35

I have just read the blog post Iterating on Inclusion by Sara Chipps about all the recent changes SE made to improve the site experience. These are all great ideas but they have one single common downside: they don't teach users how to get better.

You write that:

For some time, we’ve been using machine learning to flag comments that might be potentially troublesome in the community, especially if they could be seen as hostile to new users. (Our tool for evaluating and flagging those comments, called the Unfriendly Comment Classifier, is pretty great! We’ve got details on how it works here.)

(emphasis is mine)

It would be great if you didn't just flag them but notified the user about the fact that it might be potentially troublesome in the community and gave them at least a chance to fix it before it's too late. When someone gets flagged a couple of times and earns a timeout - they might not even know what it was for so there is still a risk that they will repeat this behaviour unwillingly in future.

Additionally there should also be a link to a page explaining how to write good comments. I highly doubt that all bad comments are bad because the authors had bad intentions.

From my experience there are a lot of people who do not know how to communicate or use diplomatic language and they might be flagged even if writing in good faith.

You can help people learn by expanding the comment's help or even providing examples of good and bad comments. The current dropdown only explains what comments are for, and how to format them, with a link to the page Help Center > Privileges > Comment everywhere - which is all very technical guidance about formatting.


I find there should be much more teaching than secret flagging and penalizing. You should show people exactly which comment was bad and why. Don't leave them in the dark. I'm pretty sure most comments are written in good faith, maybe sometimes too emotional but still, I doubt they were written with the purpose to offend anyone.


The current system disciplines users by punishment. In my opinion this is not inclusive at all. The system should support us and teach us how to use it correctly while we are using it and not kick anyone out when this could have been prevented.

You now have the Unfriendly Comment Classifier. Use it proactively to protect not only readers from being offended by inappropriate comments but also their authors from being flagged.


Teaching could also be supported by moderators telling users why a particular comment was removed -- similar to the feedback that shows why a question got closed -- maybe providing some predefined reasons. Just saying, rude or abusive is too general. People have different standards and many users don't speak English as a native language, so they might not even know something was perceived as rude. We have the chance to learn how to write good questions but we're completely lost when it comes to the comment world.

  • Are you saying that if there is an auto-flagger that runs on posted comments, it could be used on the content of comments written but not yet posted, to warn the user that the thing they are about to say is dodgy? If so, that's a big yes from me! – Richard Ward Oct 19 at 18:39
  • After thinking about it, I'd like to add "...as long as they still have the option to post it anyway." – Richard Ward Oct 19 at 18:55
  • @RichardWard I believe that tool currently works pasively, this is after a comment was posted but making it work proactively should not be such a big deal. – Christine H. Richards Oct 19 at 19:51
22

I am a frequent user on quora.com, too. That site gets many things wrong, but there is a notification for any post of yours that got removed or "hidden" from public view.

Which gives you the chance to either appeal that decision, or to improve your content. Most of the time, that is just a good information "do better next time", but it also allows you learn about unjust flags (my appeals are honored maybe 60 percent of the time).

I remember my one and only suspension on stackoverflow.com, the mail said "flags are following you everywhere". Thing is: there were earlier warning mails that quoted specific examples (helpful), and I tried to adapt. I really thought "I am doing better". Thus I was a bit surprised when that suspension came up several weeks later.

Note: I am not complaining about the suspension, it was okay and helpful. But I think: if the system would have shown me "dude, you got X comments flagged this week" , or "overall you got Y flags for being rude today", I think I would have reacted and changed behavior without that suspension. And: earlier, too.

Thus: a system that somehow provides feedback on my content that got flagged, that could be very helpful. The system can count flags, and tell you the numbers, without putting any workload on overworked moderators!

I agree that there is no need in telling users about "no longer needed" flags, but some flags indicate negative feedback, and users might never hear about that. Because their content just gets deleted without the user knowing about it.

And even if exact flag count isn't given, there could be other options, like a "traffic light" warning level "Beware: you entered red zone on rude flags, so better slow down immediately".

Thus: there is plenty of room for improvement here, and I honestly do not understand why the system is designed the way it is right now.

14

I support this proposal.

Here's what I do currently as a workaround: I have often felt frustrated about how difficult it is in the current system for participants to learn from mistakes. So, I've started leaving a constructive comment in addition to raising the flag, in cases where I get the vibes the participant means well but would benefit from some polite feedback. Often, the participant then self-deletes the offending comment and tries again, with an apology and a better tone.

Once in a while, the participant gets upset and writes a nasty comment in response. In such a situation, I refrain from commenting anything more under that post, and raise another flag.

You must log in to answer this question.

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