Downvoted, unanswered questions have been subject to automatic deletion for a long time. The same is not true of downvoted answers.

On Super User, we have almost ten thousand non-accepted negatively-scored answers posted before the beginning of this year still in existence. Over two thousand of those are scored at -2 or below, having been downvoted by multiple people.

I see no reason to keep these answers around. Based on the votes, they have helped nobody. They're just sitting there, being pointless, borderline-NAA, or otherwise bad. Low quality (but not flag-worthy) posts aren't actively harmful, but don't provide a good image for the site. Additionally, I think people would be a lot more likely to downvote bad things if they knew there was a good chance for the -1 to be refunded in the future if somebody else agreed. (And hey, the poster would get the -2 refunded as well.)

To prevent people from adding single downvotes to cause quick deletion of things they don't like, it would probably be good to require a score of at most -2 before an answer is eligible for auto-deletion. Accepted answers (that aren't self-answers) shouldn't be deleted, since they helped someone. A reasonable time frame for the deletion of -2 answers would be a month after they hit that score; the process would optimally be accelerated by additional downvotes and slowed significantly by any upvotes.

I understand now that some downvoted answers should stick around to show an example of something subtly bad. To prevent those from going away, the presence of numerous comments or somewhat-highly-upvoted comments could be a factor against deletion. I still believe that many downvoted answers are just mediocre/useless and not representative of any common misconception. (See the Super User search linked above.)

Of course, such deletion shouldn't apply to meta sites, where downvotes are overloaded. I think both child metas and Meta Stack Exchange should be excepted from answer auto-deletion.

Otherwise, please make the Community user delete negatively scored answers after a while.

  • 5
    The usual argument for leaving downvoted answers around is so their obviously-bad reception will discourage garden-path mistakes. Downvoted answers are a pretty subtle way to do that, and not the only one; making sure at least one upvoted answer describes the downsides of the approach each such answer attempts is a more straightforward approach, but it requires more general attention, in place of the organic, low-profile voting that the first approach relies on. How does your proposal accommodate the extra need for this? – Nathan Tuggy Feb 27 '16 at 2:45
  • I'd suggest that views be the windowing factor, not (or at least along with) time: only the negatively-scored posts that people are actually seeing are a problem. – jscs Feb 27 '16 at 3:02
  • @NathanTuggy: that's really outside the scope of this proposal, and is a cultural, not a technical, problem. If people are using downvotes to mean "wrong, but useful as an example" in some circumstances, and "not useful/just plain wrong" in others, they're not just going to confuse the system, they're going to confuse readers, too. Editing the "garden path" answer to be explicit is the only correct move there. – jscs Feb 27 '16 at 3:07
  • 1
    @JoshCaswell: You mean you think using downvoted answers as examples of what not to do is not what downvotes are for? I don't think that's a popular viewpoint, and frankly, I personally have no trouble with overloading voting just slightly. "This is a bad idea" and "this is a bad idea, don't think it's a good one" are really remarkably close together. – Nathan Tuggy Feb 27 '16 at 3:20
  • 1
    If the procedure outlined in an answer is a terrible idea, but you think that having the post around as an example is good, then it sounds like you think the answer itself is useful, which is the generally-accepted criterion for upvoting. And again, editing it to say "This seems like a good idea, but it's not" very easily and clearly removes any ambiguity or confusion about how to vote. – jscs Feb 27 '16 at 3:39
  • 1
    @NathanTuggy I was under the impression that warning against bad things was to be done by good answers. Maybe highly upvoted comments could stop deletion? There are a boatload of useless answers that are literally an example of nothing and should vanish, yet are not flag-worthy. – Ben N Feb 27 '16 at 4:20
  • @BenN: Hmm, the idea of having comments stop answer deletion is highly impure, ideologically, but it has a certain hackish charm. It might well work. But it's definitely a hack. – Nathan Tuggy Feb 27 '16 at 4:22
  • 2
    @JoshCaswell: Editing an answer the author intended to be instructing how to do things the right way to say it's the wrong way seems like the definition of editing against original author's intent. Upvoting an answer that is not useful in solving the problem is equally inverted: their usefulness lies in not being used to solve the problem. – Nathan Tuggy Feb 27 '16 at 4:23
  • @JoshCaswell: In any case, though, the point I was trying to make is that there are quite a few who consider, and have always considered, deliberately not deleting downvote-worthy answers to be The Thing To Do for the precise reasons I mentioned; whether right or wrong, either this post needs to disprove that for good (a tough challenge) or a different post should take on that job (still tough, but a cleaner separation). – Nathan Tuggy Feb 27 '16 at 4:26

I disagree.

What about answers that do work for example but expose a bad design or security flaw. The downvotes and comments on such post actually contribute something very useful: don't do it this way!

Downvoted answers serve a purpose. < 10K users can't even see deleted answers at all. There is a possibility they will come up with the very same answer before knowing it had been deleted. This is not useful for either one.

Total nonsense answers should be deleted and we have flags to facilitate that process.

| improve this answer | |
  • Mmm, I see the importance of showing that subtly bad answers are bad. But what about the thousands of answers that are pointlessly mediocre? Maybe it's just Super User, but we have boatloads of product recommendations or "for me it was <obscure barely related thing>" that are just good enough to be not flaggable but bad enough to be pointless. How should we deal with these? – Ben N Feb 27 '16 at 18:10
  • Using downvotes to mean "wrong, but useful" is needlessly confusing. A negative score should mean "this content is not valuable". If the content is valuable as an example of something not to do, then it should be edited to make that explicit and not downvoted, or another new answer posted explaining the situation. – jscs Feb 27 '16 at 19:14
  • 1
    Moreover, your viewpoint argues that a particular cultural side effect is a substantial objection to treating downvoted posts categorically as "bad", but it is itself subject to a side effect: what happens when the user who made the post that you downvoted but want to keep decides she doesn't want the negative score hanging around her neck any more, and deletes the post herself? – jscs Feb 27 '16 at 19:14

Downvoted answers serve a function, as already pointed out. In addition, the community already can remove downvoted answers that don't qualify as "not an answer" or "very low quality". Users with 20k rep can cast delete votes, which do not expire, and three such votes remove an answer. 10k users can see a collection of posts with pending delete votes, and you can use chat and meta to draw attention to specific posts if you like.

The Community user, being a bot and not a human, can't judge whether a particular downvoted answer is pure nonsense or a valuable "don't do this" warning. But humans can judge that and have the ability to act, so if you have an accumulation of junk answers on your site, try enlisting the community rather than Community.

| improve this answer | |
  • I guess I have to keep repeating this, since everyone else is repeating the opposite view: if a post is valuable, it should not be downvoted. Make the "don't do this" explicit, and upvote the useful post. – jscs Feb 28 '16 at 19:04
  • @JoshCaswell good point, assuming the author accepts that edit. (I've seen cases where people insist that their answer is the right thing to do despite community objections. This happens more on subjective-leaning sites, probably.) – Monica Cellio Feb 28 '16 at 20:08
  • 1
    In that case, I believe the right thing to do would be posting another answer with the same proposal debunked or refuted: "A [commonly-repeated suggestion](link-to-wrong-answer) is that pet frogs should be fed beef hamburgers for every third meal, but this is not correct. Dr. E.H. Harriman, U.N. Herpetologist-in-chief, published on this just last year..." again making the warning explicit. – jscs Feb 28 '16 at 20:20
  • That certainly works (and, to come full circle, means the community can delete the original, downvoted, wrong answer if they choose). – Monica Cellio Feb 28 '16 at 20:29

You must log in to answer this question.

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