Skip to main content
Formal Specification

Sheepskin Bellows Jumping — Formal Specification

Bosnia & Herzegovina · strength-endurance · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: abstractedGap-fills: 3
# Skakanje na mijeh — AI-optimized specification ## General idea A turn-based athletic challenge where competing agents select jump trajectory, kinetic velocity, and landing posture parameters against a pressurized, low-friction target (an inflated greased sheepskin). The challenge balances extreme impact energy (needed to burst the membrane) against stability control (to avoid slipping off and scoring zero). ## Formal model - **Abstraction taken:** The biomechanical physics of the jump, foot impact pressure, and surface friction are modeled via a multi-variable stochastic function: Fimpact=mass⋅vapproach2⋅cos⁡(θ), while P(slip)=σ(kfriction−balance_control). A burst occurs if Fimpact≥Tburst and slip==False. - **Player/agent count:** 2 to N agents (default: 4 agents in round-robin bracket). - **Turn model:** Strictly sequential turns in fixed rotation order across 1 to 3 elimination rounds. ## Formal rules R1. **Setup:** Each agent has immutable attributes: `power` ∈[0,1], `balance` ∈[0,1], and `mass` ∈[60,100]. The sheepskin target has membrane threshold Tburst=100.0 and friction coefficient μgrease=0.15. Round counter r=1, max rounds Rmax⁡=3. R2. **Action selection:** On its turn, the active agent selects a continuous parameter vector (v,θ,p) where: - v∈[0,1]: Approach speed / commitment. - θ∈[0,π/2]: Angle of attack / trajectory arc. - p∈{FEET_FIRST,SEATED_DROP}: Chosen landing posture. R3. **Trial resolution:** - Effective downward force: F=(70+30⋅power)⋅v⋅sin⁡(θ)⋅(1.2 if p==FEET_FIRST else 0.9)+ϵF, where ϵF∼𝒩(0,5.0). - Stability check: Slip probability P(slip)=11+exp⁡(balance⋅10−v⋅12μgrease⋅10). Sample s∼Bernoulli(P(slip)). - Outcome determination: - If F≥Tburst and s==0: Outcome is **BURST** (Score = 1000+F). - If F<Tburst and s==0: Outcome is **HOLD** (Score = F). - If s==1: Outcome is **SLIP** (Score = 0). R4. **Round progression:** After all agents complete round r: - If at least one agent achieved **BURST**, the agent with the highest single score among them wins immediately, and the contest terminates. - If no agent achieved **BURST**, [gap-fill] the top 2 agents by cumulative score advance to round r+1, while the bottom agents are eliminated. R5. **Match termination:** If round Rmax⁡ concludes with no BURST, the agent with the highest single valid HOLD score is declared winner. ## Edge cases - **Tie at the end condition:** [gap-fill] If two agents tie with identical highest scores after round Rmax⁡, a single sudden-death jump off is conducted with an increased approach velocity cap. - **A player with no legal move:** Not applicable; continuous parameter selection within the unit box [0,1]2×{FEET,SEATED} is always available. - **Repetition / cycles:** Bounded strictly by Rmax⁡=3 rounds; no infinite state cycles are possible. - **Simultaneous claims on the same resource:** Not applicable; jumpers take turns sequentially. - **Player elimination mid-round / turn order repair:** Eliminations only occur at round boundaries; the rotation order for the next round maintains the original relative sequence among surviving agents. - **Resource exhaustion:** [gap-fill] Fresh replacement sheepskins are modeled with identical statistical properties upon each successful burst. - **Termination guarantee:** The game terminates in at most Rmax⁡ rounds of N discrete turns, since rounds strictly advance and terminate at r=Rmax⁡. ## State representation - **Persistent state:** `scores: Dict[AgentID, List[float]]`, `active_agents: List[AgentID]`, `current_round: int`, `target_membrane_health: float`. - **Per-turn ephemeral state:** `chosen_parameters: Tuple[float, float, str]`, `sampled_force: float`, `sampled_slip: bool`. - **Observable information:** Fully observable to all agents; all scores, physics parameters, and prior attempts are public knowledge. ## Action space - `jump(velocity: float, angle: float, posture: int)`: Parameters bounded by v∈[0.0,1.0], θ∈[0.0,1.57], posture∈{0,1}. ## Playability notes - **Single-player-vs-AI:** Highly natural as a solo high-score challenge or a 1v1 optimization duel against an AI bot with known physical attributes. - **Group reduction:** A tournament of N=16 gorštaci reduces seamlessly to 2 competing agents (Human vs AI Champion) without loss of mechanics. - **Physical-skill games:** Physical acrobatic control and courage are parameterized through the `power`, `balance`, and chosen (v,θ) tradeoff curve. - **What is lost:** The sheer visceral spectacle of the slippery sheepskin flying out from under a jumper and sending them tumbling into the mud is reduced to a binary slip variable. ## Implementation hints - Very low state complexity; implementation is essentially a multi-objective parameter optimization function with a stochastic boundary. - Obvious trap: Allowing agents to choose infinite velocity without penalizing slip probability proportionally. - Suggested first milestone: Implement the deterministic force/slip equations in a script to verify that high-risk aggressive jumps have high burst probability but catastrophic slip failure rates.

Download as JSON

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

Download JSON
Cite this page

Sheepskin Bellows Jumping (Bosnia & Herzegovina). (n.d.). Games of Countries. https://gamesofcountries.com/games/ba-bosnia-herzegovina/skakanje-na-mijeh

@misc{gamesofcountries-ba-bosnia-herzegovina-skakanje-na-mijeh,
  title = {Sheepskin Bellows Jumping (Bosnia & Herzegovina)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/ba-bosnia-herzegovina/skakanje-na-mijeh}
}