Skip to main content
Formal Specification

Blind Hen (Blind Man's Buff / Aklā Vista) — Formal Specification

Latvia · chase-tag · Confidence: low

← Back to game page

Implementation Suitability

Suitability: adaptation-requiredFidelity: discretizedGap-fills: 10
# Aklās vistiņas — AI-optimized specification ## General idea A sensory-asymmetric evasion and tag game where a single blindfolded catcher ("blind hen") navigates a bounded 2D arena relying exclusively on directional acoustic signals emitted by sighted runners. Sighted agents taunt and dodge within spatial bounds, balancing the risk of noise generation against the danger of cornering themselves in the arena. ## Formal model - **Abstraction taken:** Continuous physical room space and acoustic propagation are discretized into a 2D grid arena (N×M cells). Continuous audio and blind movement are abstracted into discrete time ticks where runners generate directional sound vectors (claps, footsteps, taunts) with distance-decay attenuation, and the blind agent samples a noisy directional gradient toward the sound sources. - **Player/agent count:** 1 seeker ("blind hen") and K runners (3≤K≤14). - **Turn model:** Simultaneous-tick execution (e.g. 1 tick = 0.5 seconds). All agents submit movement/action vectors simultaneously; movements resolve, acoustics propagate, and contact detections trigger. ## Formal rules R1. **Arena setup:** The arena is a bounded 2D grid of size W×H cells [gap-fill: default 15×15 grid surrounded by impassable perimeter walls]. R2. **Initial state:** Seeker starts at center (Xc,Yc) in a disoriented state (orientation randomized across 4 cardinal directions) [gap-fill]. Runners start randomly distributed in legal grid cells with distance ≥3 cells from the seeker. R3. **Action phase per tick:** - Each runner chooses one action: `move(direction)` (1 cell orthogonal/diagonal or `stay`), optionally combined with `emit_noise(level)` where level ∈{quiet,taunt,clap}. - The seeker chooses one action: `move(direction)` (1 cell orthogonal/diagonal) with outstretched reach covering their target cell and adjacent orthogonal neighbors [gap-fill]. R4. **Acoustic signal propagation:** - Moving generates base noise level 1; `stay` generates noise level 0; `taunt`/`clap` generates noise level 3. - Sound intensity decays with Manhattan distance d: signal received by seeker from runner i equals max⁡(0,leveli−⌊d/3⌋) [gap-fill]. - Seeker receives an aggregate directional vector pointing toward the weighted centroid of acoustic signals with Gaussian angular noise [gap-fill]. R5. **Boundary constraint:** Any runner whose movement attempts to step outside arena boundaries is clamped to the boundary edge and incurs an immediate forced noise level 3 penalty [gap-fill]. R6. **Tag resolution:** If the seeker's reach zone overlaps with any runner's cell at the end of tick movement, that runner is captured. R7. **Identification resolution (standard rule):** The captured runner immediately becomes the new seeker for the subsequent round. In the identification variant [gap-fill], seeker executes `guess_identity(agent_id)`: if correct, roles swap; if incorrect, the runner is teleported 3 cells away and play resumes. R8. **Round termination:** A round terminates immediately upon a successful tag and role transfer, or when tick counter reaches Tmax⁡=300 ticks without capture (triggering seeker fatigue penalty) [gap-fill]. ## Edge cases - Tie at the end condition: not applicable — a round ends on the first singular tag event; if multiple runners occupy the seeker's reach cell on the same tick, the runner with lowest agent ID is tagged [gap-fill]. - A player with no legal move: not applicable — agents always have the option to `stay` in their current cell. - Repetition / cycles: bounded by the hard tick limit Tmax⁡=300 ticks, preventing infinite evasion cycles. - Simultaneous claims on the same resource or position: multiple runners may share a cell; if the seeker steps into a multi-occupant cell, capture resolves per R8 tie-breaker. - Player elimination mid-round: not applicable — no mid-round player elimination; the game is an ongoing role-swap loop. - Resource exhaustion: not applicable — no consumable resources. - Termination guarantee: every round provably terminates in at most Tmax⁡ ticks by explicit timeout rule or earlier by tag resolution. ## State representation - Persistent state: arena dimensions (W,H), role assignment (`seeker_id`, `runner_ids`), cumulative score/rounds played. - Per-turn ephemeral state: current agent positions (xi,yi), runner acoustic emissions, noisy acoustic vector observed by seeker, tick counter t. - Observable information: - Runners have full global visibility of the arena, seeker position, and all runner positions. - Seeker observes zero visual positions; seeker's information set is strictly limited to their own coordinates, arena boundary distance, and the aggregate noisy acoustic vector. ## Action space - `move(direction)`: direction∈{North,South,East,West,NorthEast,NorthWest,SouthEast,SouthWest,Stay}. - `emit_noise(type)`: type∈{None,Quiet,Taunt,Clap}. Precondition: agent must be a runner. - `guess_identity(agent_id)`: agent_id∈runners. Precondition: agent is seeker and tag contact occurred under identification variant. ## Playability notes - Single-player-vs-AI: natural — human player plays as seeker against heuristic AI runners (who balance taunting and evasion), or as a runner dodging an acoustic-gradient-tracking AI seeker. - Group games (n>4): naturally scales across 3≤K≤14 runners on the grid. To reduce to 2 agents (1 seeker vs 1 runner), increase mandatory taunt frequency (e.g. runner must emit noise every 3 ticks) to prevent trivial stealth stalemates [gap-fill]. - Physical-skill games: blind locomotion and tactile identification abstracted into noisy acoustic vector sampling and probabilistic identity guessing. - What is lost: physical tactile feeling of clothing/face, spatial disorientation from physical spinning, and authentic vocal timbre nuances. ## Implementation hints Low-to-medium complexity. The grid movement and acoustic decay functions are straightforward to implement. The main design consideration is tuning the acoustic noise variance to ensure the seeker is neither omniscient nor hopelessly blind. First milestone: implement a 2-agent grid with deterministic acoustic signals and tag collision detection.

Download as JSON

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

Download JSON
Cite this page

Blind Hen (Blind Man's Buff / Aklā Vista) (Latvia). (n.d.). Games of Countries. https://gamesofcountries.com/games/lv-latvia/aklas-vistinas

@misc{gamesofcountries-lv-latvia-aklas-vistinas,
  title = {Blind Hen (Blind Man's Buff / Aklā Vista) (Latvia)},
  year = {n.d.},
  publisher = {Games of Countries},
  url = {https://gamesofcountries.com/games/lv-latvia/aklas-vistinas}
}