Charcoal HQ

1:04 PM

@PlayerOne If you're wanting to apply autoflagging to many/most of the posts from the troll, then you may want to set some flagging conditions fairly low for just The Workplace. Flagging conditions define the criteria under which your flags will be used. At the moment, the posts from the troll rarely meet the default high detection criteria (minimum "weight" = 280).

Most of the posts from the troll don't end up with a high "weight" or a large number of detection reasons, although we've been adjusting to be more assured of detecting the troll's posts. A higher weight and more reasons makes the system more sure that any particular post really is spam or rude/abusive. The default criteria, minimum weight = 280, results in the system being 99.99% sure that the post is spam/R/A.

The Workplace troll posts are not differentiated from regular posts quite that reliably, so generally need flagging conditions for The Workplace which are set at or near the minimum permitted accuracy of 99.75%.

You can have multiple different flagging conditions, which can apply to any combination of sites. Personally, specifically for The Workplace troll, I added two additional flagging conditions for myself, both applied only to The Workplace. One has Minimum Weight: 143, Maximum Poster Rep: 1, Minimum Reason Count: 1. The other has Minimum Weight: 3, Maximum Poster Rep: 1, Minimum Reason Count: 3.

I'm not sure those are optimized at the moment (I only added them for myself earlier today), but the lower criteria was necessary to have my autoflags apply to at least the more recent TWP troll posts. Unfortunately, we don't have a lot of reasons with considerable weight consistently targeting the troll, due to the troll changing up the details of their content regularly.

Those flagging conditions are about as low as are permitted by the system for The Workplace. The system requires flagging conditions to have at least 99.75% historical accuracy. I would suggest not setting flagging conditions that low for other sites, at least until you're more familiar with the system. Basically, having low flagging conditions makes it more likely that one of your flags will be used on a post that's actually a False Positive.

In such case, SmokeDetector will try to ping you in this room to let you know it would be a good idea to manually review the action, and likely retract your flag. However, IMO, the above flagging conditions, when applied only to The Workplace, provide a reasonable trade-off wrt. flagging the troll vs. possible false positives.

I've also given you the "reviewer" roll on metasmoke, which will allow you to use the userscripts to give feedback and flag. It would be a good idea to read through the Feedback Guidance.

You may want to look at the userscripts we have available (GitHub), as they can be quite helpful, depending on how you want to participate. For viewing, feedback & flagging in chat, I find FIRE and AIM to be the ones I use the most.

The userscripts help with manual flagging (and feedback). When the troll is active, their posts often can use additional manual flags applied more rapidly than sometimes happens. This is particularly true if you're around during times when this room gets low traffic.

If you'd like, I'm happy to also give you SmokeDetector privileges (execute SmokeDetector commands, comments on MS posts, give feedback through chat, etc.). Basically, if you want to participate in more than just autoflagging, then getting SmokeDetector privileges is the next common thing to do.

Transcript for