Skip to main content
Formal Specification

Cooper's Hill Cheese-Rolling and Wake — Formal Specification

United Kingdom · festival-ritual · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 3
# Cooper's Hill Cheese-Rolling — AI-optimized specification ## General idea A simultaneous-start extreme downhill race down a steep 200-yard natural incline where agents balance sprint velocity against catastrophic fall risk. Agents select locomotion modes (sprint, controlled slide, roll) across discrete slope zones, managing momentum and stability until crossing the finish line. ## Formal model - **Abstraction taken:** The 180-meter slope is discretized into 18 sequential 10-meter distance segments `d in [0..18]`. Continuous gravitational acceleration, slope pitch (increasing from 20 to 45 degrees), and turf irregularities are converted into a discrete-time tick engine (1 tick = 0.5 seconds). At each tick, an agent's chosen velocity interacts with a stability threshold. - **Player/agent count:** 15 to 40 agents per race. - **Turn model:** Simultaneous-declaration ticks; all agents resolve movement concurrently. ## Formal rules R1. **Setup:** All agents start at `distance = 0.0`, `velocity = 0.0`, `stability = 100.0`, and status `UPRIGHT`. R2. **Start Signal:** At `tick = 0`, the cheese is launched with fixed descent profile `Cheese_Pos(t) = 0.5 * 9.8 * sin(theta) * t^2`. At `tick = 2` (1-second delay), all agents begin downhill descent. R3. **Action Selection per Tick:** Each active agent selects an action mode: - `sprint(target_speed: float in [4.0, 10.0])`: Maximum forward speed; high risk of tripping. - `brace_slide(target_speed: float in [2.0, 5.0])`: Moderate speed; low risk of loss of control. - `tumble_recover()`: Available when `status == TUMBLING`; attempts to regain `UPRIGHT` stance. R4. **Stability and Fall Resolver:** - Slope gradient increases with distance: `gradient(d) = 0.3 + 0.03 * d`. - Fall probability per tick: `P(fall) = (target_speed / 10.0)^2 * gradient(d) * (1.0 - stability / 100.0)`. - If fall occurs: agent status becomes `TUMBLING`, `velocity` drops by 40%, and `stability` drops by 20. - While `TUMBLING`, the agent continues sliding downhill driven by gravity (`velocity_slide = 4.0 * gradient(d)`). R5. **End Condition & Victory:** - The race terminates as soon as any agent reaches `distance >= 180.0` meters. - The first agent to cross `180.0` meters is declared the Winner. - [gap-fill] Simultaneous crossing: If multiple agents cross in the exact same tick, the agent with the highest excess distance `(distance - 180.0)` wins. ## Edge cases - **Tie at the end condition:** Resolved by R5 (comparing fractional distance delta beyond the 180m line). - **A player with no legal move:** Not applicable; movement or tumbling physics automatically update every tick. - **Repetition / cycles:** Not applicable; gravity strictly drives distance monotonically forward. - **Simultaneous claims on the same resource:** Terrain cells allow shared occupancy; [gap-fill] collision risk between nearby agents is modeled as a small stability penalty (-5 stability if within 1m of a tumbling neighbor). - **Player elimination mid-round:** If an agent's stability reaches 0 (injury incapacitation), [gap-fill] that agent stops and receives DNF status. - **Resource exhaustion:** Non-depletable track; distance is bounded at 180m. - **Termination guarantee:** Guaranteed by strictly positive forward velocity driven by gravity and sprint actions down the slope. ## State representation - **Persistent state:** `race_clock: int`, `leaderboard: list of agent_id`, `race_finished: bool`. - **Per-agent state:** `distance: float in [0.0, 200.0]`, `velocity: float`, `stability: float in [0.0, 100.0]`, `status: enum [UPRIGHT, TUMBLING, INCAPACITATED, FINISHED]`. - **Observable information:** Common knowledge of all competitors' distances, status, and cheese position. ## Action space - `sprint(speed)`: Aggressive forward sprint. - `brace_slide(speed)`: Controlled descent. - `tumble_recover()`: Recovery action during tumbling. ## Playability notes - **Single-player-vs-AI:** Highly engaging push-your-luck racing game where the player balances risky high speeds against fatal tumbling cascades. - **Group games reduction (2-agent form):** Can be run as a pure 1v1 head-to-head downhill duel between two runners without altering race dynamics. - **Physical-skill games:** Physical balance, turf reading, and impact tolerance are mapped to the speed-versus-stability dynamic curve. - **What is lost in the adaptation:** The physical adrenaline, raw mud/grass texture, and chaotic physical tumbling sensation. ## Implementation hints - Low computational complexity; runs 40 simultaneous agents in real-time with sub-millisecond tick evaluations. - First milestone: Build a 1D tick simulation of a single agent testing sprint vs slide strategies over 1,000 runs to find the Pareto optimal speed curve.

Download as JSON

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

Download JSON
Cite this page

Cooper's Hill Cheese-Rolling and Wake (United Kingdom). (n.d.). Games of Countries. https://gamesofcountries.com/games/gb-united-kingdom/cheese-rolling

@misc{gamesofcountries-gb-united-kingdom-cheese-rolling,
  title = {Cooper's Hill Cheese-Rolling and Wake (United Kingdom)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/gb-united-kingdom/cheese-rolling}
}