Skip to main content
Formal Specification

Pūheʻenehene (stone-hiding guessing game) — Formal Specification

Usa · guessing-hidden · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 4
# Pūheʻenehene — AI-optimized specification ## General idea A concealment-guessing game combining a physical tell (the hider's arm-muscle tension) with a strategic guessing structure across five hiding spots. Interesting for an AI implementation because the "tell" mechanic is unformalizable in a digital version and must be replaced with a probability model. ## Formal model - **Abstraction taken:** the muscle-tell mechanic (watching the hider's arm) cannot be reproduced in software; this spec replaces it with a hidden random variable (the true bundle) and treats the guessing side's choice as a probability/strategy decision rather than a skill-based observation, `fidelity_class: probabilistic-proxy`. - **Player/agent count:** 2 teams, 5 players each in the primary form (Ellis); a 6-player, 3-bundle variant is separately attested (Corney) and may be selected instead. - **Turn model:** sequential within a round (one hide, then sequential guesses) with round-robin alternation of the hiding role between teams. ## Formal rules R1. Setup: 5 (or 3, in the Corney variant) cloth bundles are laid in a row; one stone exists as the hidden token. R2. The hiding team's chosen player secretly assigns the stone to one bundle (uniform random choice is the natural digital analogue of concealment, [gap-fill] since the human hider's actual choice process is not attested as random vs. deliberate). R3. The guessing team's players strike bundles one at a time, in a pre-agreed order, until the stone is found or all bundles are exhausted. R4. A correct strike on the first attempt scores the guessing team a win "with one stroke"; correctness on a later attempt still wins the round but is recorded as requiring more strokes (attested: rounds are compared "with fewest strokes"). R5. [gap-fill] Match length: first team to N correct finds (N not attested; suggest N=5 or N=7 as a reasonable digital default) wins the match. R6. Reverse-mode variant (attested, R8 in the entry): the guessing team instead tries to strike as many *wrong* bundles as possible without hitting the stone within an agreed number of tries — [gap-fill] exact scoring threshold not attested. ## Edge cases - Tie at end: not attested; [gap-fill] if both teams reach N finds in the same number of overall rounds, treat as a tie requiring a tiebreak round. - A player with no legal move: not applicable — striking any un-struck bundle is always legal until all are exhausted. - Repetition/cycles: not possible — each round terminates in at most (number of bundles) strikes. - Simultaneous claims: not applicable, strikes are sequential within a round. - Player elimination: not applicable. - Resource exhaustion: a round ends when the stone is found (at most 5, or 3 in the variant, strikes). - Termination guarantee: each round is bounded by the fixed bundle count, and R5's match target N is finite, so the game provably terminates. ## State representation Persistent state: match score (finds per team). Per-round state: which bundle secretly holds the stone (hidden from the guessing team), which bundles have been struck this round. Per-player: none. Information asymmetry is the core mechanic — the guessing team does not observe the stone's location; the hiding team does. ## Action space - `hide(bundle_index)`: hiding team's private action, sets the secret location. - `strike(bundle_index)`: guessing team's public action; reveals whether that bundle held the stone. ## Playability notes - Single-player-vs-AI: adaptation required — a solo player can play the guessing role against an AI hider (uniform-random or with a simple bluffing model), or the hiding role against an AI guesser that tracks strike history. - Group games (n>4): reduces cleanly to 1-vs-1 (one hider, one guesser) since team structure only affects who takes the physical action, not the game logic. - Physical-skill games: the historical version depends on reading real muscle tension, which is explicitly lost in any digital adaptation — flagged here as the single largest fidelity loss in this entire source. - What is lost: the entire "tell-reading" skill that Ellis describes as the game's central drama is unimplementable; a digital version is a pure guessing game with none of the original's psychological tension. ## Implementation hints Trivial state space (choose 1 of 5). The interesting design question is whether to give an AI hider a bluffing/pattern-avoidance model (to avoid being purely uniform-random and therefore trivially exploitable) — recommend a simple frequency-tracking counter-strategy as the first milestone before anything more elaborate.

Download as JSON

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

Download JSON
Cite this page

Pūheʻenehene (stone-hiding guessing game) (Usa). (n.d.). Games of Countries. https://gamesofcountries.com/games/us-usa/puhenehene

@misc{gamesofcountries-us-usa-puhenehene,
  title = {Pūheʻenehene (stone-hiding guessing game) (Usa)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/us-usa/puhenehene}
}