Skip to main content
Formal Specification

Stilt Racing / Bamboo and Coconut Shell Stilts — Formal Specification

Philippines · strength-endurance · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: abstractedGap-fills: 3
# Kadang-Kadang — AI-optimized specification ## General idea A simultaneous-tick sprint and balance racing game where N agents navigate a straight 1D track of length L=50 meters on stilts. Each tick, an agent selects a locomotion speed effort; higher speeds produce larger forward spatial displacements but exponentially increase the stochastic probability of falling off the stilts (which incurs a multi-tick remount penalty). The first agent to cross the finish line mounted on stilts wins. ## Formal model - **Abstraction taken:** The complex biomechanical coupling between toe grip, rope tension, and upper body center of gravity is modeled as an internal balance state B∈[0.0,1.0] and a stride power action u∈[1,4]. Higher stride power yields faster position progression but accelerates balance decay. - **Player/agent count:** 2 to 6 agents (P0..Pk−1). - **Turn model:** Simultaneous discrete execution ticks (Δt=1 tick). ## Formal rules R1. Setup: - Track length L=50.0 meters. - Each agent begins at position xi=0.0, velocity vi=0, balance Bi=1.0, and status `MOUNTED`. R2. Stride Action & State Updates: - On each tick t, every mounted agent submits an action ui∈{1,2,3,4} (stride speed gear): - Nominal forward displacement: Δxi=0.5×ui×speed_factor (meters) [gap-fill]. - Balance drain: ΔBi=−0.04×ui2. - Natural balance recovery: Bi←min⁡(1.0,Bi+0.05). - Updated balance: Bi←max⁡(0.0,Bi+ΔBi). R3. Balance Check & Fall Event: - Fall probability per tick: P(fall)=max⁡(0.0,(1.0−Bi)×(0.20×ui)×(1.0−balance_skill)). - If fall occurs (`fall == true`): - Agent status transitions to `DISMOUNTED`. - Agent is immobilized for a remount penalty duration of Tremount=4 ticks [gap-fill]. - Bi is restored to 0.80 upon completion of remount penalty. R4. Remount Progression: - Dismounted agents decrement penalty counter: `remount_counter -= 1`. - When `remount_counter == 0`, status returns to `MOUNTED`. R5. Win Condition & Race End: - An agent wins immediately when xi≥L while status is `MOUNTED`. - If multiple agents cross on the same tick, the agent with the higher coordinate xi wins. - Race hard time cap: Tmax⁡=150 ticks. ## Edge cases - Tie at the end condition: If two agents cross L on the same tick with identical xi, the agent with higher remaining balance Bi wins [gap-fill]. - A player with no legal move: In mounted state, u∈{1,2,3,4} is always available; in dismounted state, agent automatically executes wait/remount. - Repetition / cycles: Monotonic forward displacement (xi(t+1)≥xi(t)) ensures strictly non-decreasing race progress. - Simultaneous claims on the same resource or position: Agents run in independent parallel virtual lanes; collisions between lanes are ignored. - Player elimination mid-round: Falling incurs a time delay rather than permanent player elimination. - Resource exhaustion: Not applicable. - Termination guarantee: Non-negative displacement and maximum tick ceiling Tmax⁡=150 guarantee finite termination. ## State representation - Persistent state: - `positions`: array of k floats (xi∈[0.0..50.0]). - `balance_states`: array of k floats (Bi∈[0.0..1.0]). - `agent_statuses`: array of k enums (`MOUNTED`, `DISMOUNTED`). - `remount_timers`: array of k integers. - `current_tick`: integer 0..150. - Observable information: Fully observable public state (all racer positions and mounted statuses are visible). ## Action space - `stride(gear)`: integer u∈{1,2,3,4} (when mounted). ## Playability notes - Single-player-vs-AI: Highly playable as a classic drag-race style risk/speed management challenge against AI opponents with diverse risk profiles. - Group games (n>4): Scales seamlessly to 6 or more agents without increasing algorithm complexity (O(N) tick cost). - Physical-skill games: Dynamic core balance and foot stability are modeled as a controllable state resource (B). - What is lost in the adaptation: The tactile foot pain of gripping coconut ropes between the bare toes and physical bamboo peg vibrations. ## Implementation hints Simple 1D vector kinematics with stochastic penalty interrupts. Extremely fast rollout capability for Monte Carlo tree search. First milestone: Implement 1-agent runner to tune balance decay parameters and find optimal steady-state gear strategies.

Download as JSON

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

Download JSON
Cite this page

Stilt Racing / Bamboo and Coconut Shell Stilts (Philippines). (n.d.). Games of Countries. https://gamesofcountries.com/games/ph-philippines/kadang-kadang

@misc{gamesofcountries-ph-philippines-kadang-kadang,
  title = {Stilt Racing / Bamboo and Coconut Shell Stilts (Philippines)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/ph-philippines/kadang-kadang}
}