Skip to main content
Formal Specification

Salty Fish (One, Two, Three, Salty Fish!) — Formal Specification

Israel · chase-tag · Confidence: high

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 3
# Dag Maluach — AI-optimized specification ## General idea Dag Maluach is a spatial movement-coordination game of discrete, asynchronous ticks. Runners must advance across a 1D corridor toward a stationary Caller (the Fish). The Caller can periodically and unpredictably trigger a "freeze" state, during which any runner whose velocity is non-zero is detected and penalized. The strategic core is managing momentum: accelerating quickly versus being able to decelerate to a dead stop on short notice. ## Formal model - **Abstraction taken:** The playing field is a 1D line of length D=15 meters. The Caller is positioned at x=0, and N runners start at x=15. The game runs in discrete time ticks t=0,1,2... of length 500 ms. - **Player/agent count:** 1 Caller, 2 to 8 Runners. - **Turn model:** Simultaneous actions. Each tick, all runners commit a movement speed, and the Caller commits their chanting state. ## Formal rules R1. **State variables per runner:** `position` (x∈[0,15]), `velocity` (v∈[0,2] meters per tick). R2. **The Chant Cycle:** The Caller operates a state machine: `state \in [CHANTING, SPINNING, OBSERVED]`. - `CHANTING` lasts for a random integer number of ticks C∈[3,8]. - At the final tick of `CHANTING`, the Caller transitions to `SPINNING` and then `OBSERVED`. R3. **Runner Movement:** Each tick during `CHANTING`, a runner selects an acceleration a∈[−1,1]. Their velocity updates vt+1=clamp(vt+a,0,2), and their position updates xt+1=clamp(xt−vt+1,0,15). R4. **Deceleration & The Freeze Check:** - When the Caller transitions to `SPINNING` (which is public knowledge at tick start), runners must decelerate. - On the subsequent tick of `OBSERVED`, if any runner has v>0, they are "detected". - [gap-fill] A detected runner has their x position reset immediately to 15, and their velocity v reset to 0. R5. **Touching the Base:** - If a runner reaches x=0 with v≥0, they successfully touch the Fish. - [gap-fill] Once x=0 is reached, the state transitions to a simple pursuit phase: the successful runner must run back to x=15. The Fish is placed at x=0 and pursues them with velocity vfish=1.5. - If the Fish catches the runner (intersects their position), the runner becomes the new Fish for the next round. If the runner reaches 15, they win the match. R6. **Win Condition:** The game terminates when a runner successfully completes the escape or the Fish tags them, initiating a new round. ## Edge cases - **Tie at the end condition:** If two runners touch the base on the exact same tick, the one with the lower velocity is declared the primary toucher. - **A player with no legal move:** Not applicable; players can always accelerate or decelerate. - **Repetition / cycles:** If players repeatedly get reset, the Caller's chant length C decreases by 1 tick per cycle to make the game progress faster, guaranteeing eventual termination. - **Simultaneous claims on the same resource:** Not applicable; runners are independent. - **Player elimination mid-round:** If a runner drops out, they are removed from the vector. - **Resource exhaustion:** Not applicable. - **Termination guarantee:** Guaranteed by R5 (the Fish eventually catches someone or they escape) and the decreasing chant length (R4 cycle mitigation). ## State representation - **Persistent state:** - `runner_states`: Map of `runner_id` to `(x_pos, velocity, is_detected)`. - `fish_id`: ID of the agent acting as the Caller. - `caller_state`: Enum `[CHANTING, SPINNING, OBSERVED]`. - `chant_ticks_remaining`: Integer. - **Per-tick ephemeral state:** - `active_accelerations`: Map of `runner_id` to float. ## Action space - `accelerate(a)`: Set acceleration value for the current tick a∈[−1,1]. Precondition: runner is not currently detected and is active. - `turn_around()`: Caller action to trigger the spin transition. (Can be automated by AI heuristic). ## Playability notes - **Single-player-vs-AI:** Excellent. The human player controls a runner using keyboard keys to accelerate/decelerate, while AI runners manage their momentum defensively, and the AI Caller randomizes its chant length to catch the player. - **Group games (n>4):** Generalizes beautifully. - **Physical-skill games:** [gap-fill] The physical skill of balance and rapid muscle contraction is represented by the momentum decay and the requirement to bring velocity to exactly 0 before the `OBSERVED` tick. - **What is lost in the adaptation:** The physical tension of holding an awkward, unstable body posture (e.g., standing on one foot or bent over) is replaced by numerical velocity thresholds. ## Implementation hints - Implement the simultaneous-tick engine. - Build the Caller state machine and ensure the transition to `OBSERVED` correctly detects and resets any agent with v>0. - Write a basic AI for runners: "If state is CHANTING, accelerate to max; if state is SPINNING, decelerate to 0 immediately." - Calibrate the deceleration rate so players must anticipate the Caller's spin.

Download as JSON

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

Download JSON
Cite this page

Salty Fish (One, Two, Three, Salty Fish!) (Israel). (n.d.). Games of Countries. https://gamesofcountries.com/games/il-israel/dag-maluach

@misc{gamesofcountries-il-israel-dag-maluach,
  title = {Salty Fish (One, Two, Three, Salty Fish!) (Israel)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/il-israel/dag-maluach}
}