9
$\begingroup$

Just today, the Community bot bumped a three-year old question. It's also recently bumped a half-year old question.

Right now about 1/10 of the posts on the 'active questions' meta tab are bumps from Community. I don't think this helps at all, and it worsens the signal-to-noise ratio. Can this be disabled?

$\endgroup$
7
$\begingroup$

This has come up before on meta, at (Why) does this particular question keep getting bumped up by Community?. At the time, there was very little for the Community user to bump, so it just kept bumping the same question every month.

Since then, more questions have become eligible for bumping, either through having their answers downvoted or through zero-score questions receiving a stray upvote.

As of this writing, there are four Community-bumped questions on the front page:

In my opinion, none of them particularly warrant the extra attention.

  • [Edit: recently, and even more ridiculously, Winter bash 2016 - hats are live! got bumped to the top of the front page. Because that very much required additional attention once the winter bash was over, you know.]

Moreover, these are essentially the only bumpable questions at this time. This SEDE query implements the criteria in this MSE post and turns up two of the above posts (this one and this one) plus this one (which was active 9 days ago).

Of the other two posts above, this one is reported by SEDE as having an answer with score 1, for reasons that are not really substantiated in either the question or answer timelines, and this one's answer was upvoted after the bump (it's also an old bug report which is useless to bring back up).

In any case: the set of bumpable questions is small. It can grow slightly over time, as people go into the archives and upvote score-zero questions, or downvote existing positive-score answers.


More than that, though: there really is no reason to dredge old stuff back up, particularly if it is only going to add a 10% serving of noise to the meta front page. Currently, the posts eligible for bumping are

those scoring >= 0 that have gone at least 30 days with no activity, have at least one non-deleted answer scoring 0 and none scoring more than that, and no accepted answer (also, they can't be locked or closed).

I contend that, although it may at times be beneficial to have old feature requests or standing bug reports brought back to light for a while, the current bumping criteria are completely uncorrelated with the kind of stuff that might be worthwhile bringing up.

In particular, there are plenty of reasons why a discussion in meta can be left with zero score: it could genuinely need more attention, but it could also be a five-year-old debate that ended in a +5/-5 vote that says a fair bit about the community's feeling at the time. The narrow criteria above are posed to bring up a bunch of boring posts, and quite well-tuned to miss the discussions (such as those with e.g. a single answer at score -1) that could potentially use someone dropping in and adding an answer with the consensus.

More damningly, however, is the fact that the current crop of bumps wouldn't have happened without people digging through the archives and voting on things. To me this is pretty strong evidence that the rationale for keeping this mechanism ("bringing attention to old unresolved discussions, open feature requests, and unsolved bugs") is not actually that necessary - people are digging through the archive and finding these things, and the bumping algorithm relies on it.


There's really no call to have this on meta: the idea sounds nice, sure, but it is a very blunt tool that does not work at all for the intended objective. Please turn it off.

$\endgroup$
2
$\begingroup$

Requests on MSE to modify the bumping behavior have been consistently (ignored or) declined, so no, this is probably not going to happen.

I don't think it worsens the SNR anyway, because these old questions still count in the "signal" category; just because a question is old doesn't make it undeserving of attention. If you think a particular question that has been bumped should not be exposed on the front page, there are several ways to fix that:

  • Vote the question down if it's bad
  • Vote up a good answer, if one exists
  • Post a good answer, if one doesn't already exist
  • Flag the question for closure if appropriate
$\endgroup$
  • 1
    $\begingroup$ I disagree. It does worsen the SNR. There's very little to gain, for example, by dredging up this one. It is also perfectly OK for a thread to be resolved with an answer at score 0 (e.g. it can be at +5/-5, which is plenty of information about the community's stance). None of this answer really applies, except for the assertion that the team has indeed declined it (without giving any reasons at all, let alone any valid ones). $\endgroup$ – Emilio Pisanty Sep 14 '16 at 10:14
  • $\begingroup$ ... and, in any case, just because it's been declined as a network-wide implementation, that doesn't mean that it can be turned off on this site. If it goes much beyond the simple flicking of a switch I would be very surprised. $\endgroup$ – Emilio Pisanty Sep 14 '16 at 10:15
  • $\begingroup$ I suppose you have a point about the validity of zero-score answers on meta sites. But I still think there are relatively few cases where a question with a zero-score answer (and no better scoring answer) can be considered resolved, even on meta. $\endgroup$ – David Z Sep 14 '16 at 10:35
  • $\begingroup$ Tell you what - have a look through the stuff it's dredged up lately and find stuff that should rightfully be brought back up. It will make a good addition to your answer. If it's more than one in five I'll eat my hat. $\endgroup$ – Emilio Pisanty Sep 14 '16 at 10:39
  • $\begingroup$ @EmilioPisanty I'm only able to find four bumped questions (without a ton of manual labor): 1 2 3 4. One of them (#3) has been resolved, but I'm not even sure why it was bumped in the first place. Another one (#2) is unresolved and was a useful bump. Another one (#4) has a good answer, but I think it's fair to put the blame squarely on the community for not voting that up. Question #1 is the only one that meets the criteria of the bumping algorithm and yet should not be bumped IMO. $\endgroup$ – David Z Sep 14 '16 at 12:44
  • $\begingroup$ So... you disagree with your answer, then? I disagree that #2 was a useful bump - it is a bad feature request that ended with the onus on the proponents doing more data digging, which they then abandoned. #4 was formally bumpable, but it was resolved (and a good example of how 'resolved' and 'bumpable' are orthogonal concepts). I agree that #1 should not be bumped, and I'm stumped as to why #3 was dredged up. In any case, I have yet to see a useful bump on meta. $\endgroup$ – Emilio Pisanty Sep 14 '16 at 13:01
  • $\begingroup$ Of course I agree with my answer ;-) though perhaps I could stand to clarify what I meant. In my answer I wanted to say that I believe most of the questions which get bumped are truly deserving of more attention, and my comment shows that of the three valid examples I found (not counting #3), I think two of them are truly deserving of more attention. $\endgroup$ – David Z Sep 14 '16 at 13:31
  • $\begingroup$ Shog9 just helped on #3 - the answer was upvoted after the bump. Still a useless question to dig back up, though, IMO. $\endgroup$ – Emilio Pisanty Sep 14 '16 at 18:15
  • $\begingroup$ @EmilioPisanty Ah, okay. On one hand, I think the OP should have returned to edit the answer to put the issue to rest; on the other hand, I do think status-no-repro should be enough reason to consider a feature-request resolved and thus not bumpable. $\endgroup$ – David Z Sep 14 '16 at 18:40

You must log in to answer this question.

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