Formal Specification
Emperor, Emperor, Master — Formal Specification
Bosnia & Herzegovina · chase-tag · Confidence: low
← Back to game pageImplementation Suitability
Suitability: engine-readyFidelity: discretizedGap-fills: 2
# Care, care, gospodare — AI-optimized specification
## General idea
An asymmetric 1D distance race where Runner agents attempt to advance from to by asking the Emperor for movement grants. The Emperor, facing away, observes only the noisy auditory signal of the asking agent and assigns step counts and types. Runners can disguise their voice identity to manipulate the Emperor into granting large steps, while the Emperor seeks to identify leaders and grant small or penalizing steps.
## Formal model
- **Abstraction taken:** The 1D track is defined on . Step types are discretized into fixed forward displacements: . Voice disguise is modeled as a categorical identity obfuscation with a deception detection check.
- **Player/agent count:** 1 Emperor () and Runners (, standard: ).
- **Turn model:** Strictly sequential runner inquiries within an iterative round cycle.
## Formal rules
R1. **Setup:** Track length . Runner positions for all . Runner status . Turn order among runners is fixed .
R2. **Runner action:** On runner 's turn:
- If : Runner must ask to leave the well (`ask_unfreeze()`). Emperor permits unfreeze with probability ; if permitted, , turn ends.
- If : Runner executes `request_time(disguise_target)`, choosing to speak either naturally () or feign the identity of runner .
R3. **Emperor action:**
- Emperor receives perceived identity . If disguised, Emperor detects deception with probability .
- If deception is detected, Emperor knows true identity is ; otherwise Emperor believes identity is .
- Emperor selects a step award action `grant_steps(count, step_type)` where and .
R4. **Movement and state update:**
- If : .
- Otherwise: Displacement .
- Position update: .
- [gap-fill] Overstepping penalty: If a runner attempts to sneak extra distance beyond the grant, detection with sends the runner back to .
R5. **Win condition:** The game ends immediately when any runner reaches . That runner wins the match.
## Edge cases
- **Tie at the end condition:** [gap-fill] Because runners take sequential turns and victory triggers immediately upon crossing , ties are structurally impossible.
- **A player with no legal move:** Not applicable; runners can always execute a request action, and the Emperor can always grant a valid step tuple.
- **Repetition / cycles:** At least one non-frozen runner is granted positive displacement across cyclic turns; total distance monotonically increases toward .
- **Simultaneous claims on the same resource:** Turn structure is strictly sequential; only one runner acts and moves at any given time.
- **Player elimination mid-round / turn order repair:** Not applicable; all runners stay active throughout the race.
- **Resource exhaustion:** Invariant; track length is fixed and movement grants are infinite.
- **Termination guarantee:** Because the minimum forward progress is unit on non-well turns and the Emperor cannot permanently freeze all runners (unfreeze probability ), the maximum runner position strictly reaches in finite turns.
## State representation
- **Persistent state:** `runner_positions: Dict[AgentID, int]`, `runner_frozen: Dict[AgentID, bool]`, `active_runner_index: int`, `finish_line: int`.
- **Per-turn ephemeral state:** `disguise_choice: AgentID`, `emperor_belief: AgentID`, `granted_count: int`, `granted_type: str`.
- **Observable information set:**
- Runners observe: Full history of past step grants and their own positions; can estimate opponent distances.
- Emperor observes: Voice audio signal (perceived ID ); does NOT observe true numerical positions directly (must deduce them from past grants).
## Action space
- Runner: `request_steps(feign_id: Optional[AgentID])`.
- Emperor: `grant(count: int, step_type: Enum{ANT, MOUSE, CAT, HORSE, GIANT, WELL})`.
## Playability notes
- **Single-player-vs-AI:** Natural in both configurations: Player as Runner (optimizing bluffing vs AI Emperor) or Player as Emperor (deducing true identities and managing runner spreads).
- **Group reduction:** Multi-runner group () reduces cleanly to 1 Emperor vs 2 Runners without altering bluffing or pacing mechanics.
- **Physical-skill games:** Physical gait variations (tiny heel-to-toe ant steps vs giant leaps) are mapped to discrete integer step multipliers.
- **What is lost:** The comedic joy of watching children awkwardly waddle with heels touching toes or spin around in "umbrella" circles is formalized into distance increments.
## Implementation hints
- Very low computational footprint; straightforward discrete Markov decision process with partial observability for the Emperor agent.
- Obvious trap: Allowing the Emperor full observability of runner coordinates, which destroys the incentive for voice disguises.
- Suggested first milestone: Implement the 1 Emperor vs 2 Runners state machine with a heuristic belief model for the Emperor.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Emperor, Emperor, Master (Bosnia & Herzegovina). (n.d.). Games of Countries. https://gamesofcountries.com/games/ba-bosnia-herzegovina/care-care-gospodare
@misc{gamesofcountries-ba-bosnia-herzegovina-care-care-gospodare,
title = {Emperor, Emperor, Master (Bosnia & Herzegovina)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/ba-bosnia-herzegovina/care-care-gospodare}
}