Skip to main content
Formal Specification

Buzzer — Formal Specification

Serbia · guessing-hidden · Confidence: high

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 3
# Zuca — AI-optimized specification ## General idea A asymmetric hidden-action and deduction party game where one Guesser receives a physical signal (a strike of parameterized intensity and acoustic angle) from one of `N-1` hidden Circle Agents, followed by uniform masking noise from all circle agents. The Guesser forms a Bayesian belief distribution over who struck and submits a single guess to transfer the Guesser role. ## Formal model - **Abstraction taken:** The physical palm slap is modeled as a tuple `Signal = (intensity: float, angle_noise: float, latency: float)`. Circle agents possess signature strength ranges and bluffing competencies. - **Player/agent count:** `N` agents (`N >= 3`; default `N = 4`). One designated `Guesser`, others `Circle`. - **Turn model:** Round-based sequential protocol: 1. Circle selection: One circle agent `S in Circle` is chosen (by secret coordination or random turn assignment). 2. Action execution: Agent `S` executes `strike(intensity, angle)`. 3. Observation: Guesser receives noisy observation `O = Signal + Noise`. 4. Deduction & Guess: Guesser selects candidate `G in Circle`. 5. Resolution: If `G == S`, roles swap (`Guesser = S`); otherwise, Guesser incurs 1 penalty point and remains Guesser. ## Formal rules R1. Setup: Agents are indexed `0..N-1`. Role `Guesser` is initialized to Agent 0. Penalty counters `penalties[i] = 0` for all `i`. R2. Striker Selection: In each round, the circle agents choose a single striker `S in Circle` [gap-fill: via round-robin rotation or secret uniform lottery among circle agents]. R3. Signal Generation: Striker `S` selects strike parameter `intensity in [0.1, 1.0]`. The simulator generates observation vector for Guesser: `obs = (intensity + Normal(0, sigma_int), angle_S + Normal(0, sigma_ang))`. R4. Guesser Decision: Guesser submits guess `G in Circle`. R5. Role Transition: - If `G == S`: Success. `penalties[Guesser] += 0`. The old guesser joins the circle; Agent `S` becomes the new `Guesser`. - If `G != S`: Failure. `penalties[Guesser] += 1`. Role remains unchanged. R6. [gap-fill] Game Termination: The game ends after `M` rounds (default `M = 20`) or when any single agent accumulates `P_max = 5` penalties as Guesser. The player with the fewest penalty points wins. ## Edge cases - Tie at the end condition: If multiple players tie on minimal penalty points, the player who spent fewer total rounds in the Guesser role wins. - A player with no legal move: Not applicable; Circle agents can always strike, and Guesser can always select an agent from `Circle`. - Repetition / cycles: Bounded by R6's fixed maximum round count `M`. - Simultaneous claims on the same resource or position: Not applicable; only one agent strikes per round. - Player elimination mid-round / turn order repair: [gap-fill] If an agent disconnects, they are removed from `Circle`; if the disconnected agent was the Guesser, a new Guesser is randomly drawn from the remaining circle. - Resource exhaustion: Not applicable; penalty counters and round counts are unbounded numbers. - Termination guarantee: Guaranteed by the hard round limit `M` and penalty cap `P_max`. ## State representation - Persistent state: `current_guesser: AgentID`, `penalties: dict[AgentID, int]`, `round_count: int`. - Per-round ephemeral state: `chosen_striker: AgentID`, `guesser_observation: tuple[float, float]`. - Observable information: - Circle agents know: who the Guesser is, who the active Striker is, public penalty history. - Guesser knows: only their noisy signal `obs` and past history of player profiles. ## Action space - `select_striker(agent_id: int)`: Circle coordinator action. - `strike(intensity: float)`: Striker action `[0.1, 1.0]`. - `submit_guess(target_agent_id: int)`: Guesser action, choosing from valid Circle agent IDs. ## Playability notes - Single-player-vs-AI: Natural; player as Guesser analyzes AI striker tendencies, or player as Circle agent selects subtle strike forces to fool AI Guesser. - Group games (n>4): Scales smoothly up to 10 agents with increasing deduction difficulty. - Physical-skill games: Physical auditory direction and palm pain sensation are mapped into gaussian noise distributions around angle and force parameters. - What is lost in the adaptation: Visual poker faces, involuntary eye contact, and physical body temperature cues are abstracted into mathematical noise parameters. ## Implementation hints Extremely lightweight Bayesian inference testbed. First milestone: implement state machine with random guessing and uniform signal distributions, then build a basic maximum-likelihood estimator for AI Guesser.

Download as JSON

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

Download JSON
Cite this page

Buzzer (Serbia). (n.d.). Games of Countries. https://gamesofcountries.com/games/rs-serbia/zuca

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