Skip to main content
Formal Specification

Three-Flick Land Grab — Formal Specification

South Korea · dexterity-throw · Confidence: low

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: probabilistic-proxyGap-fills: 1
# Ddangttameokgi — AI-optimized specification ## General idea A turn-based 2D territory-expansion game where agents execute three consecutive flick displacement vectors from their home territory. If the third flick successfully navigates back into the agent's existing sovereign territory without going out of bounds, the enclosed polygon is annexed into the agent's permanent domain. The agent controlling the largest surface area when space is exhausted wins. ## Formal model - **Abstraction taken:** Continuous dirt surface is modeled as a 2D discrete grid M∈{0,1,2}W×H (W=50,H=50, total 2500 cells). Neutral cells =0, Agent 1 territory =1, Agent 2 territory =2. - **Player/agent count:** 2 agents (A1,A2). - **Turn model:** Sequential single-sortie turns. A successful territory annexation grants an immediate bonus sortie up to max 3 consecutive sorties per turn. ## Formal rules R1. **Initial State:** Grid dimensions 50×50. Agent 1 starts with a 5×5 corner base at (0,0); Agent 2 starts with a 5×5 corner base at (49,49). Unclaimed territory fraction U=1.0−50/2500=0.98. R2. **Sortie Action Definition:** An active agent submits a 3-step flick trajectory action 𝐀=⟨𝐯1,𝐯2,𝐯3⟩ where each 𝐯k=(rk,θk) with magnitude rk∈[2,12] and angle θk∈[0,2π). R3. **Kinematic Execution & Perturbation:** - Starting point 𝐩0∈∂Territoryi (chosen on current border). - Waypoints resolved with noise: 𝐩k=𝐩k−1+𝐯k+𝝐k, where 𝝐k∼𝒩(0,σflick2/θdex,i). R4. **Annexation Verification:** - Valid Sortie Checks: 1. All 𝐩1,𝐩2,𝐩3∈[0,W−1]×[0,H−1] (in bounds). 2. No point 𝐩k lands inside opponent territory. 3. 𝐩3∈Territoryi (must terminate inside own territory). - If valid: Polygon P=Polygon(𝐩0,𝐩1,𝐩2,𝐩3) is rasterized onto grid M. All cells inside P not owned by opponent become Territoryi. - Agent earns +1 bonus sortie on current turn (capped at 3). - If invalid: Sortie fails; grid M is unchanged, and turn ends immediately. R5. **Turn Handover:** Turn passes to the other agent. R6. **Match Termination:** The match terminates when neutral cells drop to U≤0.05 (95% of the board claimed) or turn count reaches Tmax=40. R7. **Win Condition:** The agent with the strictly higher count of controlled grid cells |Territoryi| wins. R8. [gap-fill] Area Tie-breaker:** If cell counts are exactly equal, the agent with the higher single-sortie expansion maximum wins. ## Edge cases - Tie at the end condition: Handled by R8 via peak single-sortie area; if still tied, a sudden-death 1-sortie challenge is played. - A player with no legal move: Not applicable; an agent can always attempt a minimal 3-flick sortie from any border cell. - Repetition / cycles: Bounded; every successful sortie strictly increases |Territoryi| monotonically, and turns are capped at Tmax=40. - Simultaneous claims on the same resource or position: Turns are sequential; territory claimed becomes sovereign immediately on rasterization. - Player elimination mid-round; turn order repair: If an agent's base is completely swallowed (|Territoryi|=0), that agent is immediately eliminated. - Resource exhaustion (pieces, tokens, space, time): Neutral grid cells decrease monotonically toward zero. - Termination guarantee: Monotonic cell consumption combined with hard turn limit Tmax=40 provably guarantees termination. ## State representation - **Persistent state:** Grid dimensions (W,H), agent dexterity parameters {θdex,i}, turn limit Tmax=40. - **Per-turn ephemeral state:** Territory grid matrix M, current territory areas (|Terr1|,|Terr2|), active player i, consecutive bonus sortie counter. - **Observable information:** Fully observable; perfect information on grid cell ownership and player boundaries. ## Action space - `execute_sortie(start_point, v1, v2, v3)`: Parameters: `start_point` ∈∂Terri, displacement vectors 𝐯1,𝐯2,𝐯3. Precondition: agent is active and start point is on friendly boundary. ## Playability notes - Single-player-vs-AI: Highly strategic spatial geometry game; AI computes optimal triangle vectors to maximize area while minimizing overshoot failure risks. - Group games (n>4): Natively scales to 3 or 4 players starting from 4 corners of the board. - Physical-skill games: Tactile thumb flicks and pebble friction on dirt are discretized into 2D vector perturbations. - What is lost in the adaptation: The tactile feel of scratching line borders into dirt with a sharp twig and measuring disputed land with thumb-to-pinky hand spans (*ppyeom*). ## Implementation hints Low-to-medium complexity. The grid rasterization requires standard scanline polygon filling or `matplotlib.path.Path.contains_points`. First milestone: implement the 3-flick vector chain and verify polygon rasterization on a 50×50 2D numpy array, then add the border-return check.

Download as JSON

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

Download JSON
Cite this page

Three-Flick Land Grab (South Korea). (n.d.). Games of Countries. https://gamesofcountries.com/games/kr-south-korea/ddangttameokgi

@misc{gamesofcountries-kr-south-korea-ddangttameokgi,
  title = {Three-Flick Land Grab (South Korea)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/kr-south-korea/ddangttameokgi}
}