Skip to main content
Formal Specification

Heavy Stone Put — Formal Specification

Kosovo · dexterity-throw · Confidence: high

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 5
# Hedhja e gurapeshit — AI-optimized specification ## General idea A turn-based athletic field contest where agents compete over a fixed sequence of rounds to achieve the maximum single-throw distance with a heavy stone. Each agent chooses a throwing technique (effort level, release angle, and approach speed), balancing maximum expected distance against the risk of committing a scratch foul. ## Formal model - **Abstraction taken:** Physical biomechanics (muscular explosive force, grip friction, foot placement) are abstracted into an agent decision vector: effort intensity e∈[0.5,1.0], launch angle θ∈[30∘,55∘], and foot-fault risk tolerance r∈[0.0,1.0]. - **Outcome Sampling:** [gap-fill] The resulting distance D is sampled from a parameterized normal distribution: D∼𝒩(μ(e,θ,Skill),σ2) subject to a foul probability P(foul)=f(e,r,Skill). If a foul occurs, D=0. - **Player counts:** 2 to N agents (default benchmark: 2 agents). - **Turn model:** Strictly sequential throws over 3 preliminary rounds and 3 final rounds. ## Formal rules R1. **Setup:** Each agent i∈{1,…,N} possesses a fixed strength/skill parameter Si∈[0,100]. Throw sequence order is initialized randomly. R2. **Rounds Structure:** The tournament consists of K=6 sequential rounds (3 preliminary, 3 finals). R3. **Turn Execution:** On round r∈{1,…,K}, agent i submits an action tuple (e,θ,risk). R4. **Throw Resolution:** - Calculate foul threshold: [gap-fill] Pfoul=max⁡(0.02,0.5×(e2+risk2)×(1−Si/150)). - Sample ξ∼𝒰(0,1). If ξ<Pfoul, throw is a FOUL (di,r=0). - Otherwise, sample distance: [gap-fill] di,r=max⁡(0,Si×0.15×e×sin⁡(2θ)+𝒩(0,0.4)) R5. **State Update:** Record best throw for agent i: Dbest(i)=maxt≤r⁡di,t. R6. **Win Condition:** The agent with the highest Dbest(i) at the conclusion of round K is the winner. R7. **Tie-Breaker:** [gap-fill] If two agents have identical best distances, their second-best valid throw breaks the tie. If still equal, a sudden-death single throw is conducted. ## Edge cases - **Tie at the end condition:** Handled explicitly by R7 (second-best throw comparison, followed by single-throw sudden death). - **A player with no legal move:** Not applicable; an action tuple (e,θ,risk) is always valid within defined parameter bounds. - **Repetition / cycles:** Impossible; the tournament is strictly bounded to K=6 rounds. - **Simultaneous claims on the same resource or position:** Not applicable; turns are strictly sequential with independent metric scoring. - **Player elimination mid-round / turn order repair:** If an agent disconnects or fails to submit within timeout, [gap-fill] that throw is recorded as a foul (d=0) and turn order continues uninterrupted to the next agent. - **Resource exhaustion:** Not applicable; the throwing stone and pitch are non-depleting. - **Termination guarantee:** The game state advances strictly from round 1 to round 6. Termination is guaranteed in exactly N×6 turn steps (plus bounded tie-breaker if triggered). ## State representation - **Persistent state:** - `agent_ids`: List of N agents. - `current_round`: Integer ∈[1,6]. - `active_agent_idx`: Integer ∈[0,N−1]. - `throw_history`: Matrix of size N×6 containing floats (distances) or FOUL markers. - `best_distances`: Array of N floats. - **Observable information:** Complete public visibility of all past throw distances, rankings, and active round number. ## Action space - `throw(effort, angle, risk)`: - `effort`: Float ∈[0.50,1.00] (fraction of maximum physical exertion). - `angle`: Float ∈[30.0,55.0] (launch angle in degrees). - `risk`: Float ∈[0.00,1.00] (proximity to baseline during release). ## Playability notes - **Single-player-vs-AI:** Natural. Single human plays against one or more AI agents configured with various skill/risk profiles (e.g. conservative thrower vs high-variance risk-taker). - **Group games (n>4):** Scales trivially to any N participants without changing mechanics; reducing to 2 agents preserves 100% of the game's decision structure. - **Physical-skill games:** Physical power and throwing arc are fully represented by the risk-reward tradeoff between effort/angle and scratch fouls. - **What is lost in the adaptation:** The physical tactile strain of gripping a cold natural boulder and the sensory spectacle of alpine spectator reactions. ## Implementation hints - Extremely low state complexity; suitable for lightweight headless simulation or mobile implementation. - Optimal policy: If trailing on the final throw, agent should maximize effort and risk (e=1.0,r=1.0) to maximize the upper tail of the distance distribution. - First milestone: Build the probabilistic distance sampler and verify that higher effort increases both expected distance and foul variance.

Download as JSON

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

Download JSON
Cite this page

Heavy Stone Put (Kosovo). (n.d.). Games of Countries. https://gamesofcountries.com/games/xk-kosovo/hedhja-e-gurapeshit

@misc{gamesofcountries-xk-kosovo-hedhja-e-gurapeshit,
  title = {Heavy Stone Put (Kosovo)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/xk-kosovo/hedhja-e-gurapeshit}
}