9

What is the "IPS Comment Bot?" Why is it posting in our site chat? What do the messages it posts mean? Can I interact with it? If it posted one of my comments, does that mean my comment was bad?

| |
9

What is IPS Comment Bot?

IPS Comment Bot is a bot that thesecretmaster was awesome enough to create for us. The bot's sole purpose in life is to record every comment posted and report any it thinks may be in need of moderation. It logs every comment that gets made on the site and analyzes them for things like toxicity level, whether they're made on an inactive post and whether they match user made regexes.

The bot lives in The Closet, where it posts every comment it records.

But why is the bot posting in The Awkward Silence?

If the bot sees something funny, it'll also report the comment in The Awkward Silence (TAS) to alert users that the comment may need attention. Reasons that the bot will post include:

  1. The comment matches one of the user defined regexes (the most common reason).
  2. The comment was made on a post that has been inactive for >30 days.
  3. The comment has toxicity rating >.7 (toxicity is assigned by Google's Perspective API).
  4. The bot suspects the comment may be abusive or offensive.

What are the messages the bot posts?

The bot will only post when reporting a comment. These reports will consist of three chat messages. For example:

does anybody ever enter descriptions??

  1. The actual comment content itself1.
  2. Information about the comment including the author, the parent post, the toxicity, etc...
  3. The titles of the user defined regex reasons if any were matched.

If the bot posted a comment in TAS, does that mean the comment was bad?

Not at all!! The bot (and regexes behind it) is by no means perfect. Just because a comment has been reported, it doesn't mean it's time to immediately run out and flag it to death.

The bot exists to alert us to comments that possibly require moderation, but you still need to use your own judgement when evaluating them.

Wait, the bot can be wrong? Can I rub that in its face??

Err...let's be nice to the bots, you never know when they'll get too smart. But if you'd like to help the bot learn, it'll take feedback on comments it reports.

If the comment was incorrectly reported and is actually a good comment that belongs on the site, you can reply to one of the messages the bot posted with "fp" for "False Positive." Conversely, if the bot correctly reported a comment and you want to give it a pat on the head, you can reply with "tp" for "True Positive."

The bot should reply to your feedbacks to let you know you've been heard. It'll say something like:

Marked this comment as caught correctly (tp). Currently marked 3tps/1fps

This will help us analyze how the bot's doing so that we can tweak things to make it more accurate in the future. Every report helps!

Hold up, you didn't answer my question about XYZ!

If you ever have more questions about the bots, you can always ask in chat and there will usually be someone around who knows someone who can ping someone to help.

If you think the question will be too long to answer in a few chat messages, I've also created the tag so that you can post a new meta question here.

Have fun, but don't hurt the bots!!


1. This will be replaced with a link and the text ⚠️☢️Offensive/Abusive Comment☢️⚠️ if the bot thinks the comment may be offensive/abusive.

| |
  • I'd add a section regarding "my comment was permanently archived, normally I can delete it". – gparyani Nov 1 '18 at 22:44
  • @gparyani I’m not sure what you mean (or how it relates to the bot). Can you elaborate? – scohe001 Nov 1 '18 at 23:04
  • Some users might object to the fact that their comments become permanently archived and searchable if they trip one or more regexes, or overall in the testing rooms, and want to retain the ability to delete them as they please (which is possible on other sites, unless a mod manually moves it to chat). SE's position is that this isn't an issue. – gparyani Nov 1 '18 at 23:16
  • @gparyani ahh I see. It may be worth making another self-answered meta question about that. Or even adding another answer here. I feel like this answer is already a bit of an information overload and I haven't heard anyone complain about that here yet. – scohe001 Nov 2 '18 at 1:02
  • 1
    @gparyani I agree with Scohe on the fact that it'd be worth adding another answer to address this topic in you think that's necessary. – avazula Nov 3 '18 at 18:25
  • @scohe001 I posted one here: interpersonal.meta.stackexchange.com/questions/3396/… – gparyani Nov 11 '18 at 1:18
  • @avazula I posted one here: interpersonal.meta.stackexchange.com/questions/3396/… – gparyani Nov 11 '18 at 1:18
  • 1
    What if the bot catches a comment as "aic" (answer-in-comments), but the comment is actually chatty? The comment is still unnecessary but it was the wrong regex that caught it. Is it a tp or a fp then? – kscherrer Nov 14 '18 at 14:11
  • 2
    @Cashbee Good question! tp/fp are simply measures of whether the comment needs moderation or not. You can even tp/fp comments in The Closet that the bot hasn't flagged at all. It gives us some data for what kind of comments are good or not so good. If you see the same regex consistently throwing the wrong reason (but still correctly reporting bad comments), bring it up in chat and we can change what the reason says. – scohe001 Nov 14 '18 at 14:30

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .