Formal Specification
Nyengeli-nyengeli mune (fruit-stone memory game) — Formal Specification
South Africa · guessing-hidden · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: directGap-fills: 3
# Nyengeli-nyengeli mune — AI-optimized specification
## General idea
A memory-tracking game: fruit stones are silently redistributed between groups over several rounds while one player's back is turned, and that player must later recall the exact count in each group from having tracked the moves by ear alone (not sight, since their back is turned).
## Formal model
- **Abstraction taken:** none major — the core information-tracking mechanic transfers cleanly to a digital memory-game format.
- **Player/agent count:** 1 guesser plus 1 or more questioners/movers (questioner role can be merged into a single AI or human opponent).
- **Turn model:** sequential — one move-and-announce cycle per group, repeated, followed by a final recall challenge.
## Formal rules
R1. Setup: N fruit stones are arranged into some number of starting groups of 2 each.
R2. The guesser's back is turned (in a digital adaptation: the guesser's view of the board state is hidden) for the remainder of the round.
R3. For each group in turn, the questioner announces the group being referenced ("how many stones are there?"); one stone is moved from that group to another group (the guesser is told a stone was moved, but not, in the true traditional form, shown the board — this is the core hidden-information mechanic).
R4. If a group reaches zero stones, the guesser is informed ("there are no more") — this is the one piece of ground-truth feedback the guesser receives during play.
R5. [gap-fill] This process (R3-R4) repeats for an agreed number of rounds or until the questioner decides to end the setup phase — no fixed round count is attested.
R6. Recall phase: the questioner asks, for each group in turn (in any order, "passes rapidly from group to group"), how many stones it currently holds; the guesser must answer for each.
R7. [gap-fill] Scoring: this spec awards the guesser one point per group correctly recalled, since no attested win/forfeit condition survives in the source.
## Edge cases
- Tie at end: not applicable under the R7 gap-fill point-per-group scoring in a solo recall exercise; only relevant if adapted into a competitive 2-guesser format, in which case [gap-fill] equal correct-group counts share the win.
- A player with no legal move: not applicable, questioner always has a legal move-and-announce action while stones remain to move.
- Repetition/cycles: not applicable, the setup phase is bounded by R5's gap-filled round limit.
- Simultaneous claims: not applicable, strictly sequential.
- Player elimination: not applicable.
- Resource exhaustion: a group reaching zero stones (R4) is an attested sub-event but not a game-ending one; the setup phase ends per R5.
- Termination guarantee: guaranteed given the R5 gap-filled round bound and R6's finite recall phase (bounded by the number of groups).
## State representation
Persistent state (hidden from guesser): true stone count per group. Guesser-visible state: which group was just referenced, and "no more" announcements per R4. Per-turn ephemeral: the current group index being processed. Information asymmetry is the entire point of the game — the guesser must never see the true board state during setup.
## Action space
- `move_stone(from_group, to_group)`: questioner's private action during setup, invisible to the guesser except for the group-referenced announcement and any "no more" event.
- `recall(group_index, claimed_count)`: guesser's action during the recall phase.
## Playability notes
- Single-player-vs-AI: natural — a human guesser against an AI questioner performing randomized moves, or vice versa with the AI as guesser exercising a memory model.
- Group games (n>4): the questioner role trivially supports multiple players taking turns announcing moves; the guesser role is inherently solo.
- Physical-skill games: none — purely a memory exercise.
- What is lost: essentially nothing structural; this is one of the most directly portable games in the Junod set, since it is already an abstract information-tracking game rather than a physical-skill one.
## Implementation hints
Trivial to implement as a memory minigame: track true state server-side, hide it from the guesser's client, and score the final recall phase against ground truth. Main design choice left open by the source (and left as a documented gap) is how many setup rounds to run and how to score partial correctness.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Nyengeli-nyengeli mune (fruit-stone memory game) (South Africa). (n.d.). Games of Countries. https://gamesofcountries.com/games/za-south-africa/nyengeli-nyengeli
@misc{gamesofcountries-za-south-africa-nyengeli-nyengeli,
title = {Nyengeli-nyengeli mune (fruit-stone memory game) (South Africa)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/za-south-africa/nyengeli-nyengeli}
}