we are three researchers (Felix, Jonathan, Johannes) and over the past years, we've been doing research on polarization and misinformation. An earlier project from us is Quality News (https://news.social-protocols.org/). This is our first prototype for a platform for collaborative fact-checking.

What we have so far:

- automated detection for rhetorical fallacies in posts. This is done using gpt-4o-mini. We might release that as a library. Let us know if you're interested.

On the roadmap:

- calculate fact check results taking all the votes and comments in a thread into account

- comment ranking by measuring how mind-changing a comment is. We'll decide this by analyzing voting patterns using bayesian statistics.

- game theoretical reputation system with incentives for honesty - bias correction for votes with matrix factorization (similar to Community Notes)

- crowdfunding of fact checks (money distributed to contributors based on the effectiveness of their comments and votes)

We're aware that the true/false vote buttons are not ideal for all the content people will post. But we think it's a start. If you have ideas in this direction, please let us know.

We appreciate any feedback!