[dotancohen]

I believe that lobste.rs worked in a similar fashion. I know that I was invited by a community member, I don't know if I misbehave if that affects his reputation.

Perhaps a weighted scale would work. Alice could invite Zachary, taking responsibility for, by her choosing, 1 to 100 percent of his behaviour. Bob could invite Zachary too, also taking responsibility for 1 to 100 percent of his behaviour. Zachary could only begin using the service after 100 percent of his behaviour is accounted for. Then, when Zachary is reprimanded, Alice, Bob, and all others take a hit as well, weighted by their responsibility.

I might be willing to take 100% responsibility for my brother, but I would take about 20% responsibility for RMS and perhaps 1% responsibility for some guy I work with.

[fdw]

That's an interesting idea. This would probably curb bad behavior, when you get reprimanded from several folks who have an actual stake in it.

But - and sorry if I'm overly negative here - I see other issues: First, you wouldn't want to dilute the stakes too much. If my bad behavior is a very small problem for 100 people, they probably don't care too much and nor do I. The more pressing problem might be that growth is now really heard. The easiest way is if someone takes 100% responsibility, but then you only get very similar people and no new ideas. Or you need several people inviting the same person, so your community cannot grow exponentially (I think).

This might be a good approach if you strive for homogeneity and accept small growth.

[mariusor]

Yes, the invitation model is borrowed from lobste.rs but penalizing the chain of people which resulted in a specific user joining the instance will never happen. The main reason being that it would be very difficult to model this behaviour using the ActivityPub vocabulary that the website is built on top of.