Formal Specification
Five Bottle Caps Flicking Game — Formal Specification
Malaysia · dexterity-small · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 3
# Ceper — AI-optimized specification
## General idea
A turn-based sequential dexterity game played with five metal disc tokens. Each turn consists of an initial hand-toss flip (*menimbang*) to accumulate provisional score points, a 2D planar scatter, and a precision target flick (*menjentik*) where a designated disc must pass through a line-gap between two obstacles and strike an opponent-chosen target disc without collateral contacts.
## Formal model
- **Abstraction taken:** Hand tossing is modeled as a binomial cascade (). The planar scatter is simulated on a continuous 2D plane meters. The flick is modeled as a 2D ray collision query parameterized by agent flick accuracy and force control.
- **Player/agent count:** 2 to 4 agents taking sequential individual turns.
- **Turn model:** Turn-based round-robin. An agent executes Phase 1 (*menimbang*), generates a scatter, receives a target assignment from the next clockwise agent, and executes Phase 2 (*menjentik*).
## Formal rules
R1. State tracking:
- Global state: `scores[N]`, `current_turn_agent` , `target_score` (default 50 points).
- Turn ephemeral state: `provisional_points` , token coordinates for .
R2. Phase 1 — Menimbang (Hand-flip tossing):
- Agent tosses 5 stacked tokens:
- Back-of-hand catch count: .
- Final palm catch count: .
- Provisional points awarded:
R3. Phase 2 — Planar scatter:
- 5 token coordinates are sampled uniformly from with minimum pairwise distance .
- If any two tokens land overlapping ( where ), an *ayam* separation test is triggered: agent must execute an adjustment with success probability ; failure forfeits the turn immediately ( [gap-fill]).
R4. Target assignment:
- The defending agent (agent ) inspects the scatter geometry and selects:
- Striker disc index .
- Target disc index .
- Gap obstacle pair .
R5. Phase 3 — Menjentik (Flick execution & ray collision):
- The active agent computes flick angle from position toward .
- Actual trajectory angle: , where .
- Ray tracing resolution:
- The ray from along must cross the line segment connecting and without intersecting discs or (clearance ).
- The ray must intersect target disc within distance .
- No other disc may be struck before .
R6. Score banking & turn transition:
- If the flick satisfies all R5 criteria, the flick is successful: `scores[i] += provisional_points`, and agent earns a continuation turn ( [gap-fill]).
- If the flick fails, `provisional_points` are lost, and turn passes to agent .
R7. [gap-fill] Win condition: The first agent whose confirmed `scores[i] >= target_score` wins the match immediately upon completion of their turn.
## Edge cases
- Tie at the end condition: If multiple players reach target score, the player with the higher total score wins; if exactly tied, resolved by a single sudden-death *menimbang* round.
- A player with no legal move: In Phase 2, if no valid gap exists (e.g. discs are collinear), the defending agent must select the simplest available trajectory without obstacle constraints.
- Repetition / cycles: Impossible because scores accumulate monotonically and turn transitions ensure rotation.
- Simultaneous claims on the same resource or position: Turns are strictly sequential; physical tokens are reset per turn.
- Player elimination mid-round; turn order repair: In standard play, all players remain until target score is achieved.
- Resource exhaustion (pieces, tokens, space, time): Exactly 5 tokens; playing area is bounded.
- Termination guarantee: Non-zero probability of successful score banking () per round guarantees almost sure termination within finite turns.
## State representation
- Persistent state: `scores[N]`, `active_agent`, `target_score` (default 50).
- Ephemeral turn state: `provisional_score`, 5 token 2D positions, striker index , target index , obstacle pair .
- Observable information: Fully observable by all agents.
## Action space
- `execute_timbang()`: Triggers Phase 1 binomial evaluation.
- `choose_target(striker_idx, target_idx, obstacle_pair)`: Defensive agent action assigning target trajectory.
- `execute_flick(angle, impulse)`: Active agent action applying flick vector.
## Playability notes
- Single-player-vs-AI: Highly playable against AI that uses geometric search to assign challenging flick trajectories and heuristic angle optimization.
- Group games (n>4): Plays cleanly with up to 4 agents in round-robin sequence.
- Physical-skill games: Real-world finger dexterity is mapped cleanly to normal-distribution angle variance and geometric ray collision.
- What is lost in the adaptation: The tactile resonance of clicking metal bottle caps and micro-surface friction variations.
## Implementation hints
Easily implemented as a 2D geometric vector physics loop in 200 lines of Python or JavaScript. First milestone: implement 2D circle-segment intersection tests for gap-clearing verification before wrapping with the scoring state machine.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Five Bottle Caps Flicking Game (Malaysia). (n.d.). Games of Countries. https://gamesofcountries.com/games/my-malaysia/ceper
@misc{gamesofcountries-my-malaysia-ceper,
title = {Five Bottle Caps Flicking Game (Malaysia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/my-malaysia/ceper}
}