Hey @Alex, nice to meet you! I'm not into reputational points but I am interested in heuristics for understanding trust, and supporting people to ask different questions of the data here.

re: flags, I think these could be really neat to explore, but it's important to me that we flag behaviours not people. This allows people to be treated more humanely, and allows actions to be questioned and unaccepted, as opposed to people being flattened into labels and rejected. It also has other benefits like you have to point at some post or content to be able to "flag" some action, which immediately gives the flag a specificity and context that just saying "is sexist" lacks. It also means we can see a range of behaviours over time, and potentially choose to hide some content if it's been flagged with certain things.

Finally I'm quite keen on tuning how our systems replicate more dynamically. At the moment we replicate our to "2 hops" (follows, and people those people follow). What if we made it "2 hops, and prune replicating people who'd got more than 10 flags from people we follow". I don't know what this would do but I think we'd learn a lot. My hope is simple rules like this (that each person can tune themselves) could make for a really effective way to distribute the labour of moderation.