Skip to main content
Formal Specification

Knucklebones (Romanian Astragalus Game) — Formal Specification

Romania · dexterity-small · Confidence: high

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 3
# Arșice — AI-optimized specification ## General idea A turn-based, precision target knocking and token acquisition game. Agents take turns throwing weighted shooter bones at a line of staked target knucklebones arranged inside a circular arena. Collisions displace target bones; any target knocked completely outside the circle boundary is permanently captured by the active thrower. ## Formal model - **Abstraction taken:** 2D continuous arena with radius R=0.5m. Shooter trajectory is modeled as a 2D landing coordinate (x,y) with 2D Gaussian dispersion around the target point. Rigid-body elastic collisions transfer linear momentum to target bones. - **Player/agent count:** 2 to 4 agents (default 2). - **Turn model:** Sequential round-robin turns until all target bones are cleared from the ring. ## Formal rules R1. Setup: - Circular arena of radius R=0.5m centered at (0,0). - Each of the K agents wagers 2 target bones. Total targets N=2K arranged along the diameter line y=0 at equidistant intervals x∈[−0.35,+0.35]. - Per agent i: `captured_count` ∈ℕ (starts at 0). R2. Shooter Action: - Active agent chooses aim coordinate (xaim,yaim) along the target line and throw force F∈[1..5]. - Shooter landing coordinate: 𝐩land=(xaim+ϵx,yaim+ϵy)(𝝐∼𝒩(𝟎,σ2𝐈/skill)) R3. Collision Resolution: - For each remaining target bone 𝐭k: - If ‖𝐩land−𝐭k‖≤rimpact (r=0.08m): - Displacement vector Δ𝐭k=𝐭k−𝐩land‖𝐭k−𝐩land‖×(0.3×F). - 𝐭k←𝐭k+Δ𝐭k. R4. Boundary Knockout & Capture: - For any target bone 𝐭k, if ‖𝐭k‖>R (outside ring): - Target is permanently captured by active agent: `captured_count[active] += 1`. - Target is removed from the arena. - [gap-fill] Bonus Shot on Boierul: With probability P=0.15×skill, the shooter lands *Boierul* face up, granting 1 immediate bonus throw to the active agent. R5. Turn Handoff: - Turn passes to the next agent in round-robin sequence. R6. Win Condition: - The game ends when all N target bones are cleared from the ring. The agent with the highest `captured_count` wins. ## Edge cases - Tie at the end condition: If two agents finish with identical captured tallies, [gap-fill] a 1-throw sudden-death target toss resolves the tie. - A player with no legal move: Not applicable — shooting action is always legal while target bones remain in the ring. - Repetition / cycles: Bounded — each round has non-zero probability of displacing targets outside the ring; targets are monotonically removed until 0 remain. - Simultaneous claims on the same resource or position: Not applicable — sequential single-shooter turns. - Player elimination mid-round; turn order repair: Not applicable (all agents participate until full arena clearance). - Resource exhaustion: Exactly N=2K target bones; total captured count ∑ci=N. - Termination guarantee: Monotonic decrement of target pool guarantees termination in finite turns. [gap-fill] Hard 40-turn cap awards uncaptured bones to thrower with nearest miss. ## State representation - Persistent state: - `target_positions`: List of 2D coordinates of remaining bones. - `captured_counts`: Array of K integers. - `active_agent`: Integer index 0..K−1. - `turn_count`: Integer counter. - Per-turn ephemeral state: - `shooter_landing`: `(x, y)`. - `bonus_earned`: Boolean. - Observable information: Fully observable. ## Action space - `shoot_bone(aim_x, aim_y, power)`: - `aim_x, aim_y`: Floats within arena radius. - `power`: Float ∈[1.0,5.0]. ## Playability notes - Single-player-vs-AI: Engine-ready. Highly engaging geometric targeting puzzle balancing cluster strikes against boundary knocking angles. - Group games (n>4): Parameterized for K=2..6 agents. - Physical-skill games: Physical weighted bone throwing and collision physics converted into 2D trajectory distributions and momentum transfers. - What is lost: Tactile click of real sheep bones, smooth surface of leaded batali, and sunny village porch atmosphere. ## Implementation hints - Implement as a simple 2D collision step. - AI shooting heuristic: evaluate candidate aim points (x,y) that maximize expected target displacements past R=0.5.

Download as JSON

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

Download JSON
Cite this page

Knucklebones (Romanian Astragalus Game) (Romania). (n.d.). Games of Countries. https://gamesofcountries.com/games/ro-romania/arsice

@misc{gamesofcountries-ro-romania-arsice,
  title = {Knucklebones (Romanian Astragalus Game) (Romania)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/ro-romania/arsice}
}