Formal Specification
Pot Smashing (Striking the Earthen Pot) — Formal Specification
Czechia · festival-ritual · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 3
# Tlučení hrnců — AI-optimized specification
## General idea
A sequential spatial search and decision problem under sensory occlusion and noisy auditory feedback. An agent starting from an unknown orientation must estimate the location of a target coordinate using ambiguous "hot/cold" directional feedback and decide when to expend a strictly limited budget of 3 discrete strike attempts to shatter the target.
## Formal model
- **Abstraction taken:** The continuous physical village square and clay pot shattering are modeled on a bounded 2D coordinate grid with the pot located at hidden coordinates . Disorientation is modeled by initializing the agent with an unknown heading angle . Audience shouts are modeled as noisy temperature feedback signals: .
- **Player/agent count:** 1 active Striker agent () and 1 Audience environment generator (or adversarial Misleader agent).
- **Turn model:** Strictly sequential phases: (1) Locomotion step, (2) Sensory observation, (3) Strike decision.
## Formal rules
R1. Setup: A continuous arena of size metres (centered at origin ). Target pot is placed at fixed position with strike radius . Agent starts at distance at a random angle , with initial heading .
R2. In each turn (default paces):
- Agent chooses `step(step_length, turn_angle)` with and .
- State updates: .
R3. [gap-fill] Sensory Feedback Generation: After moving, agent receives a noisy auditory temperature signal :
- Let current distance .
- Signal probability distribution:
- If : .
- If : .
- If : .
R4. Action Execution: At any turn , the agent may execute `strike()`.
- Striking consumes 1 strike token from initial budget .
- [gap-fill] Strike Outcome Resolver: If agent is within distance (with reach = ), hit probability is . If hit succeeds, the pot shatters, and the agent wins immediately.
- If hit fails, .
R5. Win / Loss Conditions:
- **Win:** Pot is shattered before strike budget reaches 0. Score = (incentivizing fewer steps and preserved strikes).
- **Loss:** Strike budget reaches 0, OR pace limit is exhausted without a hit.
R6. [gap-fill] Multi-player Tournament Mode: In competitive play with agents, each agent plays an independent trial; ranking is determined by (1) shattered boolean, (2) fewest strikes used, (3) fewest paces taken.
## Edge cases
- Tie at the end condition: In tournament mode, tied agents with identical strike and step counts share the rank or resolve via a single tie-breaker round.
- A player with no legal move: Not applicable; `step` and `strike` actions are always formally available until termination conditions are reached.
- Repetition / cycles: Strictly prevented by the hard step counter limit and non-replenishable strike budget .
- Simultaneous claims on the same resource or position: Not applicable; trials are executed strictly sequentially one agent at a time.
- Player elimination mid-round: An agent whose budget is reached ends the trial immediately with a loss.
- Resource exhaustion: Strike budget is strictly non-replenishable; step counter is bounded.
- Termination guarantee: Monotonically decreasing strike budget and strictly incrementing turn counter guarantee finite termination in at most steps.
## State representation
- Persistent state: Arena radius, target pot coordinate , strike radius , maximum pace limit , tournament leaderboard.
- Per-step ephemeral state: Agent coordinate , agent orientation , remaining strikes , step index , last received auditory cue .
- Observable information:
- Agent observes: only its own action history, remaining strike count , and sequence of noisy auditory signals . True coordinates and pot position are hidden.
## Action space
- `step(step_length, turn_angle)`: , .
- `strike()`: spend 1 strike token to attempt shattering the pot.
## Playability notes
- Single-player-vs-AI: Directly playable as a POMDP / belief-space planning benchmark. The AI maintains a posterior probability distribution over target position based on Bayesian filtering of acoustic cues.
- Group games (n>4): Naturally formatted as a tournament where sequential single-agent trials populate a common high-score leaderboard.
- Physical-skill games: Physical disorientation and bat trajectory are abstracted into Gaussian angular drift and probabilistic strike resolvers.
- What is lost: The physical sensation of wearing a cloth blindfold, hearing the laughter and acoustic resonance of a live village square, and the physical resistance of wooden stick against clay pottery.
## Implementation hints
Model the agent's internal state as a particle filter representing the estimated position of relative to the agent's current position and heading. Update particle weights upon receiving each discrete cue according to the observation likelihood model. Strike should be triggered only when the integrated probability mass within the strike radius exceeds a calculated threshold .
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Pot Smashing (Striking the Earthen Pot) (Czechia). (n.d.). Games of Countries. https://gamesofcountries.com/games/cz-czechia/tluceni-hrncu
@misc{gamesofcountries-cz-czechia-tluceni-hrncu,
title = {Pot Smashing (Striking the Earthen Pot) (Czechia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/cz-czechia/tluceni-hrncu}
}