Formal Specification
Red Light, Stop! — Formal Specification
Denmark · chase-tag · Confidence: high
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: discretizedGap-fills: 3
# Rødt lys, stop! — AI-optimized specification
## General idea
A stop-and-go racing and motor-inhibition game. A designated leader stands at the finish line with back turned, reciting a traditional countdown phrase at variable tempos. Runners advance across the field during the recitation; the instant the phrase terminates, the leader spins around, and all runners must freeze completely motionless. Any runner caught in motion or failing a balance check is returned to the starting line.
## Formal model
- **Abstraction taken:** Continuous 2D field running is discretized into 1D track coordinates . The auditory recitation is modeled as an interval sampled from a distribution selected by the Leader's chosen tempo. Kinetic inertia and deceleration are modeled as a stochastic balance stability check upon freeze transition.
- **Player/agent count:** 4 to 20 agents (1 Leader, Runners).
- **Turn model:** Repeated discrete phases: (1) Secret Pace Selection, (2) Simultaneous Sprint Execution, (3) Inspection & Freeze Evaluation.
## Formal rules
R1. **Track Setup:** Track length is meters. Leader is stationed at . All Runners start at .
R2. **Pace Selection Phase:** The Leader submits action `select_pace(tempo)` where . This determines the random recitation duration .
R3. **Sprint Allocation:** Each Runner (without observing exact ) submits a target sprint duration and velocity . The runner's prospective displacement is .
R4. **Turnaround & Detection:** At exact time , the Leader turns to face the field:
- *Overrun Detection:* If , Runner is actively moving when the Leader turns and is detected.
- [gap-fill] Kinetic Wobble Check:* If , the runner stopped in time but must pass an inertia stability check: , where is the runner's balance skill and is deceleration decay. If wobble triggers, runner is detected.
R5. **Penalty Enforcement:** Any detected runner is sent back to the baseline (). Undetected runners successfully update their position: .
R6. **Victory Condition:** The first runner to reach while remaining undetected during the inspection wins the round and becomes the new Leader for the next game.
R7. [gap-fill] Simultaneous Crossing Resolution:** If multiple runners reach in the same cycle, the runner with the largest total displacement wins; if tied, lowest agent index wins.
R8. [gap-fill] Maximum Cycles Cap:** If no runner reaches within recitation cycles, the runner with the maximum coordinate is declared the winner.
## Edge cases
- Tie at the end condition: Resolved by R7 (displacement margin and index order).
- A player with no legal move: `sprint(duration=0, velocity=0)` is always a legal conservative action.
- Repetition / cycles: Resetting caught runners to 0 can theoretically cycle; bounded strictly by the global cycle limit in R8.
- Simultaneous claims on the same resource or position: Runners operate on independent parallel track lanes; lane collisions do not occur.
- Player elimination mid-round; turn order repair: If a runner disconnects, they are removed from the active array. If the Leader disconnects, the runner closest to becomes the new Leader.
- Resource exhaustion (pieces, tokens, space, time): Spatial track is fixed and reusable; time is bounded by .
- Termination guarantee: why the game provably ends: Guaranteed by finite cycle limit combined with positive probability that careful runners advance monotonically without penalty.
## State representation
- **Persistent state:** Track length , max speed , player roster, runner balance skills .
- **Per-round ephemeral state:** Leader tempo choice, realized duration , runner sprint parameters , runner positions , cycle counter .
- **Observable information:**
- Runners observe: Current positions of all runners, Leader identity, cycle history.
- Hidden from Runners: Exact duration during the current sprint allocation.
- Leader observes: Positions of all runners at inspection time, motion status, and wobble results.
## Action space
- `select_pace(tempo)`: Leader action. Parameter .
- `sprint(duration, velocity)`: Runner action. Parameters seconds, m/s.
## Playability notes
- Single-player-vs-AI: Excellent; human can play as Leader timing turns to catch AI agents, or as a Runner calculating risk-reward stopping intervals against an AI Leader.
- Group games (n>4): Easily reduced to 2 agents (1 Leader vs 1 Runner) without degrading the timing and risk-estimation dilemma.
- Physical-skill games: Physical balance and motor braking are modeled via velocity-dependent kinetic wobble probabilities.
- What is lost in the adaptation: The somatic strain of freezing in awkward contorted body positions, eye-contact deception, and infectious group giggling.
## Implementation hints
Low complexity. Can be implemented as a clean tickless discrete-event model: sample , collect runner tuples, evaluate detection and wobble booleans, and update . First milestone: implement 1 Leader vs 3 AI runners with basic risk-taking heuristics and verify baseline reset logic.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Red Light, Stop! (Denmark). (n.d.). Games of Countries. https://gamesofcountries.com/games/dk-denmark/roedt-lys-stop
@misc{gamesofcountries-dk-denmark-roedt-lys-stop,
title = {Red Light, Stop! (Denmark)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/dk-denmark/roedt-lys-stop}
}