Skip to main content
Formal Specification

Pigeon Flies (Bird Flies) — Formal Specification

Guernsey · dexterity-small · Confidence: medium

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 2
# Pigeon Vole — AI-optimized specification ## General idea A fast-paced stimulus-response and inhibition control game. A Caller agent announces prompt pairs (Entity,Fly!) at rapid intervals while consistently executing a deceptive physical action (raising finger). Respondent agents must inhibit visual mimicry and select actions based solely on the true semantic flight property of the named entity: raising their finger if and only if the entity is capable of flight. ## Formal model - **Abstraction taken:** Semantic knowledge is formalized as a public ontology dictionary 𝒟:Entity→{CAN_FLY,CANNOT_FLY}. Reaction timing and cognitive inhibition are modeled as discrete decision windows Δt where agent error probability is influenced by the visual distractor. - **Player/agent count:** 1 `CALLER` agent, and N≥2 `RESPONDENT` agents (indexed 1…N). - **Turn model:** Synchronous round ticks. The Caller broadcasts a prompt; all active Respondents submit binary actions within a fixed time window. ## Formal rules R1. Knowledge Base and Entities: - Finite set of entities ℰ=ℰfly∪ℰnonfly, where ℰfly∩ℰnonfly=∅. - Ground truth: V(e)=1 if e∈ℰfly, and V(e)=0 if e∈ℰnonfly. R2. Round Loop: - At tick t, Caller samples an entity et∼𝒰(ℰ) and broadcasts the event (entity=et,caller_action=RAISE). R3. Respondent Action: - Each active Respondent i declares action ai,t∈{RAISE,STILL} within time window Twindow=500ms. R4. Error Evaluation and Elimination: - Correct action: at∗=RAISE if V(et)==1, and at∗=STILL if V(et)==0. - For each active Respondent i: - If ai,t≠at∗ or response time >Twindow: Respondent i commits an error. - In elimination mode: Respondent i is eliminated (`OUT`). R5. Win Condition: - The round sequence continues until exactly 1 active Respondent remains, who is declared the winner (`SURVIVOR`). - [gap-fill] Tie-break on mutual elimination: If all remaining active Respondents commit an error on the exact same tick, all those players are reinstated for a sudden-death round. R6. [gap-fill] Max Round Horizon: If >1 Respondent survives after M=30 prompts, the winner is the agent with the lowest average reaction latency across all correct responses. ## Edge cases - Tie at the end condition: Resolved via R5/R6 (reinstatement or lowest average latency). - A player with no legal move: Not applicable; Respondents always have exactly two binary choices (RAISE or STILL). - Repetition / cycles (and the loop-breaking rule): Bounded by the prompt horizon M=30. - Simultaneous claims on the same resource or position: Not applicable; all Respondent actions are evaluated independently against ground truth V(et). - Player elimination mid-round; turn order repair: Eliminated Respondents transition to `OUT` and submit no further actions. - Resource exhaustion (pieces, tokens, space, time): Entity dictionary ℰ is finite; ticks are bounded by M≤30. - Termination guarantee: Because each prompt has non-zero error rate under cognitive speed pressure, and the prompt sequence is hard-capped at M=30, termination is strictly guaranteed. ## State representation - Persistent state: - `tick`: Integer counter ∈[1,30]. - `active_players`: Set of active Respondent IDs ⊆{1,…,N}. - `reaction_latencies`: Matrix storing response times for active agents. - Ephemeral state (per prompt): - `current_entity`: String / token et∈ℰ. - `ground_truth`: Binary V(et)∈{0,1}. - `declared_responses`: Map of agent ID →(ai,t,τi,t). - Observable information: Fully observable public game state; current entity prompt and player statuses are public. ## Action space - Caller: `broadcast_prompt(entity)` where entity∈ℰ. - Respondent i: `submit_response(action, latency)` where action∈{RAISE,STILL} and latency∈[0,Twindow]. ## Playability notes - Single-player-vs-AI: Very engaging reaction-time mini-game; human vs AI with simulated Stroop-effect cognitive delay. - Group games (n>4): Scales naturally to any N∈[2,20] without rules adaptation. - Physical-skill games: Cognitive reaction speed and inhibitory control are parameterized via an agent reaction distribution: τ∼LogNormal(μ,σ2) and an inhibition failure probability on non-flying entities: P(mimicry_error). - What is lost in the adaptation: Visual peripheral distraction, live table-tapping cadence, and comedic social pressure. ## Implementation hints - Simple and lightweight state machine: array of boolean flight mappings 𝒟. - Can easily be integrated as a fast mini-game within a broader Channel Island folk collection suite. - First milestone: Implement prompt broadcaster and deterministic correctness checker before adding latency metrics.

Download as JSON

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

Download JSON
Cite this page

Pigeon Flies (Bird Flies) (Guernsey). (n.d.). Games of Countries. https://gamesofcountries.com/games/gg-guernsey/pigeon-vole

@misc{gamesofcountries-gg-guernsey-pigeon-vole,
  title = {Pigeon Flies (Bird Flies) (Guernsey)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/gg-guernsey/pigeon-vole}
}