Per this answer, the Community user autodeletes any answer that matches this criteria (among others):

The system will automatically delete closed (not as a duplicate), unlocked questions with zero or negative score having no upvoted or accepted answers or pending reopen votes, that were closed 9 or more days ago and haven't been edited in the past 9 days.

I would like to see this changed from "no upvoted answers" (which is defined here to mean has no answers with a score > 0) to has no answer with a score > 0 *and* the highest-scored answer was downvoted by more than one person.

This upholds the intent of the above autodelete rule. Specifically, this rule aims to delete bad (closed) questions with not-good answers (0 score or below). However, a single user could find bad questions with good answers (specifically, with score = 1), downvote the answer (regardless of whether or not it is good), and silently trigger the autodelete script.

In the spirit of community agreement before action (e.g. 5 close votes are required to close, etc.), my proposed change would prevent a single user from triggering the autodelete script on a question with upvoted answers.

"But this sounds hypothetical--why change the rule?" It's not hypothetical. This one user's behavior is also making some very high-rep users on our site want to leave.

  • I'd word this as 'at least two downvotes'. I do feel this overcomplicates matters though. Aug 2, 2014 at 16:02
  • 9
    Why do you want to keep a question you think is bad?
    – jscs
    Aug 2, 2014 at 19:26
  • @MartijnPieters "a single user could..." -- oh but proposed change is still vulnerable isn't it. If system is changed like that, "two users could..." do that. Wonder where such a begging for garbage protection could stop "please dear system keep all my posts where I have got a positive net rep no matter what"
    – gnat
    Aug 2, 2014 at 20:17
  • 1
    @gnat I'd be happy with two users being able to delete. I just don't want unilateral deletion.
    – apnorton
    Aug 2, 2014 at 20:20
  • 2
    @anorton given that it takes 5 users to just close the question, I would call it anything but unilateral - meaning, none of the close voters considered answer worth voting up
    – gnat
    Aug 2, 2014 at 20:25
  • 1
    @gnat Most old questions on Math.SE (where I'm from) get closed via the review queue, and most reviewers don't ever look at answers, good or bad.
    – apnorton
    Aug 2, 2014 at 20:27
  • 8
    if question is that bad that reviewer isn't even slightly curious to look at the answer(s), it means essentially the same: 5 community members decided it's not worth it (I for one regularly open reviewed questions outside of queue and look at the answers)
    – gnat
    Aug 2, 2014 at 20:34
  • 3
    It is possible that a question gets (auto)deleted without anybody whatsoever voting on it in any way. You seem to misunderstand or misrepresent the motivation of requiring several votes on other occassions.
    – quid
    Aug 3, 2014 at 0:20
  • 1
    @quid posts discussed here (9-days roomba) take 5 votes - from 3K users to close. For post to be auto-deleted without any voting at all, requirement is much much higher: "unlocked, unanswered questions with score of zero (or one if the owner is deleted), fewer than 1.5 views per day on average, and fewer than two comments after 365 days" - meaning in particular question has a full year of exposure before system gives up
    – gnat
    Aug 3, 2014 at 1:02
  • 1
    @gnat Yes, I know that but thanks for recalling the details. My point is that (massive) absence of positive feedback is considered as valid way to arrive at a "community decision" and thus claims that there is an issue with the situation described in OP (it is not a community decision as only one user decideds) are moot, since first it is not really one user that decides and second since in some other cases even nobody at all decideds in the sense used in OP.
    – quid
    Aug 3, 2014 at 9:46
  • 2
    @quid absence of positive feedback for the full year sounds like a sensible approximation of community decision to me. Anyway, in the context of this question it's moot since we discuss questions having fairly solid negative feedback: 5 votes from 3K users for closing the question
    – gnat
    Aug 3, 2014 at 9:59
  • @gnat it seems we are talking sideways but I am really not sure in which way. Anyway, neither of us seems to see a problem with the current behavior, so perhaps let us leave it there.
    – quid
    Aug 3, 2014 at 10:10
  • 5
    If all it takes is a single downvote to trigger Community deletion, than perhaps users should upvote those answers they deem helpful. All it takes is one more upvote so that it doesn't get automatically deleted. If an answer can't even garner that much, perhaps it's not worth keeping?
    – fbueckert
    Aug 3, 2014 at 14:28

2 Answers 2


I'm uncomfortable with the idea of a "good answer" to a "bad question".

