Skip to main content
Formal Specification

The Cicada (Buzzing Slap Game) — Formal Specification

Cyprus · guessing-hidden · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: directGap-fills: 3
# Ζίζυρος — AI-optimized specification ## General idea A asymmetric hidden-identity guessing game where one defender with back turned receives a physical slap on an outstretched rear palm from an unknown circle agent. All circle agents simultaneously present identical noisy distractor signals (raised finger and buzzing). The defender inspects imperfect observational cues (latency, eye gaze, bluffing signals) to accuse the striker, swapping roles on a correct guess. ## Formal model - **Abstraction taken:** The physical sound of the slap and collective cicada buzzing is modeled as an imperfect information signaling round. The striker generates a subtle tell vector `T_striker`, which is mixed with noise from innocent agents' bluffing distributions. - **Player/agent count:** N agents (3 ≤ N ≤ 15). 1 designated Defender (D); N-1 Circle Agents (C_1...C_{N-1}). - **Turn model:** Sequential rounds. Within each round: (1) Circle selection phase, (2) Simultaneous reveal/buzzing signal phase, (3) Defender accusation action. ## Formal rules R1. Setup: Agent index 0 is assigned role `DEFENDER`. All other agents assigned role `CIRCLE`. Round counter `r = 1`. R2. Striker Selection: - Circle agents execute a coordination lottery or sequential silent bid to choose exactly 1 active `STRIKER` `s in {1..N-1}`. - [gap-fill] In autonomous multi-agent play, `s = random_choice(CIRCLE_AGENTS)`. R3. Signal Generation: - The striker generates a signal vector: `S_s = strike_impact + N(0, sigma_s)`. - Every circle agent `i` produces an observable tell: `Tell_i = (i == s) ? (base_tell + delta_striker) : (base_tell + delta_bluff)`. - All circle agents broadcast the uniform state `BUZZING = TRUE`. R4. Defender Observation & Accusation: - Defender observes private state `{Tell_1, ..., Tell_{N-1}}` with perceptual noise `N(0, sigma_D)`. - Defender selects action `accuse(target_id)`. R5. Resolution: - If `target_id == s`: Accusation succeeds. Agent `s` becomes `DEFENDER`; previous defender becomes `CIRCLE`. Round ends. - If `target_id != s`: Accusation fails. Defender remains `DEFENDER`. Round ends. R6. [gap-fill] Termination / Match Scoring: Game is played across a fixed match budget of `M` rounds (default: `M = 10 * N`). Each agent tracks `turns_as_defender`. The agent with the minimum rounds spent as defender wins the match. ## Edge cases - Tie at the end condition: If multiple agents tie for lowest `turns_as_defender`, the agent with the highest success rate as defender (successful accusations / total rounds as defender) breaks the tie. - A player with no legal move: Not applicable; Circle agents always have legal striker nomination, and Defender always has an `accuse(target_id)` action. - Repetition / cycles: Role swapping is stochastic and bounded by the match round limit `M`, guaranteeing termination. - Simultaneous claims on the same resource or position: [gap-fill] Exactly one striker is selected per round via discrete lottery. - Player elimination mid-round: If an agent disconnects/leaves, remaining agents adjust the circle roster; if the departing agent was Defender, a random circle agent assumes Defender. - Resource exhaustion: Not applicable; no physical consumables exist. - Termination guarantee: Enforced by the fixed round limit `M`. ## State representation - Persistent state: `match_round: int`, `max_rounds: int`, `current_defender_id: int`, `cumulative_defender_rounds: dict[agent_id, int]`, `successful_guesses: dict[agent_id, int]`. - Per-round ephemeral state: `striker_id: int` (hidden from Defender), `observed_tells: dict[agent_id, float]` (publicly available to Defender). - Minimal observable information: - Common knowledge: Current defender ID, match round, cumulative scores, circle agent roster. - Hidden information: `striker_id` is hidden from Defender until after `accuse()` resolves. ## Action space - Circle Agent: `nominate_strike()` (participate in striker selection), `emit_bluff(level: float)`. - Defender Agent: `accuse(target_id: int)` with precondition `target_id in CIRCLE_AGENTS`. ## Playability notes - Single-player-vs-AI: Highly playable; a human player acts as Defender inspecting simulated facial/motion cues from AI circle agents, or plays as a circle agent attempting to time the slap and minimize visual tell. - Group games (n>4): N-agent game scales seamlessly without structural modification. - Physical-skill games: Deceptive body language and deduction mapped to a continuous Bayesian inference problem over noisy scalar features. - What is lost: The visceral auditory immersion of an entire crowd loudly buzzing inches from the player's face. ## Implementation hints Very low state footprint. Implement as a standard matrix bluffing game where Circle agents optimize a bluff scalar `b in [0, 1]` and Defender trains a classifier or Bayesian estimator to predict `P(striker == i | Tells)`.

Download as JSON

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

Download JSON
Cite this page

The Cicada (Buzzing Slap Game) (Cyprus). (n.d.). Games of Countries. https://gamesofcountries.com/games/cy-cyprus/zizyros

@misc{gamesofcountries-cy-cyprus-zizyros,
  title = {The Cicada (Buzzing Slap Game) (Cyprus)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/cy-cyprus/zizyros}
}