Skip to main content
Formal Specification

Finger Flashing Sum Guessing (Morra) — Formal Specification

Montenegro · guessing-hidden · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: directGap-fills: 2
# Šijavica — AI-optimized specification ## General idea A simultaneous-choice matrix game of ancient Mediterranean origin (*Morra*). Two agents simultaneously throw a hand displaying 0 to 5 fingers while simultaneously calling a predicted sum from 0 to 10. If exactly one agent correctly calls the true total sum of all extended fingers, that agent scores a point (*punat*). Strategic depth stems from rapid pattern recognition, opponent frequency analysis, and entropy manipulation. ## Formal model - **Abstraction taken:** Direct mathematical representation of the traditional rules. No physical abstraction needed; actions map 1-to-1 to discrete numbers. - **Player/agent count:** 2 agents (Agent A and Agent B). - **Turn model:** Simultaneous discrete rounds (throws). ## Formal rules R1. Setup: State initializes `score_A = 0`, `score_B = 0`. Target score is `WIN_SCORE = 16`. R2. Simultaneous action: Each round, both agents simultaneously submit an action tuple `throw_and_call(fingers, call)`: - `fingers in [0, 1, 2, 3, 4, 5]` - `call in [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]` - [gap-fill] Consistency constraint: An agent's call must be mathematically possible given their own thrown fingers: `call >= fingers` and `call <= fingers + 5`. Actions violating this constraint are invalid and automatically default to `call = fingers + 2`. R3. Resolution: - True sum is evaluated: `actual_sum = fingers_A + fingers_B`. - Predictor correctness is checked: - `correct_A = (call_A == actual_sum)` - `correct_B = (call_B == actual_sum)` R4. Point allocation: - If `correct_A and not correct_B`: `score_A = score_A + 1`. - If `correct_B and not correct_A`: `score_B = score_B + 1`. - If `(correct_A and correct_B)` or `(not correct_A and not correct_B)`: Draw throw (*fola*); no score change. R5. Match victory: The match ends immediately when an agent reaches `score >= WIN_SCORE`. The first agent to reach `WIN_SCORE` wins. R6. [gap-fill] Turn limit safety: If a match exceeds 200 throws without reaching `WIN_SCORE` (extremely rare), the agent currently leading in points is declared winner; if tied at throw 200, play continues under sudden death until the first scored point. ## Edge cases - Tie at the end condition: Points are awarded one at a time; because draws award 0 points to both agents, two agents can never reach `WIN_SCORE` simultaneously on the same throw. - A player with no legal move: Not applicable; 36 valid `(fingers, call)` combinations are always available. - Repetition / cycles: Draws award 0 points but do not cycle state backward; R6 provides a hard loop-breaking threshold. - Simultaneous claims on the same resource or position: Simultaneous correct calls resolve cleanly as a draw per R4. - Player elimination mid-round; turn order repair: 2-agent game; forfeiture awards immediate victory to the opponent. - Resource exhaustion: No consumable tokens or pieces. - Termination guarantee: Non-zero scoring probability per throw (under mixed strategies, `P(unique win) > 0`) combined with R6 guarantees termination. ## State representation - Persistent state: `score_A: int`, `score_B: int`, `throw_count: int`. - Per-throw state: `action_A: (int, int)`, `action_B: (int, int)`, `actual_sum: int`, `round_winner: Optional[AgentID]`. - Observable information: Full past history of all throws and calls is public common knowledge; current-round choices are strictly private until simultaneous reveal. ## Action space - `throw_and_call(fingers: int, call: int)`: Parameters `fingers in {0..5}`, `call in {fingers..fingers+5}`. ## Playability notes - Single-player-vs-AI: Engine-ready; profound game-theoretic benchmark (mixed Nash equilibrium, opponent modeling, Markov chain frequency prediction). - Group games (n>4): Structured as rotating king-of-the-hill or team tag relays. - Physical-skill games: Not applicable (pure discrete cognition, speed, and psychology). - What is lost: Table-slamming acoustic drama, rapid-fire Venetian vocal cadence, and physical eye-stare intimidation. ## Implementation hints One of the cleanest simultaneous-move zero-sum games in existence. Matrix payoff size is 36x36. Milestone 1: Implement simultaneous resolver and test uniform-random vs. Nash equilibrium agent. Milestone 2: Implement n-gram Markov opponent modeler.

Download as JSON

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

Download JSON
Cite this page

Finger Flashing Sum Guessing (Morra) (Montenegro). (n.d.). Games of Countries. https://gamesofcountries.com/games/me-montenegro/sijavica

@misc{gamesofcountries-me-montenegro-sijavica,
  title = {Finger Flashing Sum Guessing (Morra) (Montenegro)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/me-montenegro/sijavica}
}