If the answer is good and useful it implies that the question is actually a good question too. It might be expressed badly or have suffered from some initial down-votes before it was knocked into shape, but ultimately it must have some use if the answers are good.

I think the solution here is therefore to:

  1. Make sure that the question is good. Edit it into shape. Remove any extraneous information and bring out the things that make the answer(s) good.
  2. Upvote the resultant question so that it's no longer a candidate for deletion.
  3. Upvote the good answers so that it takes more than one person to send the question into auto-delete territory.

If the question it truly un-salvageable then I would have doubts as to whether any answers could be considered "good".

  • Although I do believe bad questions can inspire good answers, I think that this route is probably the best way of fostering a well-developed site. Thus, I am accepting this answer. (Unless, of course, my proposed change gets implemented by some odd event... :P)
    – apnorton
    Aug 3, 2014 at 3:08

I've sat and ruminated on this for awhile. I think that ChrisF's answer is indeed the right answer (saving one from one of the Community Managers to really say one way or the other).

On occasion, I do exactly this - use a down vote to try to cut down the amount of crap that is out there. I've even got a query on Data.SE (Cliff questions (v2)) to help find these questions... and there is still more crap out there.

These questions clutter search results and set bad examples - both for how a question should be asked and answered. On the bigger sites, this is likely fighting against the tide with a little bucket. On P.SE, I've got a bigger tool to use (the delete vote) and there are like minded people who actively act to try to present the best foot forward for the site (and there are also people who likely disagree with this stance - and there are avenues of recourse to address this).

The like minded people is something important to consider... This also appears to be something spilling over from Math.SE (you guys have some wonderful drama going on in your meta)... so lets look at that:

From Around how many questions get deleted in Stack Exchange and by who?

Site Name            | Mathematics | Programmers
AllDeleted           | 21419       | 9857
AuthorDeleted        | 10663       | 1748
SpamOffensiveDeleted | 102         | 46
Deleted10K           | 218         | 851
DeletedCommunity     | 10028       | 4955
DeletedModerators    | 378         | 2559

Your 10ks appear to be very, very lacking in the "lets clean up the crap", your mods aren't being 'evil'. For some reason you want that 10,028 count for community deleted to be even lower (even at that, its still only 2x of what P.SE's is). You've got 10x the questions of P.SE, 10x the questions/day of P.SE and 1/4 the 10k delete votes.

I would like to say that I would very much like to have the roomba scripts revokable delete with 10k undelete votes and not act upon questions that have been undeleted... but thats a different matter... that appears to have been silently addressed. Yeah Haney♦!

I also think that the roomba scripts should do more. Again, thats a different matter.

Failing getting a current statement from the community managers, the best I can do is find previous statements and cherry pick them. This is certainly subject to quote battles.

From A modest proposal for voting about controversial historical questions

Outside of duplicates, rare is the question that should be kept around, closed but not deleted. And we have locking for those. -- Shog9

If you think there are closed questions that should not be deleted, fix them. Make them good questions. Reopen them.

If they aren't you are going to find yourself wallowing in crap questions from history as new users keep asking "well, that is ok there..." and new answers are "You multiply before adding, so the middle term is zero, so the answer is 41" (and the like) because those are acceptable answers that have been up voted in the past and continue to display how the site is to be used.

Guess what: if your site is full of crappy questions, your site sucks - even if they're not highly-ranked by your own users, folks are finding them via Google, and that's where the vast majority of your readers are coming from. You can work to fix that - as painful as that process is - or you can bury your head in the sand and blame it on all of those stupid people from elsewhere. -- Shog9

Granted, thats in the context of hot questions... but you need to fix it too. You can burninate and blacklist the homework tag, and write a FAQ about it (heh, but nobody reads that) but that doesn't change what the site is about.

Trying to change the roomba scripts to delete less to preserve some rep of some people who got drive by up votes on questions is likely not the right way to get higher quality site content.

I say this as someone from a site that has learned this hard lesson and is continuing to try to improve its image. If Math.SE wants to be something other than MathHomework.SE, it likely needs to learn these lessons too. If this is the case, it is unfortunate that its gotten to the point where it is now before having to delete your crap and throw away those fingernail clipping quality old questions.

Yes, this answer may have gotten a little ranty.

Watching the 'Recently Deleted' in the 10k tools scroll by when the roomba runs on P.SE, I would hate to see what would happen if it was even more selective about what it deleted. I would likely be terrified of what Stack Overflow would be like in that case.

