Formal Specification
Mauritanian bone-doll and tent roleplay (Al-Awzar) — Formal Specification
Mauritania · guessing-hidden · Confidence: medium
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: directGap-fills: 1
# Ardal — AI-optimized specification
## General idea
A multi-round turn-based social simulation and 20-questions style deduction game. In each round, the Host agent arranges a set of categorized characters within a miniature tent structure and secretly conceals a token (ring) on one character. The Guest agent queries categorical attributes (role, wealth, livestock, position) under a strict question budget to deduce the target character's identity.
## Formal model
- **Abstraction taken:** The physical crafting of bone dolls and miniature tents is formalized as a discrete character database where each character possesses attributes: , , and .
- **Player/agent count:** 2 agents (Host , Guest ), swapping roles after each round.
- **Turn model:** Alternating inquiry rounds.
## Formal rules
R1. **Character Domain & Setup:**
- Database initialized with unique characters with distinct attribute tuples .
- Host secretly selects target character .
R2. **Guest Query Phase:**
- Guest has a budget of exactly binary attribute queries:
- Query format: `query_attribute(attribute_name, candidate_value)`.
- Host must return honest boolean response .
R3. **Final Deduction & Scoring:**
- After 3 queries, Guest executes `declare_character(character_id)`:
- If : **Correct Deduction!** Guest scores points.
- If : **Failed Deduction!** Host scores points.
R4. **Hospitality Dialogue Bonus:**
- [gap-fill] Both agents exchange structured dialogue tokens representing traditional greeting formulas; submitting valid linguistic tokens within the dialogue phase awards bonus point to each agent.
R5. **Round Swapping & Victory:**
- Roles swap () after each round.
- Match continues for a fixed total of 4 rounds (2 rounds as Host, 2 as Guest).
- The agent with the higher cumulative score after 4 rounds wins.
## Edge cases
- **Tie at the end condition:** If scores are tied after 4 rounds, sudden-death 1-round deduction with 2 queries determines the winner.
- **A player with no legal move:** Query actions from the valid attribute space are always legal.
- **Repetition / cycles:** Bounded by fixed 4-round structure and finite 3-query budget per round.
- **Simultaneous claims on the same resource or position:** Turns within a round are strictly ordered (Placement Query sequence Guess).
- **Player elimination mid-round; turn order repair:** Not applicable — 2-agent round-robin structure.
- **Resource exhaustion:** Exactly characters and 3 queries per round.
- **Termination guarantee:** Guaranteed by strictly terminating 4-round match horizon.
## State representation
- **Persistent state:** Match scores , round index , current Host ID .
- **Per-round state:** Target character ID , remaining query budget , query history list.
- **Observable information:** Imperfect information: Guest observes character database and query responses, but is hidden until declaration.
## Action space
- For Host: `select_secret_character(char_id)`, `answer_query(boolean_val)`.
- For Guest: `query_attribute(attr_name, value)`, `declare_character(char_id)`.
## Playability notes
- **Single-player-vs-AI:** Highly engaging: AI Guest applies information-entropy maximization (decision tree search) to select optimal queries; AI Host plays optimal mixed distributions over character sets.
- **Group games ():** Can be extended to multi-team encampment competitions.
- **Physical-skill games:** Not applicable (pure categorical deduction and dialogue simulation).
- **What is lost in the adaptation:** The physical sensory beauty of handling decorated bone figurines, indigo dyes, and miniature woolen tents.
## Implementation hints
- Implement character database as an array of structs/dicts and evaluate information gain per query using Shannon entropy.
- First milestone: Implement 6-character database with 3-query binary search and verify scoring and role-swap logic.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Mauritanian bone-doll and tent roleplay (Al-Awzar) (Mauritania). (n.d.). Games of Countries. https://gamesofcountries.com/games/mr-mauritania/ardal
@misc{gamesofcountries-mr-mauritania-ardal,
title = {Mauritanian bone-doll and tent roleplay (Al-Awzar) (Mauritania)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/mr-mauritania/ardal}
}