There is an automated system that eventually deletes all unanswered negative scoring questions (as long as they're not locked).

I understand why this exists for regular sites – a negative score indicates that there is something wrong with the question, and if the question retains its negative score for that long, then it might not be worth keeping on the site. However, on meta sites, a negative score – particularly when it comes to s – is merely indicative of disagreement. A question can have a very negative score without there being anything wrong quality- or research-wise with the question. It might merely be a suggestion that more people disagree with than agree with. Currently, even if the disagreers only have one more member than the agreers, the question will still be automatically deleted.

This does not seem like a good system. Questions that are otherwise fine may be getting deleted just because one or more people disagreed with the suggestion than agreed with it. In fact, I just noticed a case where this happened. This question has 10 upvotes and 25 downvotes for a total score of -15, and was therefore automatically deleted under the policy of RemoveDeadQuestions. However, there doesn't appear to be anything wrong with the quality of the question (full disclosure: it's my question); it's just that more people disagreed with it than agreed with it. By deleting this post, the record of the discussion is lost (except to those who have 10,000 reputation).

As the record of the prior discussion is lost, it also means that users who later come up with the same idea will not be aware of the prior discussion, or the fact that it had been proposed before and received negatively, and may sometimes be hit with an immediate onslaught of downvotes simply because they didn't know of the prior outcome. Also, such questions cannot be closed as duplicates, impeding content curation - users will have to repeatedly explain the reasons why the request is not a good idea.

I think it would make sense to not apply the policy of RemoveDeadQuestions to Meta questions. Or, at the very least, the criteria should be tweaked. Perhaps if a question has a negative score and no upvotes it should be deleted, as there it is more likely that the negative score is because it's a bad-quality question. But when the question has a significant amount of upvotes it should not get deleted simply because one more person voted against the suggestion than in favor of it.

Some may argue that since nobody bothered to answer the question (which stops it from automatic deletion), the discussion isn't valuable, but in many cases, there may be a valuable discussion in the comments, and as Monica's answer states, feature requests are expected to make their case in the question, not in answers. Even if the author or someone else edits the question to have better arguments, the question would still be automatically deleted if it doesn't accrue enough upvotes to counter the downvotes.

(I am aware of this question which asks if RemoveDeadQuestions applies to Meta. The point of this question is to suggest that that should not be the case.)

  • 7
    Related, maybe duplicate: Turn off the roomba for child meta sites
    – Laurel
    Commented May 23, 2019 at 4:42
  • Just to clarify, are you requesting this to be added to all meta sites or just Meta Stack Exchange? If you're talking about all meta sites, it might be better to tag it [per-site-meta] instead of [meta].
    – Picachieu
    Commented May 23, 2019 at 16:12
  • Could you hint what the deleted request was about? I don't have 10K and although it's not absolutely essential to know what the post was about, it would help me and other regular users and contributors who don't have that kind of rep, if it does deserve to be undeleted/preserved. Commented May 23, 2019 at 17:02
  • 1
    @Mari-LouA here you go
    – rene
    Commented May 23, 2019 at 19:46
  • @rene thanks, the request is well argued. I never had any problems with overcoming the minimum limit when I was a newbie but then again, maybe on EL&U it's easier to fix posts by using italics, block quotes, line breaks etc. Commented May 23, 2019 at 21:36
  • 1
    Another variation of this was asked on MSO about staff announcements specifically.
    – Laurel
    Commented Apr 17, 2023 at 1:27

3 Answers 3


I agree; there's a big difference between +0/-15 and +10/-25 (or +100/-101). Controversial is not the same as uniformly rejected. Sometimes the controversy might not even be about the question itself but rather about other intertwined issues.

In a related question, Shog declined a request to turn off the Roomba entirely on child metas, arguing that somebody should have written a supporting answer (that gets upvotes) to prevent the deletion. That's a workaround, but it seems like unnecessary gymnastics to me and, worse, it runs counter to how we usually do feature requests. Specifically, we expect the FR to make a case in the question, and "yes I agree" answers aren't useful. If we move the "payload" on FRs into answers, then the question itself becomes weak and that will lead to more downvotes!

A pile of upvotes demonstrates (some) support; that should factor in.

