Formal Specification
Blind Billy-Goat / Blind Goat Tag — Formal Specification
Kazakhstan · chase-tag · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: discretizedGap-fills: 6
# Soqyrteke — AI-optimized specification
## General idea
Soqyrteke is a traditional blindfolded circle-tag and tactile identification game. One child is blindfolded as the "blind goat" (*soqyrteke*) in the center of a circular boundary while evasive players circle around, chanting taunts and lightly tapping the seeker. When the blindfolded seeker captures an evader, they must correctly identify the captive using tactile feel or voice cues alone to pass on the tagger role.
## Formal model
- **Abstraction taken:** The continuous 2D court and physical tactile probing are discretized into a bounded circular arena of radius metres. [gap-fill] The blindfolded seeker moves via noisy heading estimates and executes grab sweeps; upon contact, the identification challenge is modeled as a Bayesian classification problem over the set of active players, parameterized by seeker tactile discrimination skill and player distinctiveness traits (height, texture class).
- **Player/agent count:** agents (default ; reducible to 2 agents: 1 Seeker vs 1 Evader with boundary constraints).
- **Turn model:** Simultaneous tick-based positional movement, followed by conditional sequential identification checks.
## Formal rules
R1. **Court Boundaries and Role Assignment:**
- Court layout: Continuous 2D disc of radius m centered at .
- Exactly 1 agent is assigned the role of Seeker (); agents are assigned the role of Evaders ().
- Each evader has public static profile features .
- Stepping outside the radius () is an out-of-bounds foul; a fouled evader is immediately caught and tested for identification.
R2. **Movement & Evasion Tick:**
- Each tick, all evaders simultaneously choose velocity vectors such that m/s and .
- The blindfolded Seeker receives a noisy acoustic vector (where ) and submits a movement/grab vector with reach envelope radius m.
R3. **Capture and Freeze:**
- If for any evader :
- Evader is captured.
- Play pauses, and the game enters the **Tactile Identification Phase**.
R4. **Tactile Identification Phase:**
- The Seeker receives a noisy feature vector , where noise variance is inversely proportional to seeker skill: .
- The Seeker submits a guess `identify(candidate_id)`.
- **Correct Guess (`candidate_id == j`):**
- The blindfold is transferred: evader becomes the new Seeker for the next round.
- Seeker earns +1 point and joins the evader pool.
- Court positions are reset to the starting configuration.
- **Incorrect Guess (`candidate_id != j`):**
- Evader is released.
- Seeker receives 0 points and must continue seeking.
- [gap-fill] If the Seeker fails 3 consecutive identifications on distinct captures, the role automatically transfers to the most recently caught evader.
R5. **End Condition and Match Winner:**
- A match lasts for a fixed number of successful role transitions (e.g. rounds) or a tick budget of 500 ticks.
- The player who accumulated the most points (successful seeker escapes/identifications) wins the match.
- [gap-fill] In case of tied scores at match end, joint victory is awarded.
## Edge cases
- **Tie at the end condition:** Explicitly resolved by R5 [gap-fill] (most successful identification escapes wins; ties share joint victory).
- **A player with no legal move:** Not applicable; movement vectors within bounds and identification guesses are always well-defined.
- **Repetition / cycles (and the loop-breaking rule):** An inept seeker could cycle indefinitely without identifying captives; resolved by R4 [gap-fill] 3-strike rule automatically transferring the role.
- **Simultaneous claims on the same resource or position:** If seeker grab radius covers multiple evaders simultaneously, the closest evader is selected as the captive.
- **Player elimination mid-round; turn order repair:** Players are never eliminated; roles cycle dynamically between seeker and evader.
- **Resource exhaustion (pieces, tokens, space, time):** Spatial court bounds and player counts remain constant.
- **Termination guarantee: why the game provably ends:** Every round ends either upon a correct identification or via the 3-strike forced transfer rule in R4 [gap-fill]; the match terminates deterministically after rounds.
## State representation
- **Persistent state:**
- `seeker_id`: Agent ID currently in seeker role.
- `evader_ids`: List of agent IDs in evader pool.
- `player_scores`: Map of `agent_id -> int`.
- `consecutive_failed_ids`: Integer .
- `completed_rounds`: Integer .
- **Per-tick ephemeral state:**
- `player_positions`: Map of `agent_id -> (x, y)`.
- `current_captive_id`: Nullable agent ID.
- `noisy_cue`: Float vector.
- **Observable information:** Asymmetric / hidden information. Evaders observe full 2D ground truth positions. The Seeker observes only noisy directional acoustic cues and noisy tactile feature vectors during identification.
## Action space
- `move_evader(velocity_vector)`: Evader action; m/s.
- `seek_and_grab(direction, speed)`: Seeker action; speed m/s.
- `identify(candidate_id)`: Seeker action during identification phase; candidate .
## Playability notes
- **Single-player-vs-AI:** Natural. Human player as Seeker (receiving audio/visual noise cues) or as Evader dodging an AI seeker.
- **Group games (n>4):** Reduces cleanly to 2 agents (1 Seeker, 1 Evader in a bounded disc arena), with identification reduced to a binary bluff/feature test.
- **Physical-skill games:** Modeled via seeker tactile discrimination skill and auditory tracking accuracy.
- **What is lost in the adaptation:** Physical fabric tactile sensations under fingertips, auditory spatial localization of giggling children, and playful dodge body gestures.
## Implementation hints
- Implement the 2D kinematic engine with circular clamping ().
- Model the identification step as minimum Euclidean distance matching between observed noisy vector and known static profiles .
- First milestone: Implement 2D circle bounding movement, seeker proximity collision detection, and tactile feature classification.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Blind Billy-Goat / Blind Goat Tag (Kazakhstan). (n.d.). Games of Countries. https://gamesofcountries.com/games/kz-kazakhstan/soqyrteke
@misc{gamesofcountries-kz-kazakhstan-soqyrteke,
title = {Blind Billy-Goat / Blind Goat Tag (Kazakhstan)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/kz-kazakhstan/soqyrteke}
}