Skip to main content
Formal Specification

Five Stones (Cypriot Knucklebones & Pebble Tossing) — Formal Specification

Cyprus · dexterity-small · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 1
# Pentovola — AI-optimized specification ## General idea A turn-based sequential dexterity game where players progress through 6 hierarchical stages of tossing, scooping, and catching 5 pebbles. Physical dexterity is abstracted into a stochastic skill-and-difficulty model; the player's tactical decisions involve risk management (choosing scatter layouts, picking the mother stone, and balancing risk vs reward in the final scoring snatch). ## Formal model - **Abstraction taken:** The continuous spatial scattering and hand-toss trajectory are replaced by a 2D spatial point distribution of 5 pebbles on a plane [0,1]2, coupled with parameterized Bernoulli success trials for each level's catch attempt: P(success)=σ(skill−difficulty(stage,distances)). - **Player/agent count:** 2 to 6 agents (P1,…,PM). - **Turn model:** Sequential turn rotation. A player continues taking actions until failing an execution trial (committing a foul), at which point the turn passes to the next player. ## Formal rules R1. **Initial Setup:** Each agent starts with score Si=0 and current stage level Li=1 (where L∈{1,2,3,4,5,6}). R2. **Turn Initiation & Scatter:** - Active agent Pi executes `scatter_stones()`, generating 5 2D coordinates {𝐩1,…,𝐩5} sampled from a bivariate uniform distribution on [0,1]2. - Agent selects one index m∈{1..5} as the Mother Stone (`select_mother(m)`). R3. **Stage Execution (Levels 1–5):** - **Level 1 (Singles):** 4 successive sub-trials. In each, the agent targets one resting stone. Success probability P=σ(βskill−δ1). On success, stone is collected. If all 4 succeed, Li←2. - **Level 2 (Pairs):** 2 successive sub-trials, each targeting a pair of stones. Difficulty increases with Euclidean distance between paired stones: δ2=δ0+α‖𝐩j−𝐩k‖. If both pairs succeed, Li←3. - **Level 3 (Triples):** 2 sub-trials (pick 3 stones together, then pick 1). If both succeed, Li←4. - **Level 4 (Quads):** 1 sub-trial (pick all 4 resting stones in one grab). If successful, Li←5. - **Level 5 (Arch/Gate):** 4 sub-trials (flick each stone through the designated finger arch). If all succeed, Li←6. R4. **Stage 6 (Final Scoring Snatch):** - The agent tosses all 5 stones. The back-of-hand catch samples a count Cback∼Binomial(5,Pback). - If Cback>0, the agent executes a snatch flip. The caught count Cfinal∼Binomial(Cback,Psnatch). - Score updates: Si←Si+Cfinal×20. - Level resets: Li←1 (completed a cycle). R5. **Foul / Turn Failure:** - If any sub-trial in R3 or R4 fails (sampled outcome = 0), the active agent's turn immediately ends. - State preservation: Li remains at the failed level. Turn passes to player (imodM)+1. R6. **Win Condition:** - The first agent to reach or exceed Starget (e.g. Starget=100) wins immediately. R7. [gap-fill] If all players fail consecutively for 5 complete round rotations without advancing any level, baseline skill threshold βskill is boosted by +0.5 to break potential stagnation. ## Edge cases - **Tie at the end condition:** Not applicable — turns are sequential; the first agent to cross Starget on their turn triggers immediate game victory. - **A player with no legal move:** On an active turn, an agent always has the legal action to attempt the current stage trial. - **Repetition / cycles:** Stalling at a difficult stage is bounded by R7 (adaptive skill boost guarantees eventual progression). - **Simultaneous claims on the same resource or position:** Not applicable — strictly sequential turns with isolated trials. - **Player elimination mid-round:** Not applicable — no player elimination; all players rotate turns until a player reaches Starget. - **Resource exhaustion:** 5 stone tokens are permanent; scores accumulate. - **Termination guarantee:** Guaranteed because P(success)>0 for all trials, Starget is finite, and R7 guarantees eventual level completion. ## State representation - **Persistent state:** Number of players M, active player index i∈{1..M}, score vector 𝐒∈ℕM, level vector 𝐋∈{1..6}M, target score Starget. - **Per-turn state:** Pebble coordinates {𝐩1,…,𝐩5}, selected mother index m, sub-trial progress index k. - **Observable information:** Fully observable to all agents (all stone positions, stages, and scores are public). ## Action space - `scatter_stones()`: Initial stochastic action to place pebbles on the surface. - `select_mother(stone_index)`: Choose which pebble to toss (m∈{1..5}). - `attempt_trial(target_indices)`: Execute the required grab/flick action on specified stone subset. - `execute_snatch()`: Final scoring throw in Stage 6. ## Playability notes - **Single-player-vs-AI:** Natural solitaire challenge (minimize total turns to reach 100 points) or competitive race against an AI bot with equivalent skill parameters. - **Group games (n>4):** Standard turn rotation handles n players seamlessly without modification. - **Physical-skill games:** Physical fine-motor grasping and eye-hand timing are modeled via parameterized sigmoid difficulty curves based on inter-stone distances. - **What is lost:** The micro-tactile feel of stone surface texture, thumb arch placement, and physical finger dexterity is replaced by numerical probability checks. ## Implementation hints Model each stage trial as an explicit function taking the 2D coordinates of resting stones and computing difficulty based on geometric clustering. First milestone: implement deterministic sequential stage transitions (L1 to L6) with a fixed pass probability before adding distance-dependent difficulty formulas.

Download as JSON

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

Download JSON
Cite this page

Five Stones (Cypriot Knucklebones & Pebble Tossing) (Cyprus). (n.d.). Games of Countries. https://gamesofcountries.com/games/cy-cyprus/pentovola

@misc{gamesofcountries-cy-cyprus-pentovola,
  title = {Five Stones (Cypriot Knucklebones & Pebble Tossing) (Cyprus)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/cy-cyprus/pentovola}
}