Recently a co-worker described to me a concept that he called the "bozo room" as a way to control spam, trolling and sock-puppeting in message boards, etc. I'd like to know if it has ever been formally studied, or has actually been successfully put into practice on any successful website, online game, etc. Specifically, I'd like to know:
- Does this strategy actually control the number of sock puppet accounts, as compared to simple account banning or closure?
- Does this strategy actually control the number of sock puppet accounts, as compared to the simpler hellban strategy?
- Does this strategy carry significant negative side effects (like chasing away users), as compared to these other strategies?
The idea is, if a user is flagged as a troll or spammer by a moderator, no notification is given, and that user's experience is completely unchanged, but only flagged users (themselves and others) can see their posts. Flagged ("bozo") users can see all posts, even those of other flagged users, but unflagged ("good") users see only the posts of other "good" users.
The idea of letting "bozo" users see each others' posts is what leads to the term "bozo room", because they can still interact, but only with other "bozos", perhaps increasing the chance that they will not realize they have been flagged.
An an example, if we imagine that StackExchange has implemented this, then if you and I are in the bozo room, we can read this question, vote on it, answer it, and so on; but non-bozos never see it in searches, on the main page, or anywhere else. But there's no indication in our profiles or anywhere else that we can see that tells us we're bozos, and therefore there is no way to appeal to have the ban lifted (if we're sure we are banned), besides screaming into the void.
In my own research, starting from this metafilter post I found references to "hellban", "ghost posting", "sending to Coventry" and the "twit bit". From there I found Drupal Cave. Apparently the hellban has been used by Something Awful and on BBSes as far back as 1987, and Atwood's post claims that the hellban is secretly used on large-scale sites, but without naming names. But in the normal hellban scheme, only the bozo sees his own posts, not other bozos. By allowing bozos to see each others' posts, different metrics may or may not result. I found no discussion of attempts to measure unwanted consequences of using hellban.
This idea is also distinct from a simple bozo filter or killfile because those strategies are just simple filters; whereas the bozo room strategy takes care to conceal from the "bozo" user the fact that he has been flagged and his posts are not generally being seen.