Formal Specification
Rolling Disc Target Game — Formal Specification
Australia · dexterity-throw · Confidence: low
← Back to game pageImplementation Suitability
Suitability: adaptation-requiredFidelity: probabilistic-proxyGap-fills: 2
# Gorri — AI-optimized specification
## General idea
A moving-target throwing contest with a hard time window: the disc is only live between two markers, so every thrower is solving the same interception problem simultaneously and the round is over in seconds. The decision is when to release, not where to stand.
## Formal model
- **Abstraction taken:** the disc's roll becomes a position advancing one cell per tick along a lane; a throw becomes an interception check against the disc's position at arrival time. No physics.
- **Player/agent count:** 1 roller, 10–20 throwers.
- **Turn model:** ticked and simultaneous — all throwers may release on any tick while the disc is live.
## Formal rules
R1. Throwers stand behind a baseline parallel to the lane, 5–10 units from the rolling path. The lane is 20–30 units long.
R2. The roller calls *gool-gool* and releases the disc, which advances along the lane one unit per tick.
R3. While the disc is between the boundary markers, any thrower with missiles remaining may `throw(lead_distance)`.
R4. A throw arrives after a flight delay and strikes if the disc's position at arrival is within the hit radius of the aim point. Success probability decreases with the thrower's distance and the disc's speed. [gap-fill] No accuracy model is given.
R5. Throwing stops immediately once the disc passes the end marker.
R6. One point is awarded per direct strike that hits or knocks over the target.
R7. Each thrower has 2–4 missiles per round.
R8. The highest cumulative hit count over 5–10 rounds wins.
R9. [gap-fill] The roller does not throw and rotates each round. The source does not say how the roller is chosen.
## Edge cases
- **Tie at the end condition:** equal hit counts after the final round are common with many throwers. Resolve with one extra round; the source supports nothing more specific.
- **Player with no legal move:** a thrower out of missiles has no action for the rest of the round and is simply idle — legal and expected.
- **Repetition / cycles:** none. The disc advances monotonically and each round is a single pass down the lane.
- **Simultaneous claims:** several throwers may strike the disc on the same tick. All are credited — the source describes points per strike, not a single first-hit award.
- **Mid-round elimination and turn-order repair:** nobody is eliminated. All throwers act in parallel, so there is no order to repair.
- **Resource exhaustion:** missiles (R7) are the resource; running out ends that thrower's round.
- **Termination guarantee:** the disc traverses a finite lane in finite ticks, and the round count is fixed by R8. The game provably ends.
## State representation
Persistent: lane length, marker positions, baseline offset, missiles per thrower, round count.
Per-thrower: missiles remaining, cumulative hits, skill parameter.
Per-turn ephemeral: disc position, in-flight missiles with arrival ticks and aim points.
Information set: fully observable — everyone watches the same disc.
## Action space
- `roll()` — precondition: actor is the roller and the disc is not live.
- `throw(lead_distance)` — precondition: actor is a thrower, the disc is live, missiles remain.
## Playability notes
- **Single-player-vs-AI:** natural and genuinely non-trivial, because the agent must lead a moving target under a flight delay. That is a real estimation problem, unlike static target games.
- **Group games (n>4):** throwers do not interact at all, so it reduces to one thrower without structural loss — though the source describes it as a large communal event, and reducing it loses that character entirely.
- **Physical-skill model:** accuracy as a function of thrower distance, disc speed, and lead error. Lead error is the interesting parameter.
- **What is lost:** the expectation of modesty in victory, which the source records as part of the game's conduct. Also the fact that this trained hunting skill against live moving prey — the spec models the interception and none of its purpose.
## Implementation hints
The flight delay is what makes this more than a dexterity check — implement it before the accuracy model, or you get a static-target game with extra steps. Branching factor per thrower per tick is small (throw or hold, plus a lead parameter), so the interesting work is in the continuous lead estimate rather than in search.
Download as JSON
Download the full game corpus entry for this game as structured JSON.
Download JSONCite this page
Rolling Disc Target Game (Australia). (n.d.). Games of Countries. https://gamesofcountries.com/games/au-australia/gorri
@misc{gamesofcountries-au-australia-gorri,
title = {Rolling Disc Target Game (Australia)},
year = {n.d.},
publisher = {Games of Countries},
url = {https://gamesofcountries.com/games/au-australia/gorri}
}