Formal Specification
Clay Quoits (The Long Game) — Formal Specification
United Kingdom · dexterity-throw · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Clay Quoits — AI-optimized specification
## General idea
A precision throwing game played across ends where two agents alternate pitching heavy cast-iron rings (quoits) into a 2D continuous clay bed targeting a central flush pin (the hob). The game mirrors lawn bowls proximity scoring: points are awarded for every quoit closer to the hob than the opponent's best quoit, with strategic mechanics involving blocking lines and physically knocking opponent quoits out of scoring position.
## Formal model
- **Abstraction taken:** The 18-yard pitch trajectory is projected onto a 2D Euclidean coordinate space representing the 36 in x 36 in square clay bed: `[-18.0, +18.0] x [-18.0, +18.0]` with the hob at origin `(0.0, 0.0)`.
- **Quoit geometry:** Each quoit is modeled as an annular disc with outer radius `R_outer = 4.5 in` and inner radius `R_inner = 2.5 in`.
- **Player/agent count:** 2 agents taking alternating throws.
- **Turn model:** Alternating delivery within an end: Agent 1 throws Quoit 1, Agent 2 throws Quoit 1, Agent 1 throws Quoit 2, Agent 2 throws Quoit 2.
## Formal rules
R1. **Match Structure:** The match is played over a sequence of ends until one agent reaches or exceeds 21 points.
R2. **End Execution:** The leading agent pitches first in the end. Each agent pitches 2 quoits in alternating sequence: `A1 -> B1 -> A2 -> B2`.
R3. **Action Execution:** An agent submits a target coordinate `(x_target, y_target)` and delivery angle / power.
R4. **Impact Sampling:** The quoit's landing coordinate `(x, y)` is sampled from a 2D Gaussian centered on `(x_target, y_target)` with variance inversely proportional to agent precision skill:
- If `(x, y)` lands outside `[-18, 18]^2`, the quoit is marked `OUT_OF_BED` (distance = infinity).
- If `(x, y)` collides with an existing stationary quoit in the clay bed: a 2D elastic collision model updates both quoits' positions with high damping from clay resistance.
R5. **Distance Evaluation:** At the conclusion of all 4 throws, Euclidean distance from the edge of each valid in-bed quoit to the hob `(0, 0)` is computed: `d_i = max(0.0, sqrt(x_i^2 + y_i^2) - R_outer)`.
- If a quoit covers the origin `(0, 0)`, `d_i = 0.0` (Toucher/Gater).
R6. **Scoring:** Let `min_A` and `min_B` be the shortest distances for Agents A and B:
- If `min_A < min_B`: Agent A scores 1 point. If Agent A's second quoit `d_A2 < min_B`, Agent A scores 2 points. Agent B scores 0 points.
- If `min_B < min_A`: Agent B scores 1 point (or 2 points if `d_B2 < min_A`). Agent A scores 0 points.
- [gap-fill] If `min_A == min_B`: 0 points awarded to both; previous end leader pitches first.
R7. **Match Termination:** The first agent whose cumulative score reaches `score >= 21` wins the match.
## Edge cases
- **Tie at the end condition:** If both agents reach >= 21 points with identical totals at the end of a frame, additional ends are played until a leader is established.
- **A player with no legal move:** Not applicable; pitching into the clay bed is always available.
- **Repetition / cycles:** The strictly positive probability of scoring in each end guarantees score progression toward 21.
- **Simultaneous claims on the same resource:** Handled by R5 measuring micro-distance from hob; [gap-fill] true mathematical ties score 0 for that pair of equidistant quoits.
- **Player elimination mid-round:** Not applicable to 1v1 match.
- **Resource exhaustion:** Exactly 2 quoits per player per end; clay bed is reset/smoothed between ends.
- **Termination guarantee:** Guaranteed termination by finite target score (21) and strictly positive expected point gain per end under stochastic delivery.
## State representation
- **Global state:** `scores: [int, int]`, `current_end: int`, `end_leader: int in [0, 1]`, `active_quoits_in_bed: List[QuoitState]`.
- **Per-quoit state:** `owner_id: int`, `center_pos: (float, float)`, `status: Enum(IN_BED, OUT_OF_BED)`.
- **Observable information:** Fully transparent board coordinates and caliper measurements.
## Action space
- `pitch_quoit(x_target: float, y_target: float, spin_spin: float)`: Continuous 2D aim coordinates targeting the clay surface.
## Playability notes
- **Single-player-vs-AI:** Very natural 1v1 tactical match. AI evaluates whether to draw directly to the pin, place a front-guard blocker to obstruct the opponent's line, or strike and dislodge an opponent's scoring ring.
- **Group games reduction:** Standard pairs play maps directly to 1v1 without rule changes (one teammate pitches from each end).
- **Physical-skill games:** Skill is modeled as parameter `sigma` of the 2D Gaussian distribution, capturing pitch consistency under heavy ring physics.
- **What is lost in the adaptation:** Tactile appraisal of clay moisture, lip-edge biting into the mud, and visual reading of surface ruts.
## Implementation hints
Treat the clay bed as a 2D physics surface with heavy damping (restitution coefficient `e ~ 0.15` to represent muddy clay stopping power). Distance sorting and scoring logic matches standard bowls engines.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Clay Quoits (The Long Game) (United Kingdom). (n.d.). Games of Countries. https://gamesofcountries.com/games/gb-united-kingdom/clay-quoits
@misc{gamesofcountries-gb-united-kingdom-clay-quoits,
title = {Clay Quoits (The Long Game) (United Kingdom)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/gb-united-kingdom/clay-quoits}
}