Skip to main content
Formal Specification

Lifting the Boulder (Ditzi Lifting) — Formal Specification

Cyprus · strength-endurance · Confidence: high

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 1
# Διτζιήμιν — AI-optimized specification ## General idea A traditional Cypriot boulder-lifting and distance-carrying strength contest. Agents attempt to hoist a heavy unanchored stone (*διτζί*) from ground level onto their chest or shoulder, then sustain a forward carry across measured discrete paces under an exponential fatigue and grip-decay curve. ## Formal model - **Abstraction taken:** The continuous musculoskeletal biomechanics of lifting and carrying are discretized into a two-phase check: (1) an initial threshold lift check parameterized by agent strength vs. stone mass, and (2) a turn-by-turn pace advance loop where an agent manages a stamina/grip resource with probabilistic slip/drop failure on each step. - **Player/agent count:** 1 to N agents competing sequentially in a leaderboard format. - **Turn model:** Sequential trials. During an agent's trial, ticks represent discrete paces (0.8 metres per pace). ## Formal rules R1. Setup: Stone mass `M` set (default: 80 kg). Each agent `i` possesses static attributes `strength_i` and `endurance_i` in `[0.0, 1.0]`. Attempt counter initialized to 2 per agent. R2. Phase 1 — Hoist Check: - Lift probability: `P(hoist) = clamp(strength_i * 1.5 - (M / 100), 0.0, 1.0)`. - If `hoist == FALSE`: Attempt scores 0.0 metres. - If `hoist == TRUE`: Agent enters Phase 2 with initial grip stamina `G_0 = endurance_i * 100`. R3. Phase 2 — Step Advance Loop: - On step `k` (starting at `k = 1`), agent commits action `take_pace()`. - Stamina depletion: `G_k = G_{k-1} - (M / 10) * (1.0 + 0.05 * k)`. - Drop probability on step `k`: `P(drop_k) = max(0.0, (100 - G_k) / 100)^2`. - If drop occurs or `G_k <= 0`: The carry ends; final distance for the attempt is `(k - 1) * 0.8` metres. - If drop does not occur: `distance += 0.8m`; advance to step `k + 1`. R4. An agent may execute up to 2 attempts. The agent's final score is `max(attempt_1_dist, attempt_2_dist)`. R5. Win condition: The agent with the maximum recorded distance across all competitors wins. R6. [gap-fill] Tie-breaker: If top agents achieve identical distance, the stone mass `M` is increased by 10 kg and a sudden-death single attempt is performed. ## Edge cases - Tie at the end condition: Resolved via R6 by increasing boulder mass by 10 kg in sudden-death. - A player with no legal move: Not applicable; an active agent always chooses between attempting a lift or ending their turn. - Repetition / cycles: Strictly impossible because `G_k` decreases monotonically per step, guaranteeing `P(drop) -> 1.0` within finite steps. - Simultaneous claims on the same resource or position: Not applicable; competitors perform solo trials sequentially. - Player elimination mid-round: If an agent fails their hoist on both attempts, their score is 0.0m; leaderboard ranking handles 0.0m values cleanly without disrupting the turn order. - Resource exhaustion: Not applicable; stone is persistent and resets to starting line after each trial. - Termination guarantee: Monotonic stamina drain guarantees each attempt terminates in finite paces; fixed agent count and attempt limits guarantee total simulation termination. ## State representation - Persistent state: `stone_mass: float`, `leaderboard: dict[agent_id, float]`, `current_agent_idx: int`, `attempt_num: int`. - Per-trial ephemeral state: `current_pace_count: int`, `stamina_pool: float`, `hoist_established: bool`. - Observable information: Fully observable. All agents know the stone mass, opponent scores, and live distance. ## Action space - `attempt_lift()`: Trigger the initial Phase 1 hoist check. - `take_pace()`: Advance 1 pace forward in Phase 2 carry loop. - `drop_stone()`: Voluntarily end the carry to prevent catastrophic failure. ## Playability notes - Single-player-vs-AI: Natural single-player arcade or benchmark challenge (beat AI target distances). - Group games (n>4): Fully scalable; N-agent tournaments run natively as sequential leaderboard trials. - Physical-skill games: Muscular power and stamina modeled as continuous parameters feeding the stamina drain equation. - What is lost: The visceral balance, grip friction, spinal alignment, and breathing technique of actual heavy stone handling. ## Implementation hints Extremely lightweight discrete state machine. State is a single counter and scalar stamina float. Suggested first milestone: implement the Phase 1/Phase 2 equations with fixed random seeds to verify convergence and leaderboard sorting.

Download as JSON

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

Download JSON
Cite this page

Lifting the Boulder (Ditzi Lifting) (Cyprus). (n.d.). Games of Countries. https://gamesofcountries.com/games/cy-cyprus/ditzimin

@misc{gamesofcountries-cy-cyprus-ditzimin,
  title = {Lifting the Boulder (Ditzi Lifting) (Cyprus)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/cy-cyprus/ditzimin}
}