Skip to main content
Formal Specification

Bavarian / Austrian ice stock curling — Formal Specification

Austria · winter-ice · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Eisstockschießen — AI-optimized specification ## General idea A turn-based team target game played on an ice sheet. Two teams of 4 agents take turns delivering gliding stocks toward a mobile target puck (*Daube*). Deliveries dynamically alter the target field geometry through collisions, displacements, and knockouts. The active throwing side is determined dynamically by which team currently trails in closest proximity to the target. ## Formal model - **Abstraction taken:** The continuous ice surface and projectile physics (friction, curl, kinetic elastic collision) are abstracted into a 2D Euclidean coordinate space representing the 6×3 metre target house, with deliveries modeled as vectors with Gaussian aim/speed variance. - **Player/agent count:** 2 teams (A and B), 4 agents per team (A1…A4, B1…B4). - **Turn model:** Strictly sequential, dynamically determined by the *trailing distance* rule. ## Formal rules R1. Setup: End k∈[1,6] initializes with the Daube position D=(0,0) at the house center. Each team has 4 unthrown stocks in hand. R2. Opening throws: Agent A1 executes `deliver_stock()`, establishing initial stock position SA,1. Agent B1 executes `deliver_stock()`, establishing SB,1. R3. Dynamic turn allocation: At any state with unthrown stocks remaining: - Let dmin⁡(A)=mini⁡‖SA,i−D‖ and dmin⁡(B)=minj⁡‖SB,j−D‖ among valid in-house stocks. - If dmin⁡(A)<dmin⁡(B), Team B is trailing and must execute the next delivery with their next available agent. - If dmin⁡(B)<dmin⁡(A), Team A is trailing and must execute the next delivery. - If a team runs out of unthrown stocks, the other team delivers all remaining stocks in sequence. R4. Collision and Displacement: A delivery trajectory (v,θ) resolves potential elastic collisions with standing stocks and the Daube. - If D is displaced within the 6×3 m boundary, D updates to the new resting coordinate. - If D exits the boundary, D resets immediately to (0,0). - Any stock exiting the boundary is marked `out_of_play` and removed. R5. End scoring: Once all 8 stocks are delivered: - Let winning team W∈{A,B} be arg⁡minT⁡dmin⁡(T). - Winning team receives 3 points for the best stock. - Winning team receives +2 points for each additional own stock SW,i where ‖SW,i−D‖<minj⁡‖SL,j−D‖ (opponent's closest stock). - Losing team L receives 0 points. R6. Match victory: After 6 ends, team with the highest cumulative score wins. ## Edge cases - **Tie at the end condition:** [gap-fill] If aggregate points after 6 ends are tied, a 7th sudden-death end is played to decide the winner. - **A player with no legal move:** Not applicable; delivery is always legal as long as stocks remain in hand. - **Repetition / cycles:** Bounded strictly; each end consists of exactly 8 deliveries, and the match is bounded at 6 ends (plus at most 1 tie-breaker). - **Simultaneous claims on the same resource or position:** Deliveries are sequential. If two opposing stocks rest at the exact same measured distance from the Daube, [gap-fill] the stock delivered earlier holds precedence. - **Player elimination mid-round; turn order repair:** No eliminations during play; all 4 players per team deliver exactly 1 stock per end. - **Resource exhaustion:** Exactly 4 stocks per team per end; no stock replenishment occurs mid-end. - **Termination guarantee:** Guaranteed by finite ends (K=6) and strictly bounded 8-turn sequence per end. ## State representation - **Persistent state:** Match score scoreA,scoreB, current end index k∈[1,6]. - **Per-end state:** Daube coordinate D∈ℝ2, list of active stock coordinates {SA,i},{SB,j}, set of remaining unthrown agents per team. - **Observable information:** Fully observable state (all coordinates on the ice surface are public knowledge to all agents). ## Action space - `deliver_stock(velocity, angle, spin)`: Deliver a stock with real-valued velocity v∈[vmin⁡,vmax⁡], aim angle θ∈[−π/6,π/6], and rotation spin ∈{left,right}. ## Playability notes - **Single-player-vs-AI:** Excellent as a curling-style physics/tactics challenge. - **Group games (n>4):** Standard 4v4 reduces cleanly to 1v1 where each agent controls all 4 team deliveries per end without altering game-theoretic depth. - **Physical-skill games:** Delivery execution modeled via Gaussian perturbation around intended aim vector: (vact,θact)∼𝒩((v,θ),Σ(skill)). - **What is lost in the adaptation:** Tactile perception of ice pebble roughness, temperature effects on ice glide speed, and physical broom sweeping. ## Implementation hints - Implement 2D physics using simple disc collision math (elastic collision resolution with restitution coefficient ≈0.85). - Suggested first milestone: Implement the dynamic trailing turn resolver (R3) and the 3-2-2-2 scoring function (R5) using fixed placement coordinates before attaching continuous physics.

Download as JSON

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

Download JSON
Cite this page

Bavarian / Austrian ice stock curling (Austria). (n.d.). Games of Countries. https://gamesofcountries.com/games/at-austria/eisstockschiessen

@misc{gamesofcountries-at-austria-eisstockschiessen,
  title = {Bavarian / Austrian ice stock curling (Austria)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/at-austria/eisstockschiessen}
}