Skip to main content
Formal Specification

The Big Wind Blows (Hong Kong Musical Chairs Tag) — Formal Specification

Hong Kong · chase-tag · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: discretizedGap-fills: 4
# Daai Fung Ceoi — AI-optimized specification ## General idea A simultaneous-scramble spatial assignment and observation game based on Hong Kong musical chairs tag (*Daai Fung Ceoi*). A central Caller announces an observational trait predicate; all matching seated agents must abandon their chairs and sprint across the circular arena to claim a non-adjacent seat while the Caller simultaneously attempts to seize a chair, stranding the slowest agent as the next caller. ## Formal model - **Abstraction taken:** Physical chair layout is formalized as a cyclic graph CM with M=N−1 chairs arranged in a ring. Physical appearance traits are modeled as a discrete binary attribute vector T→i∈{0,1}K per agent. Scramble speed is modeled as ring-distance transit delay. - **Player/agent count:** N players (6≤N≤12), consisting of 1 Caller and M=N−1 Seated agents. - **Turn model:** Two-phase rounds: (1) Caller selects and broadcasts a trait predicate, (2) all affected agents and the Caller submit target chair claims simultaneously. ## Formal rules R1. **Setup & Attributes:** M=N−1 chairs are arranged on a ring indexed 0,1,…,M−1. Each agent i∈{0,…,N−1} possesses a public binary trait vector T→i=(ti,0,ti,1,…,ti,K−1). Caller starts in center state Chair=None. R2. **Callout Phase:** The active Caller submits a command (mode,k) where: - *Big Wind* (`BIG_WIND`, k): Targets all agents with ti,k=1. - *Gentle Wind* (`GENTLE_WIND`, k): Targets all agents with ti,k=0. - *Hurricane* (`HURRICANE`, None): Targets ALL seated agents. R3. **Legal Move Constraints:** Every agent targeted by the callout MUST vacate their chair and submit a target destination chair c∈{0,…,M−1}. - **Adjacent Seat Ban:** A seated agent at chair s cannot choose c if min⁡(|c−s|,M−|c−s|)≤1. - Untargeted seated agents must submit STAY (remain in chair s). - The Caller may choose ANY vacant or vacated chair index c∈{0,…,M−1}. R4. **Simultaneous Resolution:** For each contested chair c, if multiple agents target c: - The agent with the shortest transit distance d(i,c) claims the seat. Distance from center for Caller is parameterized as dcenter=1.5. - All other competing agents are deflected to their secondary preference or left unseated. R5. **Succession & Penalty:** The single agent remaining without a chair at the end of resolution receives +1 penalty point and becomes the Caller for the subsequent round. R6. **Victory:** In points mode, play proceeds for a fixed number of rounds (R=10). The player with the fewest penalty points wins. R7. [gap-fill] Trait universe: K=6 binary observable features randomly distributed among agents at initial setup with probability P(ti,k=1)=0.5. R8. [gap-fill] Simultaneous conflict resolution: If two or more agents target the same chair c with identical transit distances, the claim is awarded to the agent with the lowest agent ID. R9. [gap-fill] Non-compliance penalty: If an agent targeted by the wind fails to submit a legal non-adjacent seat action, they automatically forfeit their chair and become the new Caller. R10. [gap-fill] Tournament termination limit: Match runs for exactly 10 rounds (Rmax⁡=10); ties in minimum penalty points are broken by fewest total rounds spent as Caller. ## Edge cases - **Tie at the end condition:** Deterministically resolved by R10 (fewest total rounds as Caller, then lowest initial agent ID). - **A player with no legal move:** When a callout triggers a small subset (e.g. 2 agents), valid non-adjacent empty seats are mathematically guaranteed since M≥5. - **Repetition / cycles:** Fixed round horizon Rmax⁡=10 prevents infinite cycling. - **Simultaneous claims on the same resource or position:** Explicitly resolved by R4 and R8 (distance metric followed by deterministic agent ID order). - **Player elimination mid-round:** In elimination tournament mode, the stranded player is removed, the chair count is reduced (M←M−1), and chair indices are re-mapped consecutively. - **Resource exhaustion:** Chair count is fixed at M=N−1; trait vector length is fixed at K=6. - **Termination guarantee:** Guaranteed by strict progression of round counter r∈[1,10] to terminal evaluation. ## State representation - **Persistent state:** `num_players: N`, `chair_count: M = N - 1`, `trait_vectors: List[List[int]]`, `round_limit: 10`. - **Per-player state:** - `current_chair: Optional[int]` (None if currently Caller) - `penalty_points: int` - `rounds_as_caller: int` - **Per-turn ephemeral state:** - `active_caller_id: int` - `current_callout: Tuple[str, Optional[int]]` - `vacated_chairs: Set[int]` - `pending_seat_claims: Dict[agent_id, List[int]]` (ordered preference list) - **Observable information:** Fully observable game state. All agent positions, traits, callout commands, and penalty scores are public. ## Action space - `call_wind(mode, trait_idx)`: Parameters: `mode in {BIG_WIND, GENTLE_WIND, HURRICANE}`, `trait_idx in [0, K-1]`. Preconditions: agent is active Caller. - `claim_seat(target_chair_preferences)`: Parameters: `List[int]` of target chair indices. Preconditions: agent is targeted by wind or is Caller; target does not violate Adjacent Seat Ban. - `stay()`: Parameters: none. Preconditions: agent is not targeted by current wind. ## Playability notes - **Single-player-vs-AI:** Excellent as an observation and social deduction challenge. Human agent can strategically select traits that isolate specific bots while targeting optimal cross-circle seats. - **Group games (n>4):** Natively accommodates 6 to 12 agents. - **Physical-skill games:** Physical sprinting speed and body positioning are abstracted into graph transit distances d(s,c) and deterministic priority resolution. - **What is lost in the adaptation:** Physical pushing, seat cushion squeezing, hilarious false-start hesitations, and vocal chanting dynamics are formalized as discrete matrix assignments. ## Implementation hints - Model seat selection as a Stable Marriage / Deferred Acceptance preference matching algorithm over chairs. - State is completely discrete: ring topology is trivial to represent as modulo arithmetic. - Branching factor: Caller has 2K+1 choices (13 actions for K=6). Evaders have M−3 valid chairs to target. - First milestone: Headless 6-agent simulation verifying adjacent seat ban enforcement and tiebreaker assignment over 10 test rounds.

Download as JSON

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

Download JSON
Cite this page

The Big Wind Blows (Hong Kong Musical Chairs Tag) (Hong Kong). (n.d.). Games of Countries. https://gamesofcountries.com/games/hk-hong-kong/daai-fung-ceoi

@misc{gamesofcountries-hk-hong-kong-daai-fung-ceoi,
  title = {The Big Wind Blows (Hong Kong Musical Chairs Tag) (Hong Kong)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/hk-hong-kong/daai-fung-ceoi}
}