Skip to main content
Formal Specification

Al-Dassasa (Sudanese Sand Mound Concealment) — Formal Specification

Sudan · guessing-hidden · Confidence: medium

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: directGap-fills: 2
# Al-Dassasa — AI-optimized specification ## General idea A hidden-location deduction and deception game played with K sand mound slots. The Hider secretely buries a token in one mound while emitting noisy behavioral bluff signals. The Guesser evaluates the signal vector and chooses between high-risk direct target strike or sequential elimination sweep. ## Formal model - **Abstraction taken:** The physical sand mounds are indexed 0..K−1 (default K=5). The Hider's eye contact and facial micro-expressions are formalized as a continuous 2D behavioral observation vector 𝐱∈ℝ2 containing target-correlated signal plus Gaussian noise. - **Player/agent count:** 2 agents (Hider and Guesser). - **Turn model:** Discrete round structure with alternating roles. ## Formal rules R1. Setup: K=5 mound slots {0,1,2,3,4}. R2. Hiding Phase: - Hider secretly selects token index h∗∈{0..4}. - Hider generates an observable signal vector 𝐱∈ℝ2: 𝐱=𝛍(h∗)+𝛜,𝛜∼𝒩(0,σbluff2I) where higher bluff skill decreases signal correlation with h∗. R3. Guessing Phase: - Guesser observes 𝐱 and selects one of two guess modes: - **Mode A (Direct Strike):** Guesser chooses one index g∈{0..4}. - If g==h∗: Guesser scores **+3 points**. - If g≠h∗: Hider scores **+1 point**. - **Mode B (Sequential Sweep):** Guesser chooses an ordered elimination list [e1,e2,…,eK−1] attempting to eliminate all 4 empty mounds without touching h∗. - If h∗ is swept before the final step: sweep fails; Hider scores **+2 points**. - If all 4 empty mounds are successfully swept (leaving only h∗): Guesser scores **+2 points**. R4. Role Swap: Roles alternate every round (Hider becomes Guesser; Guesser becomes Hider). R5. [gap-fill] Match Termination: Game terminates after 10 complete rounds (5 rounds in each role). R6. [gap-fill] Win Condition: First agent to reach 10 points (or highest score at round 10) wins. ## Edge cases - Tie at the end condition: If both agents have equal score after 10 rounds, 2 additional sudden-death rounds are played (1 in each role). - A player with no legal move: In Guessing Phase, valid indices in {0..K−1} are always available. - Repetition / cycles: Strictly bounded by the 10-round match limit (R5). - Simultaneous claims on the same resource or position: Actions are strictly sequential. - Player elimination mid-round; turn order repair: No mid-round elimination; both agents play all 10 rounds. - Resource exhaustion (pieces, tokens, space, time): Exactly 1 token hidden per round; 10 rounds fixed. - Termination guarantee: Monotonic round counter ensures termination at round 10. ## State representation - Persistent state: - `agent_scores`: `{AGENT_0: int, AGENT_1: int}`. - `current_round`: integer in `1..10`. - `active_hider`: `AGENT_0 | AGENT_1`. - Per-round ephemeral state: - `hidden_mound`: integer h∗∈{0..4} (private to Hider). - `observed_signal`: vector 𝐱∈ℝ2 (public). - Observable information: Signal 𝐱, score, and chosen mode are public. Exact index h∗ is hidden until round resolution. ## Action space - Hider: `hide_token(mound_index, bluff_strategy)`. - Guesser: `guess_direct(mound_index)` or `guess_sweep(elimination_order)`. ## Playability notes - Single-player-vs-AI: Engine-ready; play as Guesser running Bayesian inference over AI signals, or as Hider trying to out-bluff the AI. - Group games (n>4): Not applicable; strictly 2-agent duel. - Physical-skill games: Direct translation; discrete decision problem with noisy observation channel. - What is lost in adaptation: The tactile feel of scooping and sifting fine sand. ## Implementation hints Discrete hidden-parameter Bayesian estimation game. Guesser agent can compute posterior P(h∗=k|𝐱) and evaluate expected value for Direct vs Sweep modes. First milestone: implement hidden index selection, signal emission, and Mode A scoring.

Download as JSON

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

Download JSON
Cite this page

Al-Dassasa (Sudanese Sand Mound Concealment) (Sudan). (n.d.). Games of Countries. https://gamesofcountries.com/games/sd-sudan/al-dassasa

@misc{gamesofcountries-sd-sudan-al-dassasa,
  title = {Al-Dassasa (Sudanese Sand Mound Concealment) (Sudan)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/sd-sudan/al-dassasa}
}