Formal Specification
Knucklebone flicking — Formal Specification
Mongolia · dexterity-small · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 1
# Shagai nyaslakh — AI-optimized specification
## General idea
A tabletop flicking game where a pool of sheep knucklebones rests scattered across a 2D plane in various animal stances. The active player identifies two separate bones showing identical stances, selects an unobstructed line of fire, and executes a finger flick to strike the target bone without touching any third obstacle bone. Strategy revolves around geometric target selection and risk evaluation.
## Formal model
- **Abstraction taken:** The 2D physical scattering of knucklebones is represented as a set of points `{(x_i, y_i, stance_i)}` on a bounded continuous plane `[0, W] x [0, H]`. A flick action specifies `source_index` and `target_index`; collision resolution is determined by line-of-sight clearance plus a stochastic accuracy variance.
- **Player/agent count:** 2 to 4 players.
- **Turn model:** Sequential discrete turns; successful hits allow continuation, while misses/fouls pass the turn.
## Formal rules
R1. Setup: `N = 36` knucklebones are randomly distributed on the board with coordinates `(x, y)` and random stance assignments (`Horse`, `Camel`, `Sheep`, `Goat`).
R2. On a turn, the active player chooses two bones `i` and `j` where `stance_i == stance_j` and designates `i` as striker and `j` as target.
R3. Legal action precondition: `i != j` and both bones currently exist on the board with matching stances.
R4. Stochastic collision check: A ray is cast from `i` to `j`. The probability of a clean hit is `P(clean_hit) = max(0, 1 - distance(i, j)/D_max) * prod_{k != i,j}(1 - obstacle_penalty(k, ray))`.
R5. Valid hit resolution: If clean hit succeeds, target bone `j` is removed from the board and added to the active player's score pile; striker bone `i` is repositioned at `j` with a new randomized stance. The active player may immediately take another flick.
R6. Foul / Miss resolution: If the flick misses or grazes an obstacle bone `k`, the turn ends immediately. [gap-fill] If an obstacle bone was grazed, the player returns 1 previously collected bone to the board as a penalty.
R7. Turn rotation: Upon a miss or foul, play rotates clockwise to the next player.
R8. Termination & Win: The game terminates when fewer than 2 knucklebones of any single stance remain on the board. The player with the highest collection count wins.
## Edge cases
- Tie at the end condition: If two players hold identical counts at board clearance, the game is declared a draw, or resolved via a 4-bone sudden-death flick contest.
- A player with no legal move: If no two bones share an identical stance, the game terminates immediately under R8.
- Repetition / cycles: Knucklebones are strictly removed on each successful hit, ensuring monotonic reduction of board state and guaranteed termination.
- Simultaneous claims on the same resource: Not applicable — turns are strictly sequential.
- Player elimination mid-round: Not applicable — all players participate until board exhaustion.
- Resource exhaustion: Board exhaustion is the primary victory trigger.
- Termination guarantee: Monotonically decreasing piece count guarantees termination in at most `N` successful steps.
## State representation
- Persistent state: `bones_on_board` (list of structs `{id, x, y, stance, active}`), `player_scores` (map of Player -> integer), `active_player_index` (0..n-1).
- Per-turn ephemeral state: Selected pair `(i, j)`, sampled collision outcome (success / miss / foul).
- Observable information: Fully observable — all bone coordinates and stances are public knowledge.
## Action space
- `flick_matching_pair(source_id, target_id)`: Select a valid pair of active bones with identical stances on the board.
## Playability notes
- Single-player-vs-AI: Natural — 1 human vs. 1 AI agent optimizing line-of-sight geometry and target proximity.
- Group games (n>4): Reduces smoothly to 2 agents without modifying mechanics.
- Physical-skill games: Accuracy modeled via angular deviation `θ ~ N(0, σ_flick^2)` along the shot vector.
- What is lost in the adaptation: The tactile feel of bone surfaces, spinning bounces, and the physical micro-friction of the felt rug.
## Implementation hints
2D geometry ray-circle intersection algorithm. Milestone 1: implement line-of-sight geometric checker and deterministic hit/miss logic; Milestone 2: add Gaussian angular jitter and penalty bone restocking.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Knucklebone flicking (Mongolia). (n.d.). Games of Countries. https://gamesofcountries.com/games/mn-mongolia/shagai-nyaslakh
@misc{gamesofcountries-mn-mongolia-shagai-nyaslakh,
title = {Knucklebone flicking (Mongolia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/mn-mongolia/shagai-nyaslakh}
}