  • 1
    But specially for feature request we should optimize for pearls and keeping those mixed signals FR's around is keeping too much sand to work with. Maybe for bug reports, but I'm meh about it.
    – rene
    Commented May 23, 2019 at 14:25
  • @rene controversial FRs sometimes signal real underlying problems where the community disagrees with the specific feature proposal. (Don't have examples handy, but I've definitely seen this.) I'm not sure we want to automatically lose that; communities can already delete these questions themselves if they're really not valuable. Commented May 23, 2019 at 14:27
  • I was trying to think of a formula for a threshold that seemed reasonable to me. I note that in your +10/-25 case, the positive votes account for 28% of the total, while in the +100/-115 case, it's 47%. I certainly don't think that the latter should be automatically scrubbed—but I'm quite as convinced about the former. I'd say a threshold of at least 40% positive should keep it safe. Perhaps that number could decrease as the total number of votes increases. For instance, even if +10/-25 should be deleted, I don't think the same should be true for +100/-250—given how popular it was. Commented May 23, 2019 at 14:38
  • Also, if it's your own bug report that was downvoted, what should I put in a self-answer? "this bug should be fixed"? Commented Aug 28, 2019 at 17:21
  • 1
    @JasonBassford How about exempting them if the author would earn a net positive reputation from the post? Commented Aug 28, 2019 at 17:22
  • @SonictheAnonymousHedgehog That would only give a threshold of 20% positive (far too low as far as I'm concerned), and it also wouldn't take volume into account. Commented Aug 28, 2019 at 17:38
  • @JasonBassford No, it's 40% positive. Keep in mind that question upvotes only award 5 rep, not 10. Commented Aug 28, 2019 at 19:10
  • @SonictheAnonymousHedgehog Ah, that's right. (I rarely ask questions, so I'm primarily used to answer reputation.) That would cause the following to be deleted, but they could each be saved by a single upvote: +1/-3, +10/-25, and +100/-250. That doesn't seem immediately wrong to me. And certainly better than basing it purely on the total score. I note that this is actually 28% positive, not 40%, if I look at percentage of total votes. (And the 10 points of reputation, had that been true, would have been 17%, not 20%.) Commented Aug 28, 2019 at 19:38

Highly controversial posts, the type linked in Monica Cellio♦'s answer, usually attract a lot of response by those who either strongly agree or disagree. The post will also draw in the crowds and literally generate a bunker of comments.

If these elements (multiple answers, high view counts, a plethora of comments) are missing from a feature request, doesn't this also say something about the FR? If no one was motivated enough to post an answer explaining why the FR was or was not feasible, I'm not sure that a heavily downvoted request is really worth keeping.

What would be a heavily downvoted post is up for debate but taking into account the upvotes and downvotes cast, I reckon that would be a -20 score. So with such an unpopular post and no answers, maybe automatic deletion is an act of mercy.

A slightly different issue are posts that are tagged discussion, they too generate controversy but they might also scare potential answerers from posting contributions for fear of losing rep, so you might well have a valid post tagged discussion, with a good mix of upvotes and downvotes but with no answers posted.

P.S. I'd like to add I have not the slightest clue what the OP's request was about as I don't have 10K. Maybe the idea was pretty decent or unpractical, who knows? Well, users with 10k do.

P.P.S. is there a FR somewhere that proposes users who have earned at least 10k on SE sites but with only 3-5K here on on Meta, the privilege to see deleted posts on Meta? As the OP stated

I think it would make sense to not apply the policy of RemoveDeadQuestions to Meta questions.

I mean, Meta is different, right?


I see no reason why an unanswered question with a score of -15 should not be deleted by the Roomba.

If a question advocates something which the community is against, and no one has made the effort to write an upvote worthy answer, then removing it from the site seems appropriate to me.

If you wish to avoid having such questions deleted in future, then you could try writing a question that simply outlines an issue in a neutral way, and then quarantines what you think is a good solution to it in a self-answer. As long as your or another answer attracts upvotes the Q&A will not be touched by the Roomba.

  • 5
    But should a question with +50/-51 score be automatically deleted? I don't think so, since we don't decide things by majority vote, but by consensus. Commented May 5, 2020 at 20:07
  • 1
    @SonictheStay-HomeHedgehog when clear consensus has not been reached after 30 days when the Roomba kicks in then I can live with the slate being cleared on what it asked, and if anyone thinks they can gain clear consensus with a different wording they are free to ask a new question.
    – PolyGeo
    Commented May 5, 2020 at 20:22
  • 2
    Retaining a record of which proposals were rejected by the community is surely a valuable thing, since it allows people who have the same idea to search and find that the idea has been proposed before and was rejected by the community.
    – kaya3
    Commented Jun 26, 2023 at 16:31

You must log in to answer this question.

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