Skip to main content
Formal Specification

Five Stones (Knucklebones / Jacks) — Formal Specification

Latvia · dexterity-small · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 6
# Pieci akmentiņi — AI-optimized specification ## General idea A turn-based sequential dexterity game played with five stones. Players attempt an escalating 6-stage routine of tossing, floor scooping, arch flicking, and back-of-hand catching; committing a fault immediately ends the turn, requiring the player to resume from the failed stage on their next turn until one player completes the entire sequence. ## Formal model - **Abstraction taken:** Physical hand-eye coordination, micro-timing of airborne stone catching, and finger agility are abstracted into a stochastic stage-progression ladder. Each of the 6 canonical stages is assigned a discrete difficulty index Ds∈[0.1,0.9]. The execution outcome of stage s is sampled from Bernoulli distribution P(success|s,skill)=11+exp⁡(−k⋅(skill−Ds)). - **Player/agent count:** 1 to 6 agents (1≤N≤6). - **Turn model:** Strictly sequential, cyclic turn order; an agent retains the turn upon success and yields upon fault. ## Formal rules R1. **Stage definitions and difficulty ladder:** - Stage 1 (*Vieniņi* - Singles): 4 single tosses and pickups (D1=0.20) [gap-fill]. - Stage 2 (*Diviņi* - Pairs): 2 pair tosses and pickups (D2=0.35) [gap-fill]. - Stage 3 (*Trijiņi* - Threes): 1 triple pickup, 1 single pickup (D3=0.50) [gap-fill]. - Stage 4 (*Četriņi* - Fours): 1 quadruple drop and snatch (D4=0.65) [gap-fill]. - Stage 5 (*Tilts* - Arch/Bridge): flick 4 stones through non-throwing hand arch (D5=0.75) [gap-fill]. - Stage 6 (*Virsroka* - Back of Hand): toss 5 stones, catch on back of hand, toss and snatch out of air (D6=0.85) [gap-fill]. R2. **Initial state:** All agents begin at Stage progress Si=1. Turn order is P1,P2,…,PN. R3. **Turn execution:** - On a turn, the active player attempts their current stage Si. - The outcome is sampled: Y∼Bernoulli(P(success|Si,skilli)). R4. **Advancement and fault resolution:** - **Success (Y=1):** - If Si<6, progress updates to Si←Si+1, and the agent immediately attempts stage Si+1 (retaining the turn). - If Si=6, the agent completes the ladder and wins immediately. - **Fault (Y=0):** - Turn ends immediately. Progress remains at Si. Turn passes to the next player in rotation. R5. **Win condition:** The first player to successfully complete Stage 6 wins the match. ## Edge cases - Tie at the end condition: not applicable — turns are strictly sequential; the first agent to execute Stage 6 successfully wins instantly. - A player with no legal move: not applicable — attempting the active stage is always the singular legal action. - Repetition / cycles: bounded by the absorbing state (Stage 6 completion); since P(success)>0 for all stages under any valid skill, every player almost surely advances. - Simultaneous claims on the same resource or position: not applicable — turns are strictly sequential. - Player elimination mid-round: not applicable — all players remain active until a winner emerges. - Resource exhaustion: not applicable — 5 abstract stones used perpetually. - Termination guarantee: modeled as a finite absorbing Markov chain with absorbing state S=6; transition probabilities to next state are strictly positive, ensuring finite termination with probability 1. ## State representation - Persistent state: player IDs, player skill ratings, stage progress array (S1…SN), turn rotation list. - Per-turn ephemeral state: active player ID, sampled stage outcome Y, current turn attempt counter. - Observable information: full public observability — all player stages, attempt histories, and skill parameters are public. ## Action space - `attempt_stage()`: execute current stage routine. Precondition: agent is active turn holder. ## Playability notes - Single-player-vs-AI: natural — solitaire speedrun mode (minimize total turns to complete all 6 stages) or multi-agent race against AI opponents. - Group games (n>4): natively scales from 1 to 6 players; turn rotation order generalizes losslessly without altering the individual stage-climbing problem. - Physical-skill games: manual stone catching reflexes, hand orientation, and spatial pebble scatter are abstracted into the calibrated difficulty ladder D1…D6. - What is lost: physical sensation of smooth river pebbles, auditory clatter on wood, and delicate fingertip arch coordination. ## Implementation hints Extremely low complexity. Pure discrete Markov state ladder with simple Bernoulli sampling. First milestone: implement 6-stage state machine, Bernoulli resolver, and sequential turn loop.

Download as JSON

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

Download JSON
Cite this page

Five Stones (Knucklebones / Jacks) (Latvia). (n.d.). Games of Countries. https://gamesofcountries.com/games/lv-latvia/pieci-akmentini

@misc{gamesofcountries-lv-latvia-pieci-akmentini,
  title = {Five Stones (Knucklebones / Jacks) (Latvia)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/lv-latvia/pieci-akmentini}
}