest where agents gallop on horseback down a timed shooting corridor and release an arrow at a suspended golden gourd target. The decision problem centers on gallop speed control, draw timing, aim elevation, and risk-reward tradeoffs between fast stable passing and high-precision aiming to achieve glancing hits, punctures, or complete target drops. ## Formal model - **Abstraction taken:** Equine galloping dynamics, wind deflection, and bow mechanics are abstracted into a sequential turn run with a parameterized shot resolver. Gallop speed determines the width of the legal shooting timing window, and agent aim parameters feed into a bivariate Gaussian hit resolver centered on the suspended target. - **Player/agent count:** 2 to n individual archer agents. - **Turn model:** Strictly sequential individual runs across fixed tournament rounds (R=3 to 5 runs per agent). ## Formal rules R1. **Setup:** A gallop track of length 100 m features a designated shooting corridor between x=40 m and x=60 m. The target gourd is suspended at height H=8 m at midpoint x=50 m. R2. **Run Execution:** On each run, the active agent chooses a target gallop speed v∈[vmin,vmax] and an aim offset vector (Δy,Δz). R3. **Shot Validity & Timing Window:** - The horse traverses the corridor in time ΔT=20v seconds. - An arrow must be released within the legal corridor (40≤xrelease≤60); otherwise the shot is scored 0 (foul / out of window). R4. **Hit Resolution & Scoring:** - [gap-fill] The arrow trajectory error (ϵy,ϵz)∼𝒩(0,Σ(v,skill)) increases with gallop speed v and decreases with archer marksmanship skill. - Resulting impact distance d=‖(Δy+ϵy,Δz+ϵz)‖2: - d≤0.10 m: Shatter / cord severed →3 points. - 0.10<d≤0.25 m: Puncture (lodged arrow) →2 points. - 0.25<d≤0.45 m: Glancing strike →1 point. - d>0.45 m: Miss / pole hit →0 points. R5. **Tournament Progression & Victory:** - Each agent performs K=4 regulation runs. Points accumulate into total score Si=∑k=1Ksi,k. - The agent with the highest cumulative score Si wins. - [gap-fill] If a tie occurs for 1st place, tied agents participate in a sudden-death single-arrow shoot-off at a smaller target diameter (d≤0.20 m). ## Edge cases - Tie at the end condition: Handled by R5 — resolved via single-arrow sudden-death shoot-off until the tie is broken. - A player with no legal move: Not applicable — committing a gallop speed and aim point is always possible on one's turn. - Repetition / cycles: Bounded strictly by the fixed number of regulation runs (K) and monotonic elimination in shoot-offs. - Simultaneous claims on the same resource or position: Not applicable — runs are strictly sequential and individual. - Player elimination mid-round; turn order repair: If an agent disconnects or fails to run, remaining agents maintain their sequential queue order. - Resource exhaustion (pieces, tokens, space, time): Exactly 1 arrow is consumed per run; arrows are reset each run. - Termination guarantee: why the game provably ends: Guaranteed by the finite fixed run count K×N plus sudden-death shoot-off with non-zero hit/miss variance. ## State representation - Persistent state: Leaderboard of cumulative scores `Dict[AgentID, int]`, current round index k∈[1,K], turn queue. - Per-player state: `skill_marksmanship: float`, `skill_horsemanship: float`, `runs_completed: int`, `score_history: List[int]`. - Per-turn ephemeral state: Active agent's chosen speed v, release coordinate xrelease, sampled trajectory offset, hit tier, awarded points. - Observable information: Fully observable — all scores, run replays, and target outcomes are public. ## Action space - `ExecuteRun(gallop_speed, aim_offset, release_tick)`: - `gallop_speed`: float in [8.0,15.0] m/s. Precondition: must maintain minimum gallop speed ≥8.0 m/s. - `aim_offset`: (Δy,Δz)∈[−1.0,1.0]2 m. - `release_tick`: time of release within the corridor entry/exit window. ## Playability notes - Single-player-vs-AI: Natural — 1 human player aims and times runs against simulated AI marksmen. - Group games (n>4): Sequential turn format scales trivially to any player count n without rule modifications. - Physical-skill games: The physical difficulty of drawing a composite bow while galloping without stirrup support is abstracted into the bivariate Gaussian variance model parameterized by speed and skill. - What is lost in the adaptation: The somatic thrill of riding at 40 km/h, horse wind resistance, and the muscular strain of the full-draw release are collapsed into statistical parameter selection. ## Implementation hints Low complexity. Simple sequential state machine with statistical sampling. Branching factor is continuous (aim/speed) but can be discretized into 5 speed levels and 9 aim zones. Suggested first milestone: implement a single-agent 4-run scoring loop with the R4 Gaussian resolver, then add multiplayer turn queuing and shoot-offs.

Download as JSON

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

Download JSON
Cite this page

Golden Gourd (Mounted Archery at Suspended Target) (Turkmenistan). (n.d.). Games of Countries. https://gamesofcountries.com/games/tm-turkmenistan/altyn-gabak

@misc{gamesofcountries-tm-turkmenistan-altyn-gabak,
  title = {Golden Gourd (Mounted Archery at Suspended Target) (Turkmenistan)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/tm-turkmenistan/altyn-gabak}
}