So, we used to have 'too chatty', 'not constructive', and 'obsolete' as separate flag options, which moderation could use to see, for example, if someone was posting lots of unconstructive comments because there's some correlation between comments being flagged and a per-user statistic or something. Not being a mod, I'm unclear on the details.

Now we just have Super Bad, No Longer Needed, and Custom Flag of Mod Summoning +1. My issue is that 'No Longer Needed' incorporating most things that were never helpful seems like it would make it hard to differentiate between someone who posts lots of excellent comments (who should accumulate a lot of NLN:Obsolete flags) and someone who posts lots of annoying, argumentative, and/or unhelpful-- but not offensive-- comments (who should accumulate a lot of NLN:Too Chatty/NotConstructive flags).

Is this an actual detriment to our site policing situation, or am I just not understanding things properly?


2 Answers 2


Yes, the new flags end up working differently, being used differently, and giving us different immediate1 information. From what I've seen this is mostly positive though.

The purpose of the change as I understand it was to make it easier for flaggers to tell which flag to use, and for mods to be able to understand faster why a comment was flagged. It seems to succeed on this point so far.

In terms of tracking flags, the No Longer Needed flag is the only one that doesn't go in the “bad flags” list in the moderation tab on a user's profile. This does mean that if people use flags exactly as before and map (Not Construction, Obsoletely, Too Chatty) ⇒ (No Longer Needed) 100% of the time, then we're losing some of the intended flagging messages. The upshot is twofold though:

  • People aren't using flags 100% the same as before (as intended by the change).
  • Flags that do get cast tend to be more accurate, because the there's much clearer separation between their meanings.

As a result I'm seeing what looks like a few subtle shifts in how flags get used around problem comments.

  • Low-level annoyance comments are being more deftly dealt with.

    These used to be flagged with “bad” flags, but now tend to get a NLN flag that effectively means “please remove this but it's not a problem.” Since it's faster for mods to evaluate the accuracy of this flag, the comment gets dealt with faster. This frequently removes the irritation quicker, helping to defuse a low-level situation's potential to become a real problem. (People can't engage with annoying comments that aren't there.) Often de-escalation rather than an intervention is the right level of mod response, so this is nice as the new default. And since this doesn't go on our mod view of profiles, the list of problem comments on an account no longer contains comments that, individually, don't seem to belong when reviewed later.

  • Patterns of problem comments are better documented and involve a touch more community moderation.

    These aren't documented now by strings of low-level comments… which used to require mods to laboriously reconstruct them by trawling through lists of “why was this comment flagged ‘bad’?” comments. Instead, patterns made out of individually-meh comments spur non-mod users to escalate to descriptive Custom flags. This means that a) only active problems that merit putting in that much effort get logged, and b) the exact problem is documented textually for us and for posterity.

I think that the overall effect of the change will be that our data is sparser but richer and more accurate. There's just less noise in the signal.

As a nice side effect, accounts no longer earn “black” marks that are actually grey marks that we have to mentally filter out when quickly checking someone's posting history. Now everything in that list from after the change is actually a significant data point. And since we can still always go into yet another view1 to see all someone's flagged comments regardless of reason, we don't lose the data if we do need it later.

  1. We have another view where all comments of a user ever flagged can be reviewed, so what doesn't get shortlisted to the “bad” list is still recorded and available to review. It just doesn't immediately jump to our attention when looking at the mod tab of a profile.

In my eyes, this is a valid concern:

I often do go through my posts and then correct stuff based on comments, which I subequently flagged as obsoltete, and soon were removed.

With the new system putting the "Chatty" and the "Obsolete" together, it might create an overwhelming number of "Chatty/Obsolete" tags for some of those that used to get tons of obsolete but not chatty flags. As they now are tracked together it might become a problem IF that number is used to determine removal/suspension of comment privilege. However is it used to determine comment privileges? THAT is a question that has to be solved first, and I am not in a position to do this.

However, even if this number of removed Chatty/Obsolete is not used to determine removal/suspension of privileges, then it is still not a good thing to put "Constructivve but now incorporated and thus obsolete comment" and "nonconstructive or chatty" into the same bin.

  • \$\begingroup\$ (NB that there's no mechanism to remove just comment privileges — only a temporary suspension can prevent someone from commenting after they earn the privilege.) \$\endgroup\$ Commented Oct 8, 2017 at 9:05
  • \$\begingroup\$ Yeah, we don't have comment-specific bans, just I was under the impression that in the past we used comment tracking data to tell us stuff about how we could improve our behavior (e.g. see rpg.meta.stackexchange.com/questions/3447/… only I thought there was one with flags but maybe I was wrong about that...) \$\endgroup\$ Commented Oct 8, 2017 at 10:30
  • \$\begingroup\$ A suspension could be based on the tracked data, 7SD, so there might be a point in keeping the data seperate. \$\endgroup\$
    – Trish
    Commented Oct 8, 2017 at 11:09

You must log in to answer this question.

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