Skip to main content
Formal Specification

Sahrawi desert hare tracking game (Nayrouba / La Liebre del Desierto) — Formal Specification

Western Sahara · guessing-hidden · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: directGap-fills: 1
# Nayruba — AI-optimized specification ## General idea A two-player deductive game of hidden state and query optimization. The Tracker hides 1 primary token (Mother Hare) and 4 secondary tokens (Leverets) across 5 discrete bins. The Interrogator receives a budget of 3 structured queries (e.g. membership checks, parity checks, regional sums) before submitting a full state hypothesis. Scoring rewards hypothesis accuracy vs Tracker deception. ## Formal model - **Abstraction taken:** Direct discrete constraint-satisfaction model. State is a vector B=(b0,b1,b2,b3,b4) where each bi=(hi,ci) with hi∈{0,1} (mother hare indicator, ∑hi=1) and ci∈{0,1,2,3,4} (leveret count, ∑ci=4). - **Player/agent count:** 2 agents, Agent 0 (Tracker) and Agent 1 (Interrogator). - **Turn model:** Sequential sub-phases: (1) Secret configuration, (2) Query-response loop (3 iterations), (3) Hypothesis submission and score resolution. ## Formal rules R1. **State Space and Configuration:** - The Tracker secretly chooses a valid configuration B={(hi,ci)}i=04 such that ∑hi=1 and ∑ci=4. R2. **Query Phase:** - The Interrogator has a budget of exactly 3 queries. In each step, the Interrogator selects one query type: - `QUERY_HARE(bin_idx)`: Returns `TRUE` if hbin_idx==1, else `FALSE`. - `QUERY_PARITY(bin_idx)`: Returns cbin_idx(mod2). - `QUERY_SUM_INTERVAL(start_idx, end_idx)`: Returns ∑j=start_idxend_idxcj. - The Tracker returns the mathematically truthful answer. R3. **Hypothesis and Scoring:** - The Interrogator submits a final guess Bˆ={(hˆi,cˆi)}i=04. - Scoring resolution: - If hˆi==hi for all i: Interrogator receives +2 points. - If cˆi==ci for all i: Interrogator receives +3 bonus points. - If the Interrogator fails to locate the mother hare (hˆ≠h): Tracker receives +2 points. R4. **Role Swap:** - After each round, roles swap (Tracker↔Interrogator). R5. **Victory Condition:** - The first agent to accumulate S≥10 points wins the match immediately. R6. [gap-fill] **Round Limit:** If neither agent reaches 10 points after R=20 rounds, the agent with the highest score wins. ## Edge cases - **Tie at the end condition:** If tied at round R=20, resolved as a draw. - **A player with no legal move:** Configuration selection and query execution are always mathematically well-defined. - **Repetition / cycles:** Bounded by round limit R=20 (R6). - **Simultaneous claims on the same resource or position:** Not applicable; strictly sequential query protocol. - **Player elimination mid-round; turn order repair:** Not applicable (2-agent game). - **Resource exhaustion:** Action space and state configurations are finite (5×(4+5−14)=5×70=350 total valid states). - **Termination guarantee:** Guaranteed by finite state space, 3-query budget per round, and 20-round cap. ## State representation - **Persistent state:** - `scores`: `[score_0, score_1]`. - `tracker_role`: `0` or `1`. - `round_count`: integer ∈[1,20]. - **Per-round ephemeral state:** - `true_state`: vector of 5 pairs (hi,ci) (private to Tracker). - `query_history`: list of up to 3 `(query_type, params, answer)` tuples (public to both agents). - **Observable information:** Interrogator observes `scores`, `query_history`, and answer responses, but not `true_state` until scoring resolution. ## Action space - Tracker: `set_configuration(vector of 5 pairs (h_i, c_i))`. - Interrogator: `query_hare(idx \in 0..4)`, `query_parity(idx \in 0..4)`, `query_sum(start, end)`, `submit_hypothesis(vector of 5 pairs)`. ## Playability notes - **Single-player-vs-AI:** Natural 1v1 deductive puzzle game. AI Interrogator uses information-gain maximization (Shannon entropy reduction over the 350-state hypothesis space). AI Tracker chooses configurations that maximize remaining entropy under optimal query trees. - **Group games (n>4):** Reduces cleanly to 2 competing teams collaborating on queries and placement. - **Physical-skill games:** Not applicable (pure discrete deduction). - **What is lost in the adaptation:** The poetic Hassaniya chanting, desert tracking metaphors, and tactile handling of smooth pebbles. ## Implementation hints - Total state space is exactly 350 possible configurations. - Interrogator AI can maintain an exact candidate set S⊆{0…349} and select queries that partition S into subsets of equal size. - First milestone: Implement state generator and exact query response validator before implementing entropy-based AI search.

Download as JSON

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

Download JSON
Cite this page

Sahrawi desert hare tracking game (Nayrouba / La Liebre del Desierto) (Western Sahara). (n.d.). Games of Countries. https://gamesofcountries.com/games/eh-western-sahara/nayruba

@misc{gamesofcountries-eh-western-sahara-nayruba,
  title = {Sahrawi desert hare tracking game (Nayrouba / La Liebre del Desierto) (Western Sahara)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/eh-western-sahara/nayruba}
}