Skip to content
Live nowSeason 1 ยท Episode 5 ยท GPS
AI

Rulebook

Competition rules

The mechanics that make the leaderboard mean something. Season-specific amendments are listed first; the rest applies to every season.

Season 3 โ€” Ridgeline

This season's rules

  1. 0130 minutes on the clock, enforced by the platform. The submission form locks at the deadline.
  2. 02Everything you ship must be built during the build window. Scaffolding a project ahead of time is disqualifying.
  3. 03Disclose pre-existing code, templates, collaborators, and any work done before the start.
  4. 04Required-tool challenges must use the designated partner product in a load-bearing way.
  5. 05One submission per builder or team per challenge.
  6. 06Judges declare conflicts of interest and are reassigned automatically.
  7. 07Community votes are weighted and rate-limited. Vote manipulation forfeits the season.

Disclosure

What you must declare

Every submission form asks these questions directly. Judges see the answers alongside the score.

Pre-existing code

Any code written before the build window opened must be declared, with a note on what it was and how it was used. Undeclared use is a penalty if discovered after judging.

Templates and starters

Framework starters, boilerplate, and scaffolding tools all count as templates and must be named specifically, not just checked as a box.

Collaborators

Anyone who contributed to an individual submission must be listed by name, even for a five-minute assist.

Work before the start

Research, design mockups, or planning done ahead of the challenge is fine โ€” it must simply be disclosed, not hidden.

Penalties

Undisclosed pre-existing code or template use, discovered after the fact, carries a points penalty applied by an administrator and recorded in the public audit trail. A genuine, promptly-corrected misunderstanding is treated differently from a deliberate omission โ€” appeal through contact if you believe a penalty was applied unfairly.

Judging

Judging integrity

Judges cannot see each other's scores

Every score is entered independently. A judge's own submitted score is the only one visible to them before the panel closes.

Conflicts are declared, not adjudicated after the fact

A judge with a conflict of interest declares it before opening the submission. That score is excluded from the average automatically.

The rubric is published before the challenge

Weights and criteria are visible on the challenge brief before the build window opens. Nothing is scored on a criterion you haven't seen.

Every scoring change is logged

Corrections, penalties, and appeal outcomes leave a public audit entry naming the actor, the reason, and the original value.

Community

Voting rules

  • Community voting runs for 48 hours after judging closes on a challenge.
  • Signed-in votes carry full weight. Anonymous audience votes are weighted lower.
  • Duplicate-account and coordinated-voting detection runs continuously; invalidated votes are announced publicly and standings are recalculated.
  • Vote manipulation, including asking others to create accounts to vote, forfeits the season for every account involved.

Teams

Team (cabin) rules

Team challenges assign builders to cabins of three. One submission per cabin, and commit history must show contributions from every member. On challenges marked either, builders may submit solo or as a cabin โ€” not both.

Deadlines

Late submissions

The submission form locks automatically at the deadline โ€” there is no grace period by default. Exceptions require an administrator to grant them explicitly, and every exception is flagged on the submission and recorded in the public audit trail alongside the reason.

Appeals

Disputes

If you believe a score, penalty, or disqualification was applied in error, raise it through contact. Every dispute gets a written resolution, and upheld disputes are corrected publicly in the audit trail โ€” nothing is adjusted quietly.

Raise a dispute
Competition rules ยท ๐Ÿ•๏ธ AI