Formal Specification
Ring-passing guessing game — Formal Specification
Armenia · guessing-hidden · Confidence: medium
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 4
# Matani gcotsi — AI-optimized specification
## General idea
A hidden-token deduction parlor game distinct in its asymmetric round resolution: a correct guess triggers a three-way role rotation (recipient → Interrogator, Interrogator → Distributor, old Distributor → Seated), while an incorrect guess changes nothing structurally except a forfeit performance — the same Interrogator and Distributor simply try again. The core information asymmetry is a single hidden variable (which seated player holds the token) that only the Distributor and recipient know.
## Formal model
- **Abstraction taken:** The Interrogator's real skill — reading facial micro-expressions and body language — is abstracted into a single stochastic detection check per round, since the entry describes it qualitatively with no numeric model. The concealment pass itself (entry §2.1-2.2) is treated as instantaneous token placement, since its physical mechanics carry no separate gameplay branching.
- **Player/agent count:** 1 Distributor, 1 Interrogator, M Seated players, M in [3, 18].
- **Turn model:** Sequential rounds. Each round: a hidden token placement, then a single accusation, then resolution.
## Formal rules
R1. Setup: M seated players sit in a line with joined hands (entry §1.1); one agent is Distributor, one is Interrogator (initial assignment arbitrary/agreed before play).
R2. Token placement: the Distributor secretly selects exactly one seated player as `recipient` (entry §2.1-2.2, "at one selected player... lets the ring slide"). This is hidden state, known only to the Distributor and the `recipient` themselves.
R3. Composure: the `recipient` sets `composure_maintained` to the outcome of a stochastic check (entry §2.3 requires pressing palms tight "without flinching, blinking, or shifting posture") [gap-fill: no numeric model is given for how reliably a player can maintain composure; a flat per-player baseline probability is assumed]. `composure_maintained = false` increases the Interrogator's detection probability toward that player in R5 [gap-fill: the entry's description of the Interrogator reading "subtle facial expressions, micro-reactions" implies a link between composure and detectability, though no formula is given].
R4. Call: the Distributor signals `reveal_pass_complete()` (entry §3.1), ending the placement phase.
R5. Accusation: the Interrogator submits `point_accuse(player_id)`, naming exactly one seated player (entry §3.3).
R6. Resolution — correct (entry §4.2): if `player_id = recipient`, the accused reveals their hands (entry §4.1), the Interrogator scores +1 point, and a three-way role rotation occurs: the former `recipient` becomes the new Interrogator, the former Interrogator becomes the new Distributor, and the former Distributor rejoins the seated line [gap-fill: the entry states the first two role changes explicitly but is silent on where the old Distributor goes; returning them to the seated line is the only reading consistent with exactly one Distributor and one Interrogator existing at all times].
R7. Resolution — incorrect (entry §4.3): if `player_id ≠ recipient`, the Interrogator performs a forfeit (song, dance step, or verse — flavor only, no state effect) and remains Interrogator; the Distributor role is unchanged. A new round begins with the same Distributor and Interrogator (R2).
R8. Match end: the entry specifies no fixed score threshold, only "until an agreed score or festive conclusion is reached" (entry §5) [gap-fill: treated honestly as an external parameter — the match ends when the Interrogator's cumulative points reach an externally agreed value N, which the source does not itself specify numerically, so no number is invented here].
## Edge cases
- Tie at the end: not applicable — there is a single running point total for whichever agent currently holds the Interrogator role at any moment; points are not separately tracked per original-player-identity across role rotations in the entry's description, so no tie condition arises from the formal rules as given.
- A player with no legal move: not applicable — the Distributor always has `select_recipient`/`reveal_pass_complete` available; the Interrogator always has exactly one legal `point_accuse` target choice among the seated players.
- Repetition / cycles: possible and not structurally prevented — an Interrogator can in principle guess incorrectly indefinitely (R7 loops back to R2 with no forced state change), so a given round-pair can recur without bound if the Interrogator's detection probability is low. This does not undermine overall termination (R8's argument below) as long as detection probability is non-zero.
- Simultaneous claims on one position: not applicable — exactly one seated player is `recipient` per round (R2); there is no contest over the token.
- Elimination mid-round and turn-order repair: not applicable — no player is ever removed from the game; role rotation (R6) relabels three agents at once without removing anyone.
- Resource exhaustion: not applicable — no consumable resource; the ring/token and seating arrangement are reused every round.
- Termination guarantee: each round has a fixed positive probability of a correct accusation (R5-R6), since the Interrogator's detection check is never zero even at baseline composure. Given an externally-supplied finite point target N (R8), the probability that N has not been reached after t rounds tends to 0 as t → ∞, so the match terminates with probability 1. Absent an agreed N, the entry itself gives the activity no formal terminal state (matching its own "festive conclusion" language) — this is treated honestly rather than inventing a number, exactly as R8 states.
## State representation
- Persistent (session) state: `score[Interrogator_role]` (tracks the current point total for whoever currently holds the Interrogator role, per R6/entry §4.2's scoring), `distributor_id`, `interrogator_id`.
- Per-round hidden state: `recipient` (known only to Distributor and recipient).
- Per-round observable state: `composure_maintained[player]` is not directly observable as a boolean to other agents — it only probabilistically informs the Interrogator's detection check (R3) — everything else (who is currently Distributor/Interrogator, running score) is public.
- Observable information: all players see role assignments and score; only the Distributor and the current `recipient` know the hidden `recipient` value for the active round.
## Action space
- `select_recipient(player_id)`: Distributor only; precondition: placement phase active.
- `reveal_pass_complete()`: Distributor only; precondition: `recipient` has been selected.
- `point_accuse(player_id)`: Interrogator only; precondition: `reveal_pass_complete()` has been called.
## Playability notes
- Single-player-vs-AI: natural as human-Interrogator-vs-AI-Distributor-and-seated-players (a pure deduction/detection challenge), or human-as-recipient managing an AI Interrogator's detection probability.
- Reducing M > 4 to 2 agents: reduces the seated pool to a binary guess, which trivializes the deduction (50/50 baseline) and removes the game's core challenge of scanning many faces.
- Physical-skill games: none beyond the abstracted composure check (R3); the real game's entire challenge — reading micro-expressions — is collapsed into a single stochastic detection parameter.
- What is lost: the actual tactile concealment technique of the hand-passing, the specific psychological tension of maintaining a poker face under group scrutiny, the traditional call-and-response phrase, and the comic performance of the forfeit.
## Implementation hints
State is minimal (one hidden recipient variable per round, role labels, a score); the real design work is calibrating the detection-probability model and its sensitivity to `composure_maintained`, neither of which the source quantifies. First milestone: implement the round loop (R2-R7) with a flat detection probability ignoring composure, verify role rotation on correct guesses, then layer in composure-sensitivity once a baseline is validated.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Ring-passing guessing game (Armenia). (n.d.). Games of Countries. https://gamesofcountries.com/games/am-armenia/matani-gcotsi
@misc{gamesofcountries-am-armenia-matani-gcotsi,
title = {Ring-passing guessing game (Armenia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/am-armenia/matani-gcotsi}
}