Formal Specification
Sheaf Tossing / Sheaf Pitching — Formal Specification
Ireland · strength-endurance · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 2
# Caitheamh na Síge — AI-optimized specification
## General idea
A vertical height elimination tournament where competing throwers use a pitchfork to heave a 14-lb straw sheaf over a progressively raised horizontal bar. Strategic decisions center on energy management, deciding when to take risky maximal-effort attempts versus passing heights to avoid cumulative muscle fatigue before high-altitude rounds.
## Formal model
- **Abstraction taken:** Physical trajectory, biomechanical torque, and sheaf aerodynamics are abstracted into a stochastic height capacity function `H_throw = baseline_height + f(effort, stamina, technique) + ε` where `ε ~ N(0, σ²)`. A throw succeeds if `H_throw ≥ Bar_Height` and clearance angle check passes.
- **Player/agent count:** `N` individual agents (`P_1, ..., P_N`, standardized to 4 agents in engine benchmark).
- **Turn model:** Sequential attempt rounds ordered by bib draw across ascending height stages.
## Formal rules
R1. Initialization: Bar height initialized at `H_0 = 4.5m` (increment `ΔH = 0.3m`). Each agent `i` has stamina `S_i = 100`, strength `STR_i`, technique `TECH_i`, and attempts record.
R2. Round structure at height `H`:
- Active agents with status `Active` are called in order.
- On their turn, an active agent declares either `attempt_toss(effort_level ∈ [0.5..1.0])` or `pass_height()`.
R3. Toss resolution:
- If `attempt_toss(effort)` is chosen:
* Stamina expenditure: `S_i = S_i - (15 * effort)`.
* Achieved throw height: `H_actual = (STR_i * 0.05) + (TECH_i * 0.03) + (effort * 2.5) + N(0, 0.25)`.
* If `H_actual ≥ H`: attempt is `SUCCESS`; agent qualifies for next height stage `H + ΔH`.
* If `H_actual < H`: attempt is `FAIL`. Agent increments misses at current height.
R4. Elimination rule:
- If an agent accumulates 3 consecutive `FAIL` results at height `H`, that agent's status transitions to `Eliminated`.
- Their final score is recorded as the last successfully cleared height `H_cleared_max`.
R5. Bar progression & victory:
- When all agents are either qualified for `H + ΔH` or eliminated:
* If more than 1 agent remains `Active`: `H = H + ΔH`; attempts count at new height resets to 0.
* If exactly 1 agent remains `Active`: that agent is declared the winner.
* If 0 agents qualify at height `H` (all remaining failed 3 attempts): winner is evaluated among those tied at `H_cleared_max` by fewest misses at that height, then fewest total competition misses.
R6. [gap-fill] Jump-off rule: If two or more agents remain tied after all countbacks, a single sudden-death attempt at `H_cleared_max + 0.15m` is executed until the tie is broken.
R7. [gap-fill] Global height cap: If bar reaches `H_max = 12.0m`, remaining active agents share victory.
## Edge cases
- Tie at the end condition: Broken hierarchically by fewest misses at final cleared height, fewest total misses in tournament, then sudden-death jump-off.
- A player with no legal move: Not applicable; an agent always chooses to attempt or pass.
- Repetition / cycles: Bar height increases monotonically; max 3 attempts per height guarantees bounded execution per level.
- Simultaneous claims on the same resource or position: Throws and bar attempts are strictly sequential.
- Player elimination mid-round: When an agent incurs 3 misses, they exit the active queue; turn order of remaining throwers is preserved.
- Resource exhaustion: Stamina is finite (floored at 0, imposing massive throw height penalties); bar heights are bounded by 12.0m cap.
- Termination guarantee: Monotonically increasing bar height coupled with stamina depletion and the 3-miss elimination rule ensures guaranteed termination.
## State representation
- Persistent state: Current bar height `H`, height stage index, active competitor list, global scoreboard `[(agent_id, max_cleared, misses_final, total_misses)]`.
- Per-player state: Attributes (`STR`, `TECH`), current stamina `S_i`, attempts history at current height `[result_1, result_2, result_3]`, status (`Active`, `Qualified`, `Eliminated`).
- Per-turn ephemeral state: Active agent ID, declared effort level, evaluated `H_actual`, attempt outcome.
- Observable information: Complete public scoreboard, bar height, competitor stamina/fatigue cues, and full attempt history.
## Action space
- `attempt_toss(effort: [0.5..1.0])`: Execute physical sheaf pitch with specified muscular effort commitment.
- `pass_height()`: Forfeit remaining attempts at current height to preserve stamina for subsequent rounds.
## Playability notes
- Single-player-vs-AI: Functions cleanly as a tactical stamina-management and risk-optimization simulation.
- Group games (n>4): Naturally scales from 2 to 20 agents via sequential queue.
- Physical-skill games: Muscular power and pitchfork trajectory are captured via stochastic biomechanical models.
- What is lost: The physical smell of sweet barley straw and the tension of a swaying crossbar in crosswinds.
## Implementation hints
Model the competition as a state machine: `Round_Start -> Attempt_Phase -> Qualification_Check -> Height_Advance / Tiebreak -> Finished`.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Sheaf Tossing / Sheaf Pitching (Ireland). (n.d.). Games of Countries. https://gamesofcountries.com/games/ie-ireland/caitheamh-na-sige
@misc{gamesofcountries-ie-ireland-caitheamh-na-sige,
title = {Sheaf Tossing / Sheaf Pitching (Ireland)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/ie-ireland/caitheamh-na-sige}
}