Skip to main content
Formal Specification

Scientific Great Broom — Formal Specification

Italy · card · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: directGap-fills: 1
# Scopone Scientifico — AI-optimized specification ## General idea A four-agent, discrete, imperfect-information partnership card game where the entire 40-card deck is dealt directly to the four players (10 cards each) with an initially empty table. Agents play sequentially to capture single cards or additive combinations, using card counting and suit parity to generate sweeps (*scope*) while denying capture setups to opponents. ## Formal model - **Abstraction taken:** None; the game is a finite, discrete, perfect-recall imperfect-information extensive-form game. - **Player/agent count:** 4 agents arranged as Team 0 (Agents 0, 2) vs Team 1 (Agents 1, 3). - **Turn model:** Strictly sequential, 40 total turns per deal (10 turns per agent in fixed rotation). ## Formal rules R1. Deck structure: 40 cards, 4 suits (Coins, Cups, Swords, Batons) with values V∈{1..10}. Match target score is T=11. R2. Deal: Dealer shuffles and distributes 10 private cards to each of the 4 agents. Initial table state is Table=∅. R3. Turn action: On turn t∈[1,40], active agent i plays one card c∈Handi. - Single Match Rule: If ∃t∈Table with val(t)=val(c), agent must capture t. - Additive Match Rule: Else if ∃S⊆Table with ∑s∈Sval(s)=val(c), agent captures subset S. - Discard: Else, c is added to Table. R4. Scopa evaluation: If a legal capture results in Table=∅ on any turn t<40, the capturing team immediately scores +1 *scopa* point. R5. Last trick collection: On turn t=40, any remaining cards on the table are awarded to the capture pile of the agent who made the last legal capture (no scopa awarded on turn 40). R6. Deal scoring: - Scope: +1 per recorded scopa. - Carte: +1 to team with |Captures|≥21 (0 if 20-20). - Denari: +1 to team with |Coins|≥6 (0 if 5-5). - Settebello: +1 to team holding 7 of Coins. - Primiera: +1 to team with highest sum of best-in-suit Primiera values (7→21,6→18,1→16,5→15,4→14,3→13,2→12,faces→10). R7. [gap-fill] Match win condition: The first team to accumulate ≥11 points at the conclusion of a deal wins. In case of a tie at ≥11, additional complete deals are played until the tie is broken. ## Edge cases - Tie at the end condition: Ties at ≥11 points trigger additional deals until one team holds a strictly higher point total (R7). - A player with no legal move: Impossible; any card in hand is legally playable every turn. - Repetition / cycles: Impossible; exactly 40 total card plays occur per deal, monotonically reducing hand sizes from 10 to 0. - Simultaneous claims on the same resource or position: Not applicable; turns are strictly sequential. - Player elimination mid-round; turn order repair: Not applicable; exactly 4 fixed agents. - Resource exhaustion (pieces, tokens, space, time): The 40-card deal exhausts in exactly 40 turns. - Termination guarantee: Monotonic progress over 40 turns per deal, and positive point allocation per non-drawn deal guarantee match termination. ## State representation - Persistent state: `score_team_0`, `score_team_1`: integers. - Per-deal state: - `table`: set of cards on the table. - `hands`: array of 4 card lists (size 10 down to 0). - `captured_team_0`, `captured_team_1`: list of cards. - `scopa_team_0`, `scopa_team_1`: integers. - `last_capturer`: agent ID. - `turn_index`: integer ∈[1,40]. - Observable information: Table cards, discarded cards, public captures, and announced scope are public. Partner and opponent hands are private. ## Action space - `play_card(card_id: int, capture_subset: Optional[List[int]])` ## Playability notes - Single-player-vs-AI: Engine-ready; excellent domain for imperfect-information game AI, belief-state particle filtering, and deep reinforcement learning. - Group games (n>4): Not applicable; strictly 4 agents in 2 fixed pairs. - Physical-skill games: Direct discrete rules; no physical skill. - What is lost in the adaptation: Table psychology, glance signaling, and conversational tension. ## Implementation hints State transitions are deterministic given card choices. The decision tree has depth 40 with branching factor between 1 and 10 per agent turn. First milestone: Implement the 40-card deal, empty-table start, parity tracking, and terminal score evaluation.

Download as JSON

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

Download JSON
Cite this page

Scientific Great Broom (Italy). (n.d.). Games of Countries. https://gamesofcountries.com/games/it-italy/scopone-scientifico

@misc{gamesofcountries-it-italy-scopone-scientifico,
  title = {Scientific Great Broom (Italy)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/it-italy/scopone-scientifico}
}