The idea of automatic feedback-driven "moderation" (if you will) is good, and has been shown to work in many contexts. But, as many point out, it is susceptible to vendetta and "possees".

You may believe that "possees" may never form here, but the very openess of the platform could easily allow an individual or group of individuals to raise havoc very quickly.

You could prevent this by limiting the amount of -ve feedback a person can get, in aggregate, in one day (apologies if this was suggested before).

Antoher approach is giving members the ability to filter what they do and don't see. Many spaces would allow me to say "I don't want to see posts by this person", or "I don't want to see posts by this person in this workspace". This is good because it lets me control what I consider to be moderation, not the system.

I realize this is off-topic from the suggestion, but it's an alternative you may consider, since you only need moderate on feedback in extreme cases.