Human Reputation: A New Rating System
github.comWe are currently working on Curyo. The main idea is to create a new rating system for the age of AI, since we think we need better/new quality signals for the web. We are generally interested in getting more feedback on what a good rating system should be.
FYI: We previously submitted a link to the website to invite people to test it, but it was immediately flagged as "dead". I'm not sure why, but the project is completely open source, and there's no money involved or anything like that. It uses a public ledger, but this is because we thought a trustless ledger would make sense for a new rating system (because of the transparency). We are happy to change this if people prefer a database, because it's more user-friendly.
Assuming widespread adoption, what’s to stop this from turning into a closed “elite club” walled garden? I’ve found that oftentimes, as soon as you attempt to distribute the decision of who’s to be trusted, you’re definitely going to end up filtering for something, but it’ll rarely be what you initially intended to filter for.
It works better in maintainer-led allow/trustlists because the people who need the trust hold all the levers that decide what “trust” means. But when you distribute those levers to everyone, it makes sense that it’d eventually drift into almost a self-defined system.
I agree that this is an issue. We are currently trying a combination of two things. First, every human can verify themselves via the passport, and this comes with some "voting rights". This way, we keep it open. Second, there is a per-person, per-content limit on voting power. This aims to prevent some people from becoming too powerful. The exact limit needs to be adapted over time. But yeah, we haven't found the perfect solution here, and we need to keep working on it and improving it. That's also why we are looking for as much feedback as possible.
Great idea. Cannot wait to read the full paper.
Thanks. I'm working on it