Thanks. The signal to noise ratio in our test groups is quite high. Initially, people joined a group using a unique URL shared by the group creator. Whilst that was completely anonymous, it would allow bad actors to keep registering new throw-away accounts as long as the URL remained the same.
I've since switched to one-time email invitations. The only way to join a group is to be invited by email, and each invitation can only be used to create one account.
Group creators can ban or suspend users who don't play by the rules, so ultimately, it's up to them just how toxic or gossipy a group becomes.
I see. It'd be interesting to see what kinds of behavior the anonymity would bring out in the long run even with the restrictions.
On a different note, have you given consideration to side-channel leaks? It seems possible to me that over time, one might notice trends and correlate a certain behavior with the presence of someone specific in the group, who is known to be in the group without their username being known.
Do you expect the usage model to involve "volatile" groups, as in, you create a group for every project or side project, or are groups long term?
Either way, pretty intriguing project, particularly as a social experiment. Best of luck.
