Formal Specification
Hide and Seek — Formal Specification
Uzbekistan · hide-seek · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: discretizedGap-fills: 6
# Bekinmachoq — AI-optimized specification
## General idea
Bekinmachoq is an asymmetric spatial hide-and-seek game where one seeker attempts to locate hidden players and race back to a designated home base to strike it while declaring their names, while hiders attempt to sneak or sprint to the home base uncaptured to secure safety. Structurally distinct from simple tag or unanchored hide-and-seek, victory and capture depend on a competitive race back to the fixed base node rather than physical body contact in the field.
## Formal model
- **Abstraction taken:** The continuous physical courtyard environment is discretized into a connected spatial graph with a designated base vertex and obstruction-based line-of-sight (LOS) visibility tables. Continuous time is discretized into sequential round-ticks where all agents choose movement or stealth actions.
- **Player/agent count:** 1 Seeker agent (), Hider agents (), where (standard formal default: ).
- **Turn model:** Simultaneous tick-based commitments during the search phase, resolved in fixed order: movement updates line-of-sight / detection checks base-touch evaluations.
## Formal rules
R1. **Setup Phase:**
- The game board is initialized as graph with base vertex .
- Initial seeker is chosen [gap-fill: by deterministic round-robin or lowest index if not chosen via counting-out].
- For ticks [gap-fill: default ticks], is locked at with vision disabled ().
- During , each hider executes up to movement steps across graph to reach a hiding vertex .
R2. **Search Phase:**
- At tick , opens vision. Each tick thereafter, all active agents simultaneously submit an action (`move(v_next)` or `stay`).
- If occupies vertex and has unobstructed line-of-sight to at vertex , is flagged as `spotted`.
- Once is `spotted`, both and are in a race condition to reach .
R3. **Base Resolution (Tura urish):**
- If enters while is `spotted` and , executes `strike_base(H_i)`: transitions to state `captured`, and 's capture rank is recorded ().
- If reaches before executes `strike_base(H_i)` (whether spotted or stealthily unspotted), executes `save_self()`: transitions to state `safe` and is removed from active board play.
R4. **Round Termination and Role Transition:**
- The round ends when every hider is either `captured` or `safe`, or when max ticks expires [gap-fill: ticks; any unspotted hider at automatically becomes `safe`].
- If at least one hider is `captured`, the hider with capture rank 1 ( captured) becomes seeker for the subsequent round.
- If all hiders become `safe` (zero captures), the current seeker remains seeker for the next round (total defeat / *kuygan*).
R5. [gap-fill] Max Round Limit & Match Winner:**
- A full match consists of rounds (default ). Each agent scores 1 point per round they achieve `safe` status and 0 points when `captured`. The seeker scores points equal to the number of captured hiders. Highest total score wins.
## Edge cases
- **Tie at the end condition:** If multiple agents finish a multi-round match with equal points, the agent with fewer total rounds served as seeker wins [gap-fill].
- **A player with no legal move:** On a graph, an agent can always execute `stay` (pass turn).
- **Repetition / cycles:** Bounded strictly by per round; stalling hiders are forced to act or automatically evaluated at timeout.
- **Simultaneous claims on the same resource or position:** If and a spotted hider enter on the exact same tick, the tie is awarded to the hider ( is `safe`), mirroring the ethnographic rule favoring the runner [gap-fill]. Multiple hiders can occupy the same graph vertex simultaneously.
- **Player elimination mid-round:** When a hider becomes `safe` or `captured`, their state is set to inactive and they take no further actions during the round; turn order and tick progression continue without disruption.
- **Resource exhaustion:** Graph nodes and movement steps are infinite across ticks; time bounded by .
- **Termination guarantee:** Each round strictly terminates within ticks; a match terminates in exactly ticks.
## State representation
- **Persistent state:**
- Graph topology and line-of-sight adjacency matrix .
- Agent roles: seeker ID , hiders .
- Match scores: vector of integers of size .
- Current round number .
- **Per-round state:**
- Current tick .
- Agent positions: map .
- Hider status: map .
- Capture sequence: ordered list of captured hider IDs.
- **Observable information:**
- Hiders observe: base location, their own position, and seeker position if within line-of-sight. Other hiders' exact positions are hidden unless sharing a vertex.
- Seeker observes: base location, own position, and any hider where .
## Action space
- `move(v)`: Move to adjacent vertex . Precondition: .
- `stay`: Remain at current vertex. Always legal.
- `strike_base(H_i)`: Executed automatically/explicitly by when at against a `spotted` hider .
- `save_self()`: Executed automatically/explicitly by when entering .
## Playability notes
- **Single-player-vs-AI:** Natural. Human can play as Seeker navigating graph with line-of-sight to spot heuristic-driven AI hiders, or as Hider trying to outmaneuver a patrolling AI seeker.
- **Group games reduction:** The game reduces smoothly to hider vs 1 seeker on a compact graph without loss of core hide-and-base-race tension.
- **Physical-skill games:** Physical sprinting speed and visual camouflage are abstracted into graph edge movement costs and line-of-sight occlusion parameters.
- **What is lost:** Physical real-world 3D stealth, physical agility, micro-distractions, and the kinesthetic excitement of real sprinting.
## Implementation hints
- Model graph as a planar grid (e.g. 8x8 grid) with obstacle blocks (walls/trees) that block visibility lines.
- Branching factor: average degree of vertices (typically 2 to 4 moves per tick).
- Avoid representation trap of making seeker omniscient; visibility masks are essential for genuine hide-and-seek decision-making.
- First milestone: Implement 1v1 discrete movement on a 6x6 grid with 1 base and 2 obstacles, verifying detection and base race logic.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Hide and Seek (Uzbekistan). (n.d.). Games of Countries. https://gamesofcountries.com/games/uz-uzbekistan/bekinmachoq
@misc{gamesofcountries-uz-uzbekistan-bekinmachoq,
title = {Hide and Seek (Uzbekistan)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/uz-uzbekistan/bekinmachoq}
}