Skip to main content
Formal Specification

Swallows Pass Through — Formal Specification

Serbia · singing-circle · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 2
# Laste prolaste — AI-optimized specification ## General idea A two-stage party and athletic game. In Stage 1, marching agents are sequentially captured at a rhythmic singing gate and make discrete secret choices between two hidden team identities (*Apple* vs *Pear*). In Stage 2, the resulting two partitioned teams execute a simultaneous isometric tug-of-war to decide the match winner. ## Formal model - **Abstraction taken:** The rhythmic singing march is modeled as a cyclic queue of marching agents progressing 1 step per beat, with a deterministic trap trigger on beat B=8 ("Zatvaraj!"). Stage 2 is a 1D force-displacement tug-of-war between the two dynamically assembled teams. - **Player/agent count:** N agents (N≥6, default N=8). - **Turn model:** Stage 1: sequential capture and choice steps until the queue is empty; Stage 2: synchronous isometric tug-of-war ticks. ## Formal rules R1. Setup: Agents 0 and 1 are Gatekeepers. Agents 2..N−1 form the marching queue Q. Gatekeepers secretly assign identities: G0=Team_Apple, G1=Team_Pear. R2. Stage 1 (Capture & Team Assignment): - On each round of B=8 beats, queue Q advances. - On beat 8, the agent currently under the gate, Atrapped=Q.pop(0), is captured. - Atrapped secretly selects choice C∈{APPLE,PEAR}. - If C==APPLE, agent joins `Team_0`; if C==PEAR, agent joins `Team_1`. - Repeats until queue Q is empty. R3. Stage 2 (Tug-of-War Duel): - Central position x=0.0. - Team force calculation: F0=∑a∈Team_0strength(a), F1=∑b∈Team_1strength(b). - Per-tick dynamic displacement: Δx=(F1−F0)×kpull+Normal(0,σtug). - Update: x←x+Δx. R4. Win Condition: - If x≤−1.0: Team 0 (Apple) wins the match. - If x≥+1.0: Team 1 (Pear) wins the match. R5. [gap-fill] Stage 2 Timeout: If |x|<1.0 after 50 ticks, the team with x<0 wins for Team 0, and x>0 wins for Team 1. ## Edge cases - Tie at the end condition: If x==0.0 exactly at tick 50, a 10-tick sudden-death burst determines the victor. - A player with no legal move: Not applicable; marching agents always select from `{APPLE, PEAR}`; tug agents automatically pull. - Repetition / cycles: Bounded; Stage 1 completes in exactly N−2 capture cycles; Stage 2 terminates in at most 50 ticks. - Simultaneous claims on the same resource or position: Not applicable; queue pop is deterministic on beat 8. - Player elimination mid-round / turn order repair: [gap-fill] If an agent disconnects during Stage 1, they are removed from queue Q; if in Stage 2, their strength contribution drops to 0. - Resource exhaustion: Queue Q depletes monotonically to zero. - Termination guarantee: Guaranteed by finite queue N−2 in Stage 1 and R5 tick limit in Stage 2. ## State representation - Persistent state: `team_assignments: dict[AgentID, int]`, `gatekeepers: tuple[AgentID, AgentID]`, `phase: enum[STAGE_1_MARCH, STAGE_2_TUG]`. - Stage 1 state: `marching_queue: list[AgentID]`, `beat: int`. - Stage 2 state: `displacement_x: float`, `stamina: dict[AgentID, float]`, `tick: int`. - Observable information: - Public: marching queue order, public team rosters in Stage 2, displacement x. - Private: Gatekeeper true mapping (G0→Apple, G1→Pear) is hidden from marchers until selection is submitted. ## Action space - `choose_side(choice: enum[APPLE, PEAR])`: Stage 1 action for captured marchers. - `pull_tug(effort: float)`: Stage 2 action during team pull. ## Playability notes - Single-player-vs-AI: Playable as a team management/tug challenge where player tries to deduce AI gatekeeper preferences and lead their squad in the tug contest. - Group games (n>4): Benchmark multi-agent social partition and coalition game. - Physical-skill games: Isometric grip and mass are mapped to linear strength summations. - What is lost in the adaptation: The melody and rhythmic joy of children singing in schoolyard sunshine are abstracted to discrete beat counters. ## Implementation hints Two distinct sub-engines (queue partition + 1D physics pull). First milestone: implement Stage 1 queue pop and verify team split sizes, then pipe rosters into the Stage 2 tug resolver.

Download as JSON

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

Download JSON
Cite this page

Swallows Pass Through (Serbia). (n.d.). Games of Countries. https://gamesofcountries.com/games/rs-serbia/laste-prolaste

@misc{gamesofcountries-rs-serbia-laste-prolaste,
  title = {Swallows Pass Through (Serbia)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/rs-serbia/laste-prolaste}
}