Formal Specification
The Tireli Stick Game — Formal Specification
Jersey · dexterity-throw · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Lé Jeu du Tireli — AI-optimized specification
## General idea
A two-agent tip-cat contest that blends physical strike-timing with psychological distance bidding. The Striker pops a double-pointed wooden peg (*tireli*) into the air and bats it into the field, then bids a number of strides to measure the distance back to base. The Pitcher must decide whether to accept the bid (granting points) or challenge it (risking double points if the Striker successfully proves the distance).
## Formal model
- **Abstraction taken:** Physical peg flipping, batting trajectory, and stride stepping are abstracted into discrete stochastic outcomes parameterized by agent skill and chosen risk level. The bidding and challenge phase is modeled as a complete-information sequential bargaining game.
- **Player/agent count:** 2 agents (Agent 1: `STRIKER`, Agent 2: `PITCHER`).
- **Turn model:** Sequential sub-phases per batting turn: Pitch Strike Pitch-back Tip-strike Bid/Challenge.
## Formal rules
R1. Setup: Home base is located at coordinate . Pitching mark is at . Score target is . Both agents start with score 0.
R2. Phase 1 (Initial Pitch and Hit):
- Pitcher executes `pitch_peg()`.
- Striker selects swing intensity; outcome is `HIT(distance)` with probability or `MISS`.
- If `MISS`, Striker loses one strike attempt (max 3 misses per inning).
- If `HIT`, Pitcher has probability of catching the peg on the fly; if caught, Striker is out (`ROLE_SWAP`).
R3. Phase 2 (Pitch-Back at Base):
- From landing point , Pitcher aims at base target of width .
- If throw hits base (probability ), Striker is out (`ROLE_SWAP`).
R4. Phase 3 (Tip and Drive):
- If pitch-back misses, Striker executes up to 3 `tip_and_strike()` attempts from .
- Each attempt pops peg into air (height ) and hits forward; landing point advances to .
R5. Phase 4 (Bidding and Stride Challenge):
- Striker observes and submits a bid representing estimated stride units ().
- Pitcher observes and , then chooses `ACCEPT` or `CHALLENGE`.
- If `ACCEPT`: Striker gains points. Turn ends, Striker remains batter.
- If `CHALLENGE`: Physical stride verification is sampled:
- If actual strides (Striker's bid was conservative or accurate): Striker gains points.
- If (Striker overbid): Striker gains 0 points and suffers a dismissal (`ROLE_SWAP`) [gap-fill: overbidding on a challenged bid results in role swap].
R6. Win Condition: The first agent to accumulate points immediately wins the match.
## Edge cases
- Tie at the end condition: Not applicable; points are awarded sequentially during the bidding phase. The first agent to cross the threshold wins immediately.
- A player with no legal move: Not applicable; at every phase, the active agent has exactly one valid action category (pitch, swing, tip, bid, or accept/challenge).
- Repetition / cycles (and the loop-breaking rule): [gap-fill] If a Striker fails to score any points across 3 consecutive innings, a mandatory role swap occurs to prevent endless scoreless stalling.
- Simultaneous claims on the same resource or position: Not applicable; all turns and actions are strictly sequential.
- Player elimination mid-round; turn order repair: In 2-agent form, roles swap between `STRIKER` and `PITCHER` upon any dismissal event. In team play, batting order follows a fixed queue.
- Resource exhaustion (pieces, tokens, space, time): Only one *tireli* and one *bâton* exist; they reset position after each resolved inning.
- Termination guarantee: Points are strictly non-negative and monotonically increase upon accepted or won bids. Combined with the 3-inning rotation cap (R6/gap-fill), infinite scoreless play is prevented and an agent provably reaches 50 points.
## State representation
- Persistent state:
- `striker_id`: Index of current Striker (0 or 1).
- `pitcher_id`: Index of current Pitcher (1 or 0).
- `scores`: `[score_0, score_1]`.
- `consecutive_scoreless_turns`: Integer counter.
- Ephemeral state (within turn):
- `phase`: Enum (`INITIAL_PITCH`, `PITCH_BACK`, `TIP_AND_DRIVE`, `BIDDING`, `RESOLVED`).
- `peg_position`: Float .
- `current_bid`: Integer .
- Observable information: Fully observable public state; both agents observe exact peg positions, current scores, and bid declarations.
## Action space
- `pitch_peg()`: Pitcher throws peg to base.
- `strike(power, timing)`: Striker swings at pitched peg.
- `throw_at_base()`: Pitcher throws peg back from field.
- `tip_and_strike(angle, power)`: Striker flips peg up and strikes downfield.
- `declare_bid(b)`: Striker announces integer stride bid .
- `respond_bid(action)`: Pitcher selects `ACCEPT` or `CHALLENGE`.
## Playability notes
- Single-player-vs-AI: Very natural; human as Striker against AI Pitcher. The AI uses expected value calculation: versus .
- Group games (n>4): Easily adapted to 2 teams by rotating team members into the Striker and Pitcher positions while team scores accumulate jointly.
- Physical-skill games: Trajectory is modeled as a function of power and strike accuracy .
- What is lost in the adaptation: The tactile neuromuscular challenge of tapping the sharp conical tip of the wood and hitting a spinning projectile mid-air.
## Implementation hints
- Bidding logic is equivalent to a classic poker-style bluffing and calling mechanism with continuous underlying distance verification.
- Recommended first milestone: Implement the turn state machine with deterministic distances and AI bid evaluation before adding physics distributions.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
The Tireli Stick Game (Jersey). (n.d.). Games of Countries. https://gamesofcountries.com/games/je-jersey/le-tireli
@misc{gamesofcountries-je-jersey-le-tireli,
title = {The Tireli Stick Game (Jersey)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/je-jersey/le-tireli}
}