Skip to main content
Formal Specification

Upright Knucklebone — Formal Specification

Kazakhstan · dexterity-small · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 6
# Ompy — AI-optimized specification ## General idea Ompy is a high-stakes knucklebones target-shooting game centered around an upright, vertically balanced master bone (*ompy*) set in the middle of a line of staked pieces (*kön*). Players shoot lead-weighted master knucklebones (*saqa*) from a distant baseline, choosing between low-risk incremental shots at peripheral target bones or a high-difficulty direct jackpot shot at the central *ompy*, which immediately sweeps all remaining pieces on the board. ## Formal model - **Abstraction taken:** Projectile dynamics across the 10-metre court (throw velocity, ground skip, collision angle, and upright bone toppling physics) are replaced by a stochastic target-resolution function. [gap-fill] The probability of hitting a regular bone vs toppling the delicate vertical *ompy* is parameterized by target profile area and agent skill S∈[0,1], where the *ompy* target presents a significantly narrower tolerance window. - **Player/agent count:** 2 to 6 individual agents (or 2 teams of 2–3 agents). - **Turn model:** Sequential turn rotation with bonus throw chaining upon successful piece capture. ## Formal rules R1. **Court Setup & Stake Initialization:** - Court layout: Linear playing corridor of length 12 metres with a center transverse line at x=0 (*kön*) and two baseline throwing marks at x=−8.0 m and x=+8.0 m. - [gap-fill] Exactly 8 standard knucklebones (K1..K8) are placed lying along the *kön*, with 4 bones flanking each side of the center. - Exactly 1 bone is placed standing vertically on its narrow end at the exact midpoint (0,0), designated as the Bompy. - Total bones staked on table = 9 pieces. R2. **Turn Order Determination:** - All agents roll their *saqa* toward the baseline; the agent rolling the highest stance (Alshy>Täyke>Bük>Shik) shoots first, proceeding in descending order. R3. **Shooting & Target Selection:** - The active agent chooses a target from the remaining pieces on the *kön*: either one of the standard lying bones Kj or the standing Bompy. - The agent chooses shooting position: baseline (default) or an advanced court position (xs,ys) if an *advanced stance privilege* was earned on the previous turn. - The agent executes `shoot(target_id, position)`. R4. **Shot Resolution:** - **Regular Target Shot (Kj):** - Stochastic hit test: P(hit)=f(S,distance). - If hit and displaced ≥60 cm: Kj is removed from the court and added to the active agent's bank (+1 point). The active agent earns an immediate bonus throw. - **Ompy Jackpot Shot (Bompy):** - Stochastic hit test: P(topple)=g(S,distance), where g(S,d)<f(S,d) due to narrow profile. - If successfully toppled: The shooter claims Bompy (+2 points) *and* all remaining standard Kj bones on the *kön*, instantly ending the round. - **Miss or Insufficient Displacement:** - The shot fails to capture. - The *saqa* comes to rest at coordinate (xs,ys) with a sampled orientation. - If (xs,ys) is inside the court boundaries and the stance is Alshy or Täyke, the shooter gains `advanced_stance_privilege` for their next turn. - The turn passes to the next agent. R5. **End Condition and Round Scoring:** - A round ends immediately when: 1. The Bompy is toppled (jackpot sweep), OR 2. All standard Kj bones on the *kön* have been captured. - The agent with the highest total captured bones at the end of 3 rounds wins the match. - [gap-fill] If total points are tied after regulation rounds, a 1-on-1 single-throw sudden-death duel against Bompy is conducted. ## Edge cases - **Tie at the end condition:** Explicitly resolved by R5 [gap-fill] (sudden-death single throw against Bompy). - **A player with no legal move:** Not applicable; as long as targets remain on the *kön*, shooting is always legal. - **Repetition / cycles (and the loop-breaking rule):** Since non-zero skill guarantees positive capture probabilities, pieces will be cleared. [gap-fill] A hard cap of 40 total throws per round ends the round, dividing any remaining unclaimed bones equally. - **Simultaneous claims on the same resource or position:** Not applicable; turns are strictly sequential. - **Player elimination mid-round; turn order repair:** If an agent disconnects or forfeits, they are removed from rotation; turn advances to next active player index modulo active count. - **Resource exhaustion (pieces, tokens, space, time):** The 9 staked pieces are non-renewable per round and monotonically decrease to 0. - **Termination guarantee: why the game provably ends:** Every successful hit decreases the board piece count, and a single successful *ompy* hit clears the entire board in 1 step. Combined with the 40-throw ceiling [gap-fill], the game provably terminates. ## State representation - **Persistent state:** - `standard_bones`: Array of 8 structs `{id: 1..8, in_play: bool, pos: (0, y)}`. - `ompy_standing`: Boolean flag indicating if central *ompy* is standing. - `player_scores`: Array of N integers. - `active_player`: Integer 0..N−1. - `advanced_privilege`: Array of N structs `{eligible: bool, pos: (x, y)}`. - `round_throw_count`: Integer tracking total throws this round. - **Per-turn ephemeral state:** - `selected_target_id`: ID of chosen bone. - `shot_origin`: Baseline coordinate or advanced position. - `hit_outcome`: Enum `[MISS, REGULAR_CAPTURE, OMPY_JACKPOT]`. - **Observable information:** Fully observable. Positions and statuses of all pieces, scores, and advanced shot privileges are public. ## Action space - `shoot_target(target_id, use_advanced_pos)`: - Parameters: `target_id` ∈{K1..K8,Bompy}, `use_advanced_pos` ∈{true,false}. - Preconditions: Active turn; `target_id` is currently in play; `use_advanced_pos` requires valid privilege. ## Playability notes - **Single-player-vs-AI:** Natural. 1 human vs 1–5 AI agents. AI evaluates expected value of targeting regular bones (high P(hit), 1 point) vs jackpot *ompy* (low P(hit), k+2 points). - **Group games (n>4):** Reduces directly to 2 agents or 2 teams of 2–3 in alternating turns without loss of strategic structure. - **Physical-skill games:** Modeled via player throwing accuracy S∈[0,1]. - **What is lost in the adaptation:** Physical tension of balancing the *ompy* on its delicate natural tip, subtle ground friction variations, and ricochet collision dynamics. ## Implementation hints - Model the decision problem as a Markov Decision Process where the state includes remaining bone count on the *kön*; this yields an elegant dynamic programming solution for the optimal target choice threshold. - Set P(ompy hit)≈0.35×P(regular hit) to preserve the historical risk-reward balance. - First milestone: Implement court state, standard bone removal, jackpot board clear, and test with greedy baseline agents.

Download as JSON

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

Download JSON
Cite this page

Upright Knucklebone (Kazakhstan). (n.d.). Games of Countries. https://gamesofcountries.com/games/kz-kazakhstan/ompy

@misc{gamesofcountries-kz-kazakhstan-ompy,
  title = {Upright Knucklebone (Kazakhstan)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/kz-kazakhstan/ompy}
}