Formal Specification
Goat Pulling — Formal Specification
Tajikistan · animal-contest · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 2
# Buzkashi — AI-optimized specification
## General idea
A high-stakes equestrian field game modeling the struggle over a heavy carcass (*buz*). Agents control mounted riders navigating a 2D spatial plane, contending to pick up the carcass from the turf, shield it using horse positioning, break through defensive scrums, and deposit it cleanly inside the target scoring ring (*marra*) to score a *halol* goal.
## Formal model
- **Abstraction taken:** Continuous horseback riding, animal mass, and carcass gripping are mapped to a 2D continuous or fine-grid vector field with momentum, rider stamina, and stochastic contest resolvers for carcass lifts and stripping.
- **Player/agent count:** 2 teams of 4 agents (or N individual competing agents).
- **Turn model:** Real-time ticked (state tick length: 100ms) or simultaneous discrete turn moves.
## Formal rules
R1. Setup: Field dimensions 200 × 100 grid units. A central goal circle (*marra*, radius 5 units) is marked. The carcass (*buz*, mass 35 units) starts stationary at field center.
R2. Movement: Each tick, an agent chooses heading vector and gallop speed . Horse stamina drains proportionally to speed squared.
R3. Carcass Pickup: When an agent is within 2 units of the grounded carcass and traveling at speed , the agent may attempt `hoist_carcass()`. Success probability is .
R4. Possession & Encumbrance: A rider carrying the carcass suffers a 30% penalty to maximum speed and turning rate.
R5. Stripping / Contested Possession: If an opponent comes within 1.5 units of the possessing rider, they may execute `strip_carcass()`. Resolution resolves via relative strength, stamina, and angle of approach. On successful strip, possession transfers or the carcass drops to the ground.
R6. [gap-fill] Illegal actions (whipping opponent horse, dangerous T-bone collision) result in a 30-second penalty box removal.
R7. Scoring (Halol): When the possessing rider reaches the *marra* circle and executes `drop_carcass()`, the drop is scored: if the carcass vector settles entirely within the radius, 1 Goal (Halol) is awarded. If any portion lies outside, the drop is *harom* (invalid) and remains live on the ground.
R8. [gap-fill] A match consists of 4 quarters of 300 game ticks each. Highest team score at the end of 4 quarters wins.
## Edge cases
- **Tie at the end condition:** If tied at match expiration, a sudden-death overtime period of 200 ticks is played with first *halol* winning.
- **A player with no legal move:** Not applicable; movement and defensive blocking are always legal.
- **Repetition / cycles:** Bounded by match clock and horse stamina depletion.
- **Simultaneous claims on the same resource:** If two agents attempt to hoist the carcass on the same tick, an immediate strength-vs-strength stochastic roll decides the claimant; on a dead heat, the carcass remains grounded.
- **Player elimination mid-round:** If a rider is penalized or dismounted, their team plays shorthanded until the penalty expires.
- **Resource exhaustion:** Horse stamina recovers when galloping below 40% speed or standing still.
- **Termination guarantee:** Guaranteed by finite match ticks and discrete scoring intervals.
## State representation
- Global state: Match time remaining, scores per team, carcass state (`grounded(x,y)` or `carried_by(agent_id)`).
- Per-agent state: Position `(x,y)`, velocity vector `(vx, vy)`, `horse_stamina` [0..100], `is_penalized` (boolean), `has_carcass` (boolean).
- Observable information: Fully observable 2D field state for all agents.
## Action space
- `move(direction, speed)`: Set movement vector and gallop velocity.
- `hoist_carcass()`: Attempt to lift grounded carcass.
- `strip_carcass(target_id)`: Contest and attempt to pull carcass from opponent.
- `drop_carcass()`: Release carcass into target scoring ring or onto ground.
- `block(target_id)`: Position mount to impede opponent trajectory.
## Playability notes
- **Single-player-vs-AI:** Natural team game where player controls the captain/carrier supported by heuristic AI teammates providing escorts and blocks.
- **Group games (n>4):** Scales smoothly from 2v2 to 8v8 team squads.
- **Physical-skill adaptation:** Complex equestrian physical mastery is abstracted into stamina management, spatial positioning, and probabilistic grip rolls.
- **What is lost:** Real tactile horse handling, physical whip dynamics, and ground-level dust/wind environment.
## Implementation hints
Treat horse movement with simple 2D Newtonian physics with inertia. Implement carcass possession as an attachment to rider transform with mass drag. First milestone: single rider picking up and dropping carcass into circle before adding multi-agent tackling.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Goat Pulling (Tajikistan). (n.d.). Games of Countries. https://gamesofcountries.com/games/tj-tajikistan/buzkashi
@misc{gamesofcountries-tj-tajikistan-buzkashi,
title = {Goat Pulling (Tajikistan)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/tj-tajikistan/buzkashi}
}