Skip to main content
Formal Specification

The Drop-Handkerchief / Rotten Egg — Formal Specification

Germany · chase-tag · Confidence: medium

← Back to game page

Implementation Suitability

Suitability: engine-readyFidelity: discretizedGap-fills: 2
# Der Plumpsack geht um / Faules Ei — AI-optimized specification ## General idea A circle-topology chase and vigilance game featuring hidden token placement under perceptual constraints. N−1 seated agents form a directed ring graph, facing inward without visual sightlines to their rear quadrant. One active Runner agent circles the exterior, choosing when and behind whom to drop a discrete token (*Plumpsack*). Seated agents poll limited sensory/tactile checks to detect dropped tokens behind them. If undetected, the Runner laps the ring and claims a penalty point (*Faules Ei*); if detected, a synchronized directional sprint race ensues to claim the single open ring node. ## Formal model - **Abstraction taken:** The physical circular geometry is discretized into a directed cycle graph CN=(V,E) with nodes 0,1,…,N−1 arranged in clockwise order. Node 0 represents the center jail (*Faules Ei* zone). Visual forward focus is modeled by zeroing rearward line-of-sight: seated agents at node v observe events only within the center circle, receiving a stochastic tactile/auditory signal Psense∈[0,1] when the sack is dropped at node v. - **Player/agent count:** N agents (1 Runner prun, N−1 Seated agents). - **Turn model:** Synchronized discrete ticks Δt=0.2 s. ## Formal rules R1. **Topology & Roles:** Ring nodes 1…N−1 are occupied by seated agents. Node 0 is the center penalty zone. One agent is designated `Runner`, positioned on the exterior perimeter at node vrun=1, holding the token (`HasToken = True`). R2. **Runner Movement & Drop Action:** On each tick t while `HasToken = True`: - Runner submits `move_step(direction = +1)` (moving clockwise along outer perimeter at velocity vjog=1 node/tick) or `drop_token(target_node)`. - If `drop_token` is executed at node v, the token state becomes `DroppedAt(v)`, and `HasToken = False`. R3. **Seated Agent Vigilance & Detection:** - On each tick, each seated agent i at node vi submits `poll_rear_sense()`: - If `DroppedAt(v_i)` is true, agent receives a sensory cue with detection probability Pdetect=0.40 on the drop tick, rising by +0.20 per subsequent tick elapsed. - Upon detection, agent i transitions to `Chaser`, grabs the token, and vacates node vi (marking vi as `Vacant`). R4. **Unaware Lap Resolution (*Faules Ei*):** - If the token remains undetected and the Runner completes a full lap (reaching node v where the token was dropped), the Runner submits action `tap_unaware_player(i)`. - Agent i transitions to state `RottenEgg` and is relocated to node 0 (center). - Runner picks up the token and immediately continues as Runner for the next cycle. R5. **Pursuit Sprint Race:** When agent i becomes `Chaser`: - Both Runner and Chaser sprint clockwise along the perimeter graph toward the `Vacant` node vi. - Runner sprint speed: vsprint,run∼𝒩(μrun,σ2). - Chaser sprint speed: vsprint,chase∼𝒩(μchase,σ2). - Distance to vacant node: drun=(vi−vrun_pos)mod(N−1), while dchase=N−1. - **Outcome A (Runner Safe):** If Runner reaches vi before Chaser can tag them (satisfying treach,run<treach,chase), Runner occupies vi and becomes `Seated`. Chaser becomes the new `Runner` for the next round. - **Outcome B (Chaser Tag):** If Chaser intercepts the Runner before vi, Runner remains `Runner` for the next round, and Chaser returns to sit at vi. R6. [gap-fill] Jail Release Mechanism:** A player at node 0 (`RottenEgg`) is released back to a seated node whenever a new player enters node 0, or after Kjail=3 completed rounds. R7. [gap-fill] Match Termination & Scoring:** The match runs for Rtotal=20 completed drop cycles. Agents are scored by +1 for winning sprint races, −1 for each stint as `RottenEgg`. The agent with the highest net score wins. ## Edge cases - Tie at the end condition: Tied net scores are broken by fewest total stints as Rotten Egg; if still tied, a 1v1 sprint shootout around the perimeter is conducted. - A player with no legal move: Seated players continue polling sensory cues; Rotten Egg agents wait at node 0. - Repetition / cycles: Bounded strictly by fixed total cycle count Rtotal=20 in R7. - Simultaneous claims on the same resource or position: The single vacant node vi is claimed by whichever agent arrives first (tarrival tie broken in favor of Chaser). - Player elimination mid-round; turn order repair: An agent disconnection removes their node from the ring graph, contracting the perimeter cycle CN−1. - Resource exhaustion (pieces, tokens, space, time): The single physical sack is tracked continuously between drops, retrievals, and handoffs. - Termination guarantee: why the game provably ends: Guaranteed because each drop cycle terminates in finite ticks (either via aware sprint race or unaware full lap tap), bounded by total cycle cap Rtotal. ## State representation - **Persistent state:** Ring node count N, ring graph adjacency, score table {Scorei}, penalty jail occupancy list. - **Per-player state:** Node location vi, role (`Runner`, `Seated`, `Chaser`, `RottenEgg`), sensory awareness skill parameter. - **Per-turn ephemeral state:** Token state (`Held(runner)`, `DroppedAt(node)`), perimeter runner coordinate, pursuit delta distance. - **Observable information:** Imperfect information; seated agents observe center circle and other seated agents, but their own rear node is hidden until polled sensory cue triggers. Runner observes all player positions and token state. ## Action space - `move_step(direction)`: parameter ∈{+1}. Precondition: agent is active Runner or sprinting Chaser. - `drop_token(target_node)`: parameter v∈{1…N−1}. Precondition: agent is Runner at node v with token. - `poll_rear_sense()`: active sensory check. Precondition: agent is Seated. - `tap_unaware_player(target_id)`: tag action upon completing full lap. ## Playability notes - Single-player-vs-AI: Natural; 1 human player as Runner or Seated ring member with AI agents modeling stochastic reaction latencies. - Group games (n>4): Mass group circle reduces cleanly to a 2-agent pursuit race (Runner vs Target Chaser) while treating passive circle members as stochastic sensory filters. - Physical-skill games: Physical auditory sensing and peripheral vigilance are modeled as the detection probability hazard rate Pdetect(t). - What is lost in the adaptation: The live vocal singing of the traditional German rhyme, clapping rhythm, and physical breathless giggles of the schoolyard. ## Implementation hints Discrete cycle graph with Poisson/Bernoulli detection triggers. Milestone 1: implement directed cycle graph with Runner drop-and-lap logic, verify that sprint equations correctly assign seat ownership and *Faules Ei* transitions.

Download as JSON

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

Download JSON
Cite this page

The Drop-Handkerchief / Rotten Egg (Germany). (n.d.). Games of Countries. https://gamesofcountries.com/games/de-germany/plumpsack

@misc{gamesofcountries-de-germany-plumpsack,
  title = {The Drop-Handkerchief / Rotten Egg (Germany)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/de-germany/plumpsack}
}