A few weeks ago, I noticed an odd post on a forum that lead me to uncovering a vast reputation management operation that I’ve documented some of on GitHub. It involves new users asking basic questions while repeatedly mentioning a name, including in the title of the post, to try and hide negative news stories in search results.
I also noted another post that was a long question but contained an odd looking link to a spammy domain, and have found many other similar posts by the same username and same domain across multiple forums (user = benof, domain = igmguru dot com - search for "benof" forum).
The difficulty however, is that it is very hard for me to report this spam as a user - I have to sign up to each of the forums (and then receive endless emails), reply to the posts and hope someone notices and takes action. I can’t report posts as spam which seems like a major limitation (although understandable).
I realise this is tricky to solve and goes way beyond Discourse, but it really irks me that they waste volunteers’ time by getting answers to questions that they have no actual interest in. AI makes it much easier to make for these spammers to hide what their real motive is by asking what appear to be genuine questions.
A few thoughts of features that could help:
- Earned reputation linked to email addresses, similar to how reputation works on Stack Overflow sites, so that if I am a legit user of one forum I can at least report spam as soon as I sign up.
- Scanning of all posts made on Discourse forums to try and spot patterns e.g. of unusual domains being linked to / phrases being repeated.
- Shared spam rules between forums, including retrospective flagging so that once a pattern is spotted it can be dealt with automatically.