The Roomba process eliminates closed, negatively voted, etc. types of questions after weekly, monthly, yearly checks. However, I do not understand why the similar checks are not applied to answers.

For example, if today I google for "Is the t-butyl carbocation more stable than the benzyl carbocation?", I would get this Chem.SE hit. Now, obviously, my query is answered in the accepted answer, as well as the other two answers, but why is the fourth answer left there for me to see? It merits deletion, because it is:

  • negatively voted: indicating that it is low quality or clearly wrong
  • abandoned: user has not shown effort to improve it, it was last edited four months earlier

You can see a similar case here, and in many other questions on many other SE sites.

My question remains the same: why aren't old, abandoned, negatively-voted answers auto-deleted? Clearly they don't add any value to that question, so why not delete them?

(I have tagged this because I am interested in the philosophy behind the decision; however, if multiple users agree on the fact that such answers should be deleted we, may edit this into a )

Dan made a point that "negative scoring answers are useful as a signal of what’s known wrong." However, consider that negatively voted answers are already greyed out below a vote threshold. So, it's clear that the admins at SE don't want us to read them.

That, and if there's any topic where a concept has a possible pitfall, it should be made clear in the answer, in such a format: "This is a popular pitfall: ... Why people think it's correct:... Here's why it's wrong" so that it adds more value to the question. The current trend is mostly a bad/wrong answer and then discussion/argument in the comments about its incorrectness, which does not help future readers.

  • 2
    The cliffs notes are that negative scoring answers are useful as a signal not only of what’s known to be right, but what’s known wrong. The obvious example is real etymologies (upvoted) and common-but-fictional folk etymologies (downvoted: no, dear reader, I know you’ve heard this asserted before, but it’s steaming). This quality of wrong answers doesn’t translate to questions: wrong questions are mostly nonsensical, too broad to add any value, etc. Words that can’t teach anyone anything.
    – Dan Bron
    May 21, 2018 at 5:30
  • @DanBron Thanks for your comment. Sorry but I disagree, and I have expanded my question to elaborate on that. May 21, 2018 at 5:38
  • Thanks, I’ll come back and read the update tomorrow, and remove or add comments as needed. I think you’re asking a fine question. I’m just giving you one user’s opinion, which could easily be wrong.
    – Dan Bron
    May 21, 2018 at 5:40
  • The fact that an answer is greyed out doesn't mean that the admins don't want us to read it. It just means that its not considered useful in the sense of actually answering the question. If the admins really didn't want us to read it, then it wouldn't be visible. May 21, 2018 at 7:14
  • I'm speculatively certain its also cause no one thought it would be a problem. May 21, 2018 at 7:30
  • @JasonBassford Greying out stuff makes it harder to read. If they only wished to convey that the answer is not considered useful, the negative vote count would have been sufficient. May 21, 2018 at 9:28
  • If you can't emit a VLQ flag, but the question is really very bad, you may flag it for mod attention. Furthermore <-3 answers can be deleted by 3 20k+ users. After the first delete vote, the answer will be visible for them on the 10k+ admin pages. If all of these don't happen, maybe the answer is really not soo bad. It may also happen, that the answer is not bad, but the majority of the community disagrees it, and this is why it is voted deeply down. Pluralism dictates for the sites to show also minority opinions, even if they are downvoted.
    – peterh
    May 21, 2018 at 14:30
  • @peterh these are indeed some good points. On science sites like Chemistry, it may make sense to delete the wrong answers (because there can be either a right or a wrong). But on more subjective sites like Politics/Philosophy/etc., many negative votes may also mean that the community disagrees with the view. May 21, 2018 at 14:44
  • @GaurangTandon I think in this case, the delete votes of 3 20k+ users could solve the problem, if the answer survived the initial period of the VLQ flagging. Unfortunately, there is no review queue for the delete votes, maybe the SE doesn't want to give so much power in the 20k+ hands. There were many initiatives on the MSE to have delete review queues on the site, but it didn't happen until now. The SE doesn't want it. I think the chemistry SE is trustable in this sense, that you won't misuse it, but other sites are not.
    – peterh
    May 21, 2018 at 14:51
  • @GaurangTandon There are also "delete voter chatrooms", there 20k+ users give eachother tips, where to vote. They serve essentially as an "unofficial delete review queue". As far I know, the SE dislikes it, but they don't intervene until there is no clear misuse.
    – peterh
    May 21, 2018 at 14:54
  • @GaurangTandon I think you could create a "delete voter chatroom" and then advertise it on the meta site.
    – peterh
    May 21, 2018 at 14:55
  • @GaurangTandon The search function of the answers is continuously attacked on the meta sites, saying that it is bad, but the truth is in my opinion, exactly the opposite. For example, here you can find the answers below -3 posted in the last week. There you can find your targets.
    – peterh
    May 21, 2018 at 15:00
  • @peterh Thanks! This was certainly very helpful. I'll see what can be done here. May 21, 2018 at 15:36
  • @GaurangTandon My pleasure :-) What I see on the Chemistry SE, that you have very few 20k+ users. This may be the reason of the lack of the answer-delete activities.
    – peterh
    May 21, 2018 at 15:40
  • @peterh "that you have very few 20k+ users." that is precisely the reason why the thought of Roomba occurred to me in the first place, as I was thinking of ways other than the delete votes to get stuff deleted. On a larger science site like Mathematics, getting wrong answers deleted is not at all a problem, because they have many active users, and a chatroom like CRUDE. However, on a low activity site like Chem.SE, stuff like this is hard :( May 21, 2018 at 16:04

1 Answer 1


In a practical sense - there's many more signals to whether a question is low quality (that it's closed for reasons other than being a dupe, is low-scored and so on) than with answers.

In a sense, I assume it's less of a problem - if a question has other good content associated with it, that content floats to the top, and folk can and do flag content for low quality or other issues. (and if it attracts poor quality answers, the answers are gone with it as part of a roomba).

I also suspect that answers that are wrong, but not deleted prove to be useful - we don't delete answers that attempt to answer a question for being wrong in many cases, and they serve as what they'd call in the military to be a "Negative Example"

Yanno what would add value to the downvoted answer? A comment going "This doesn't seem right - (quick reason)". The downvotes show that it's wrong, and the other excellent answers are what people would see.

There's no real broken window effect either - so there's no clear benefit from deleting this specific instance automatically.

  • I think I already considered your point "Yanno what would add value to the downvoted answer? A comment going "This dosen't seem right - (quick reason)"." If we want to tell users that there exist wrong answers, a better way to tell them so should be: "This is a popular pitfall/wrong answer: ... Why people think it's correct:... Here's why it's wrong..." directly in the answer box, instead of making the reader to understand the comments to try figure out what's wrong. May 21, 2018 at 9:24
  • Also, sorry, but I could not get what you meant to convey by the "broken window" effect in this context. May 21, 2018 at 9:27
  • There's a theory that places with broken windows have more crime. Likewise - one of the goals of the roomba is to weed out low quality, or at least low interest questions automatically improving the overall quality of content, least in theory May 21, 2018 at 16:28
  • "In a sense, I assume it's less of a problem" I remember a case where a user used a -5 scored answer that broke its system... I never presume anything anymore
    – Braiam
    May 21, 2018 at 20:44

You must log in to answer this question.

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