Formal Specification
Setting out the pied frog — Formal Specification
Mongolia · dice-lots · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: directGap-fills: 2
# Alag melkhii örökh — AI-optimized specification
## General idea
A turn-based collection and resource-harvesting game where players roll a six-sided die to dismantle a 92-token anatomical turtle/frog layout on a central board. Each die face maps to specific body parts; if an agent rolls a number whose corresponding body parts are already exhausted, the agent must pay penalty tokens back to the board, creating a push-pull inventory dynamic.
## Formal model
- **Abstraction taken:** Direct discrete formalization. The 92 knucklebones on the board are partitioned into 6 distinct categorical inventory buckets matching the die values `1..6`.
- **Player/agent count:** 2 to 6 players (default 4 players).
- **Turn model:** Sequential discrete turns, rotating in fixed clockwise order.
## Formal rules
R1. Setup: Board inventory buckets are initialized to exact anatomical capacities totaling 92 knucklebones:
- `Bucket(1) = 2` (Heart: 1, Bladder: 1)
- `Bucket(2) = 6` (Eyes: 2, Ears: 2, Kidneys: 2)
- `Bucket(3) = 6` (Head: 3, Tail: 3)
- `Bucket(4) = 16` (4 Legs × 4 bones)
- `Bucket(5) = 20` (4 Paws × 5 bones)
- `Bucket(6) = 42` (Neck: 6 + Spine/Back: 36 in 6 rows of 6)
- Total board tokens = 92. All player banks initialize to 0.
R2. On their turn, the active player rolls a standard uniform six-sided die `d in {1..6}`.
R3. Harvest resolution:
- If `Bucket(d) >= d`: The player takes exactly `d` tokens from `Bucket(d)`. Player bank increases by `d`; `Bucket(d)` decreases by `d`.
- If `0 < Bucket(d) < d`: The player takes all remaining tokens in `Bucket(d)`. Player bank increases by `Bucket(d)`; `Bucket(d)` becomes 0.
R4. Depletion penalty:
- If `Bucket(d) == 0`: The player must return `d` tokens from their personal bank back to `Bucket(d)`.
- [gap-fill] If the player holds fewer than `d` tokens in their bank, they return all tokens they currently hold (`Bucket(d) += bank`, `bank = 0`); no negative debt is carried.
R5. [gap-fill] Extra turn on full-row harvest: If an agent successfully harvests an entire 6-token spine row or 5-token paw, they do NOT take an extra roll; turn passes immediately to prevent infinite single-turn runs.
R6. Termination: The game terminates the instant the total tokens remaining across all buckets on the board reaches 0 (`sum(Bucket(1..6)) == 0`).
R7. Win condition: The player with the highest total knucklebones in their bank at termination wins. In case of a tie, the victory is shared as joint good fortune.
## Edge cases
- Tie at the end condition: Equal collected totals result in a shared victory (consistent with Tsagaan Sar collective blessing ethos).
- A player with no legal move: Not applicable — rolling the die is always mandatory and legal.
- Repetition / cycles: While penalty returns restore tokens to empty buckets, the asymmetric capture dynamic and non-zero harvest rates ensure expected board depletion over time.
- Simultaneous claims on the same resource: Not applicable — turns are strictly sequential.
- Player elimination mid-round: Not applicable — players with 0 bank remain active and may harvest on subsequent rolls.
- Resource exhaustion: Complete board exhaustion is the explicit game termination trigger.
- Termination guarantee: The Markov chain of total board tokens has an absorbing state at 0 with probability 1 as turn count `T -> infinity`.
## State representation
- Persistent state: `board_buckets` (6-element array of integers, representing remaining tokens in buckets 1..6), `player_banks` (array of integers per player), `active_player_index` (0..n-1), `turn_counter` (integer).
- Per-turn ephemeral state: Sampled die roll `d in {1..6}`, harvest quantity or penalty quantity.
- Observable information: Fully observable — all bucket capacities and player bank sizes are public.
## Action space
- `roll_die()`: Trigger uniform stochastic sample from `{1, 2, 3, 4, 5, 6}`. No decision parameters required.
## Playability notes
- Single-player-vs-AI: Natural — 1 human vs. 1–3 AI agents tracking probability distributions of remaining buckets.
- Group games (n>4): Extends naturally up to 8 players without altering the discrete mechanics.
- Physical-skill games: Deterministic accounting around a stochastic die randomizer.
- What is lost in the adaptation: The tactile aesthetic of rearranging bone pieces into the turtle silhouette on the felt rug.
## Implementation hints
Extremely simple and lightweight state machine. Can be implemented in under 100 lines of code. Milestone 1: basic turn loop and bucket accounting; Milestone 2: Monte Carlo simulation to compute expected game duration (~40–60 rolls for 4 players).
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Alag Melkhii (Mongolia). (n.d.). Games of Countries. https://gamesofcountries.com/games/mn-mongolia/alag-melkhii
@misc{gamesofcountries-mn-mongolia-alag-melkhii,
title = {Alag Melkhii (Mongolia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/mn-mongolia/alag-melkhii}
}