> ## Documentation Index
> Fetch the complete documentation index at: https://docs.boundlessfi.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# How judging works

> How hackathon submissions are scored against a rubric, and what the AI judging assist does.

> **TL;DR:** Judges score each submission against a weighted rubric. The scores decide the winners. An AI assist can summarize long submissions to save time, but it never scores for you, people decide.

## What you'll learn

* How a judging rubric is structured
* How scoring decides winners
* What the AI judging assist does and doesn't do

***

## The rubric

A hackathon defines a **rubric**: a set of criteria that judges score against. Each criterion carries a weight, and the weights add up to 100. This keeps scoring consistent: every judge evaluates the same things in the same proportions.

Hackathons can also have **tracks** (categories such as DeFi, tooling, or consumer). Submissions enter a track, and prize tiers can be awarded within a track or overall.

<Frame caption="Judges score each submission against weighted rubric criteria that add up to 100; the totals decide the winners.">
  <img src="https://mintcdn.com/boundless-bb5af936/SRy6_o7Nl_hCFvuz/images/diagrams/judging-rubric.svg?fit=max&auto=format&n=SRy6_o7Nl_hCFvuz&q=85&s=79b99b6877eaf084964c6705ada60aae" alt="Weighted rubric criteria feed into judge scores, which decide winners" width="700" height="175" data-path="images/diagrams/judging-rubric.svg" />
</Frame>

## Scoring

Judges review each submission and score it on every criterion in the rubric. Once scoring closes, the totals rank the submissions and determine who wins each prize tier. Some hackathons set a **pass mark**: a minimum quality bar a submission must clear to win a tier, so a prize isn't awarded to weak work.

## The AI judging assist

For submissions with a lot of material, an **AI judging assist** can summarize the entry per criterion so judges can review faster. It only summarizes, it never assigns a score. The judgment is always human.

## Fairness

Scores and outcomes are recorded, so decisions are auditable. If a result is questioned, see [Disputes and moderation](/help/disputes-and-moderation).

## Next steps

* [Score submissions](/organizers/score-submissions)
* [Host a hackathon](/organizers/host-a-hackathon)
* [Tracks, prizes, and judging (participant view)](/contributors/hackathons/tracks-prizes-judging)