Formal Specification
Azerbaijani tug-of-war — Formal Specification
Azerbaijan · strength-endurance · Confidence: medium
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 3
# Kəndir dartma — AI-optimized specification
## General idea
A team rope-pull along a 1D line, distinguished from a generic tug-of-war by three named foul categories with different severities — a mid-pull stall (locking knees and sitting without backward movement), unauthorized rope-wrapping by non-anchor pullers, and a pre-start foothold-digging violation — plus a designated Anchor role that is the sole exception to the wrapping prohibition.
## Formal model
- **Abstraction taken:** Continuous grip strength, footing, and synchronized cadence across each team are aggregated into a single per-tick team force value plus a stochastic slip term, since the entry describes coordination qualitatively (call-and-response cadences) without a numeric force model.
- **Player/agent count:** 2 teams of 6-10 (equal size), or a 1v1 duel variant (entry's `1v1 İp dartması` alternative).
- **Turn model:** Synchronous ticks; both teams submit aggregate pull actions simultaneously each tick, resolved at tick end.
## Formal rules
R1. Setup: the rope's center ribbon is placed at `position = 0` on a line bounded by each team's 3 m victory marker (entry §1.1); pullers line up on opposite sides; each team designates its rearmost puller as Anchor, who may wrap the rope tail across their shoulder (entry §1.3).
R2. Pre-start foul (entry §3.3): if any team submits `dig_foothold()` before the referee's start command, that team receives a warning and the setup resets [gap-fill: the entry names this a prohibition but not its consequence; a warning-and-reset is used here rather than an outright loss, since it occurs before the contest itself begins, distinct from the in-play fouls below].
R3. Start: on the referee's three-stage call ("Tutun! Gərin! Çəkin!", entry §2.1), the pull enters ACTIVE phase.
R4. Per-tick pull: each team submits `pull(intensity)` (aggregate of all non-fouled pullers' effort, entry §2.2-2.3). `position` updates by the force differential between the two teams' committed intensities, plus a stochastic slip term.
R5. Mid-pull foul — stall (entry §3.1): if a team's aggregate state includes `locked_or_seated = true` for a sustained period without `position` moving in that team's favor, that team loses the pull immediately [gap-fill: the entry calls this "a foul" without stating the consequence; an immediate pull loss is used, consistent with how fouls are resolved severely elsewhere in this batch's strength-contest entries].
R6. Mid-pull foul — illegal wrap (entry §3.2): if any non-Anchor puller on a team wraps the rope around their hand, wrist, or elbow, that team loses the pull immediately, with the same severity as R5.
R7. Win condition per pull (entry §4.1): the pull ends the instant `position` crosses either team's 3 m victory marker; that team wins the pull.
R8. Match structure (entry §4.2): best-of-three pulls, with teams swapping sides between pulls (a fairness measure with no effect on the formal rules beyond resetting `position` to 0 for the next pull). The match ends the instant one team has won 2 pulls.
## Edge cases
- Tie at the end: not applicable at the pull level (R7 always produces a single winner, and fouls in R5/R6 are also single-team losses); not applicable at the match level since R8 requires a clean 2-pull majority.
- A player with no legal move: not applicable — a team always has `pull(intensity)` available while ACTIVE and not fouled out.
- Repetition / cycles: two evenly-matched teams could in principle hold `position` near 0 indefinitely through balanced pulling, but rising fatigue (mirroring the hazard-based termination argument used for this batch's other strength-contest entries) increases the stochastic slip term's variance over time, driving `position` toward a boundary or triggering a stall foul (R5) with non-vanishing probability [gap-fill: fatigue-scaling is assumed by analogy, since the entry frames this as an endurance contest but gives no explicit fatigue mechanic].
- Simultaneous claims on one position: not applicable — this is a 1D contest with a single shared `position` value.
- Elimination mid-round and turn-order repair: not applicable — a pull has exactly one winner and ends immediately (R5-R7); the fixed best-of-three structure (R8) is unaffected by any single pull's outcome beyond the running win count.
- Resource exhaustion: not applicable — no consumable resource; fatigue is a probability modifier (per the Repetition note), not a depletable quantity.
- Termination guarantee: `position` is confined to a bounded interval between the two 3 m victory markers, and any sustained force advantage drives it toward a boundary (R4, R7); independently, the stall-foul hazard (R5) and slip-term variance are non-decreasing with elapsed ticks and bounded away from 0, so the probability a pull remains undecided after t ticks tends to 0 as t → ∞. The match is a fixed sequence of at most 3 such pulls (R8).
## State representation
- Persistent (match) state: `pulls_won[team]`, `pull_number`, `sides_assignment` (swapped each pull, R8).
- Per-pull state: `position` (shared, bounded), `ticks_elapsed`, `locked_or_seated[team]` (bool, with a duration timer for R5), `illegal_wrap_detected[team]` (bool).
- Observable information: fully observable — both teams see the rope position and elapsed time; no hidden information.
## Action space
- `dig_foothold()`: either team, pre-start only; triggers R2 if used before the start command.
- `pull(intensity)`: either team; precondition: pull ACTIVE, team not yet fouled out.
- Fouls the referee model must recognize but that are never in the legal "advance position" action set: sustained `locked_or_seated` (R5), non-Anchor rope-wrapping (R6).
## Playability notes
- Single-player-vs-AI: workable as a human managing one team's aggregate pull-intensity strategy against an AI-controlled opposing team; less natural as a literal multi-agent simulation of 6-10 individual pullers unless each is modeled as contributing to one team-level action.
- Reducing team size to a 1v1 duel: the entry explicitly supports this via the `İp dartması` variant, which the formal model handles identically with team size = 1 and no Anchor-specific rule.
- Physical-skill games: the entire grip-strength, footing, and cadence-coordination contest is reduced to a single aggregate force value plus a stochastic slip term (R4); no genuine biomechanics or team synchronization is modeled.
- What is lost: the real physical strain and rhythmic cadence chanting that coordinates a team's pull, the ritual festival framing (believed to ensure rainfall and harvest prosperity), and the specific tactile skill of grip and heel-digging technique.
## Implementation hints
State is minimal (one shared position value plus small per-team counters); the actual design difficulty is calibrating the force-differential-to-displacement mapping and the fatigue/slip-variance growth rate, neither of which the source quantifies. First milestone: implement `pull()`-only pulls to validate the boundary-crossing win condition (R7) and best-of-three scoring (R8), then add the three foul types (R2, R5, R6) and fatigue-scaling.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Azerbaijani tug-of-war (Azerbaijan). (n.d.). Games of Countries. https://gamesofcountries.com/games/az-azerbaijan/kendir-dartma
@misc{gamesofcountries-az-azerbaijan-kendir-dartma,
title = {Azerbaijani tug-of-war (Azerbaijan)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/az-azerbaijan/kendir-dartma}
}