Skip to main content
Formal Specification

Mongolian interlocking wooden puzzles (Burr puzzles) — Formal Specification

Mongolia · dexterity-small · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 1
# Modon onis — AI-optimized specification ## General idea A deterministic sequential spatial puzzle where an agent must disassemble and reassemble a three-dimensional interlocking wooden burr structure. The structure is composed of orthogonal notched rectangular bars that constrain each other through geometry; only one specific "key piece" (*onisnii gol*) possesses an initial unblocked degree of freedom, unlocking a deterministic sequence of sliding movements. ## Formal model - **Abstraction taken:** The 3D physical burr puzzle is formalized as a discrete constraint satisfaction and state-space graph search problem. Each component timber is a 3D voxel grid or rigid bounding box with allowed translation vectors along orthogonal axes `{+X, -X, +Y, -Y, +Z, -Z}`. - **Player/agent count:** 1 solver (or 2 competing agents racing in parallel to reach the target state). - **Turn model:** Sequential discrete steps; on each step, the agent selects one piece and a translation direction. ## Formal rules R1. Setup: In the standard 6-piece classical cross burr (*Зургаан модны оньс*), 6 notched bars occupy predefined relative voxel coordinates in a locked configuration `S_initial`. R2. On each step, the agent selects an action `slide(piece_id, direction, distance)`. R3. Movement legality: An action is legal if and only if shifting `piece_id` by `distance` along `direction` does not cause its solid voxels to collide with the voxels of any other stationary piece. R4. Key piece identification: In `S_initial`, exactly one piece (the *onisnii gol*) has a non-zero legal translation distance; all other pieces have 0 legal moves. R5. Sequential unlocking: Sliding out a piece creates empty space, unlocking new degrees of freedom for adjacent pieces. R6. Complete disassembly target: The disassembly goal state `S_disassembled` is reached when all 6 pieces have Euclidean pairwise distances greater than the puzzle bounding radius (fully separated). R7. Reassembly target: From `S_disassembled`, the agent must execute the inverse transformation sequence to reconstruct `S_initial`. R8. [gap-fill] Move cap: For an n-piece burr, a move limit of `M = 5 * optimal_path_length` is enforced to prevent infinite search loops. R9. Win condition: In a 2-agent speed duel, the agent that completes `S_initial -> S_disassembled -> S_initial` in the fewest steps (or lowest simulated time) wins. ## Edge cases - Tie at the end condition: In a 2-agent race, identical step counts result in a draw. - A player with no legal move: Occurs if an illegal dead-end state is reached during reassembly; the agent must backtrack. - Repetition / cycles: Bounded by R8 move cap. - Simultaneous claims on the same resource: Not applicable — solo state search or independent parallel puzzle instances. - Player elimination mid-round: Not applicable. - Resource exhaustion: Not applicable. - Termination guarantee: Guaranteed by finite graph state space and R8 move cap. ## State representation - Persistent state: `piece_positions` (array of 6 translation vectors `(x, y, z)`), `piece_voxels` (fixed 3D geometry per piece), `step_counter` (integer), `puzzle_phase` (`disassembling` | `reassembling` | `completed`). - Per-turn ephemeral state: Selected `piece_id`, chosen direction vector, collision check result. - Observable information: Fully observable — complete 3D geometry and relative piece coordinates are common knowledge. ## Action space - `slide_piece(piece_id, axis, distance)`: Translate piece `piece_id` along `axis` in `{-1, +1}` by integer voxel units. ## Playability notes - Single-player-vs-AI: Natural — solver search algorithm (A*, BFS, or reinforcement learning agent) navigating the burr state graph. - Group games (n>4): Solitary puzzle or parallel tournament heat. - Physical-skill games: Deterministic logic puzzle; physical dexterity is abstracted into discrete move steps. - What is lost in the adaptation: The tactile friction of polished cedar wood, the subtle manual jiggle used to locate the loose key piece, and physical carpentry imperfections. ## Implementation hints Standard 3D voxel collision check or interval-overlap bounding box algebra. Milestone 1: implement classical 6-piece burr disassembly graph (level-1 burr requiring 1 key slide); Milestone 2: implement multi-step higher-level burr reassembly search.

Download as JSON

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

Download JSON
Cite this page

Mongolian interlocking wooden puzzles (Burr puzzles) (Mongolia). (n.d.). Games of Countries. https://gamesofcountries.com/games/mn-mongolia/modon-onis

@misc{gamesofcountries-mn-mongolia-modon-onis,
  title = {Mongolian interlocking wooden puzzles (Burr puzzles) (Mongolia)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/mn-mongolia/modon-onis}
}