The open-invite system works well in many cases. It works particularly well in-person but even there you can get drift over time. Our fraternity unanimously agreed on every single initiate who joined; the cohort today is still very different from the one 20 years ago.

In online systems the scales quickly get too big for open-invite. There needs to be a way to automatically update the trust network at a fine grain.

The one that jumps to mind is an inference system; when I +/- a comment, I'm really noting that I trust or distrust the author. It can be general or on a specific topic (eg I trust the author to tell the truth or I trust the author to make me laugh). I could also infer that other people with similar trust patterns are likely trustworthy. And I could likely infer that people who are trusted by people I trust are trustworthy.