  • 3
    BTW, it appears that I can exclude math questions from the hot questions list with one weird trick... rep whorers hate me. (I'll post about this separately). Yes, you correctly judged that this thread is a spillover from the lovefest on our meta. A little cross-site exchange of opinions is refreshing, like opening a window from a stuffy room where debates went on for hours. Now linked from my answer on meta.Math.
    – user259867
    Aug 2, 2014 at 23:40
  • 1
    @900sit-upsaday given that I spend time in a chat room with gnat, I can assure you I'm familiar with the workings of the hot question list. ;-)
    – user213963
    Aug 2, 2014 at 23:43
  • 1
    I was referring to the (empirically observed) fact that questions with TeX markup in the title are excluded from the list. I just tested this by renaming two questions on the list, inserting a formula: both immediately disappeared. I know something like this was requested on Meta, but never saw any announcement of the change. By the way, Roomba deletions are reversible by 10Ks now.
    – user259867
    Aug 2, 2014 at 23:45
  • 2
    As 900 mentioned, this was prompted by the drama we've had on meta. I was reluctant to air our own dirty laundry, but I do think that it was important to "surface" from our internal debates and examine what the other SE sites felt about this issue. You've given some good insight; thank you for your responses.
    – apnorton
    Aug 3, 2014 at 3:02
  • 1
    There is a fundamental flaw in the above argument, viz. thinking that Math.SE can be compared to some other SE site. It cannot. General-level math forums are extremely unique beasts. This fact even surprised some of the designers of SE (e.g. Jeff Atwood remarked about it many times on MSE). Aug 3, 2014 at 12:58
  • 4
    @BillDubuque could you help by providing links to those differences? user:1 math doesn't seem to turn up anything and this query doesn't show much either. Every site is a unique beast, but those numbers of close rates seem to indicate you've got the lowest delete rates on the network. Yet your meta is complaining about its appearance as mathhomework.se and isn't taking steps to clean up beyond trying to sweep it under the covers.
    – user213963
    Aug 3, 2014 at 14:01
  • 1
    @MichaelT Given the "queries" that you performed, I'm not at all surprised that you have little understanding of what is currently happening on Math.SE's meta. Aug 3, 2014 at 14:15
  • 3
    @BillDubuque and on Meta.StackExchange, MSE refers to Meta.StackExchange. Glancing through his answers on meta.Math.SE, I find things like this and this and most certainly this that seem to suggest that Math.SE is just like the rest of the SE sites.
    – user213963
    Aug 3, 2014 at 14:25
  • 12
    @BillDubuque I'm not sure how going, "We're a unique special flower" contributes to your point. Please outline what makes M.SE different and requiring alternate rules outside the normal SE network. I mean, if we can make video games in general work within the framework, I don't see Math being different enough to require a raft of alternate methodologies.
    – fbueckert
    Aug 3, 2014 at 17:02
  • 1
    @fbueckert Alas, explaining that to a non-mathematical audience would require a 30 page paper, since to appreciate the points requires a nontrivial understanding of mathematics and the various ideas on how it is effectively taught (online). Certainly that will not fit in an SE comment. My views on such matters are informed by a few decades of heavy involvement in such (going back to the early days of usenet newsgroups). Aug 3, 2014 at 17:11
  • 12
    @BillDubuque So, you're going to insist that M.SE is special, but refuse to actually provide any evidence of it? You're sorta making my point for me. If you want special treatment, you're going to need to point out how and why. Just saying, "You won't understand" doesn't help you any.
    – fbueckert
    Aug 3, 2014 at 17:18
  • 1
    @fbueckert I have neither "insisted" nor "refused" anything. Rather, I have attempted to explain to you that this is an innately extremely difficult task - certainly not one that can be achieved in an SE comment (or answer). Aug 3, 2014 at 17:35
  • 6
    @BillDubuque where can I learn more about General-level math forums being "extremely unique beasts"?
    – gnat
    Aug 3, 2014 at 21:40
  • 9
    @BillDubuque if you feel you cannot explain it to a non-mathematical audience perhaps you could explain it to a mathematical audience, for example to me who does not understand the unusual uniqueness either.
    – quid
    Aug 3, 2014 at 23:33
  • 1
    The lack of review queues may be part of it. But in this related question I ask why it is nearly impossible to get 10k users to delete stuff, even really bad stuff. One answer seems to be that this isn't how most people care to donate their time, but I am convinced that this is the end of the story there.
    – tchrist
    Mar 18, 2015 at 6:26

You must log in to answer this question.

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