Gavel is an automated end-to-end expo judging system. We’ve used it to automate judging at HackMIT, a 1000-person event with over 200 projects and 100 judges. Dozens of other events have also used Gavel since the software was released in private beta in late 2015.
Gavel fully automates expo judging logistics and project ranking — the system tells judges which projects to look at, collects judges’ votes, and produces a ranking of projects.
Here’s a demo of the judge interface:
Gavel’s model is pretty different from traditional scoring methods — it’s based on the method of pairwise comparisons, and the system uses fancy math to come up with a ranking. For reasons described in detail in this post, pairwise comparisons work much better than having judges input scores from 1 to 10 or something like that.
The HackMIT team has used Gavel for four events now. We implemented the first prototype, a text message based system built on top of Twilio, at Blueprint in Spring 2015. In Fall 2015, we switched to the Crowd-BT algorithm and reimplemented the system as a web app for HackMIT. We continued to use that implementation for Blueprint in Spring 2016. For this year’s HackMIT, we completely redesigned the frontend and the admin panel of the application, changing basically everything except for the core algorithms. Now the system is a whole lot more user-friendly for both judges and event organizers.
Here’s what judging looked like at HackMIT this year:
Using this system has made judging logistics super easy for us, and we believe that it has also led to higher-quality results. We thought that other events could benefit from using Gavel too, so we’re releasing the software to the public!
We’re releasing the latest version of the system, Gavel v1.0, which is what we used at HackMIT 2016. From this point on, Gavel will be developed as an open source project.
Gavel is free software released under AGPLv3, available for download here.
If you use Gavel for your event or have any feedback about anything, I’d love to hear about it.