15

I'm curious to know if there is a particular time to delete - or not delete - an answer that's clearly wrong?

Consider the first revision of https://unix.stackexchange.com/revisions/420788/1, which aside from a lack of formatting, contains a fragment of (presumably untested) code. The code attempts to answer the question so it's on topic, but it's incorrect in at least one significant way.

Ignoring the lack of formatting, I cannot find a text that explains whether to vote to delete such questions or accept it because it does attempt to answer the question.

  • Accept and move on, knowing the answer is on topic but faulty
  • Downvote and/or comment, and move on, knowing the answer is on topic but faulty
  • Vote to delete, to try and clean up "wrong" answers
  • Spend the time trying to fix what the author might have intended (potentially tricky without hitting the prohibition on changing the author's intent)
12

Well, it’s on another site’s meta (Stack Overflow), but You're doing it wrong: A plea for sanity in the Low Quality Posts queue, written by a moderator, with over 200 votes and minimal dissent, says that wrong answers should be downvoted (ideally with a comment) and not deleted.  And it is an amplification of Your answer is in another castle: when is an answer not an answer?, which is on the main Meta, is written by a moderator / community manager, and has over 500 votes.

If the answer is wrong but very close (i.e., a good approach but with a typo), it’s probably OK to edit it.

  • The top answer to that makes a good case for deleting posts that give code without any explanation. The example here is just that, and incorrect to boot. – muru Jan 30 '18 at 22:44
  • @muru: (1) Well, it says that some answers with no explanation should be deleted, but it looks more like an opinion to me; I don’t see a “good case”.   And I don’t particularly agree with it.   (2) I believe that low-quality questions deserve low-quality answers.   The author of the question didn’t put much effort into it, and is clearly asking for a command; I don’t see why an answer that provides only a command isn’t sufficient.   … (Cont’d) – G-Man Jan 30 '18 at 23:28
  • (Cont’d) …  (3) I believe that this isn’t really relevant here, but was the answer wrong other than the trivial typo? (4) The other answer to the question (which is just a light-weight synopsis of tante’s answer to a question of which this one is (maybe) a duplicate, isn’t any better. It is a command line with a little window dressing but no real explanation. … (Cont’d) – G-Man Jan 30 '18 at 23:28
  • (Cont’d) …  And it adds the minor flaw of taking commands that were on separate lines and putting them on the same line, separated by ; instead of &&.  That answer should probably be deleted because (a) if a question is a duplicate, it should be flagged/closed as a duplicate. Copying answers from the other question is not appropriate. And (b) it (sort-of) copies tante’s answer without explicitly identifying the original author. – G-Man Jan 30 '18 at 23:28
  • @G-Man ref (2). the OP did at least bother to try a couple of variants, even if neither was successful. There are (way too many) other questions where the person asking couldn't even be bothered to do that. Maybe I'm lowering the mental bar too much... :-( – roaima Jan 30 '18 at 23:37
  • 1
    @roaima: The Emperor has no clothes!     Am I the only one who noticed that the OP’s “couple of variants” are identical? – G-Man Jan 30 '18 at 23:45
  • 2
    Where's the markup for shakes head in despair and looks for the close-because-it-s-just-incomprehensibly-stupid button? – roaima Jan 30 '18 at 23:47
  • 1
    @roaima I wouldn't give much weight to OP's couple of variants (even if they weren't identical). I have felt that there's a recent trend to throw random crap dressed in code formatting in questions as a way to avoid "what have you tried" comments. In particular, I strongly believe that this particular question was edited to add the "second" try purely for bumping. – muru Jan 30 '18 at 23:55
  • 1
    @G-Man an answer that only provides a command is insufficient because it's no better than a comment. We want not only high-quality questions but high quality answers that can serve as a useful reference for other people with similar questions. What the OP asks for isn't always relevant or important. BTW, comments can get deleted any time, I don't see any reason why comment-level bad answers shouldn't be. – cas Feb 11 '18 at 5:36
8

When considering whether to delete an answer, I tend to follow the available deletion reasons:

  • commentary on the question or other answers
  • asking another, different question
  • “thanks!” or “me too!” responses
  • exact duplicates of other answers
  • barely more than a link to an external site
  • not even a partial answer to the actual question

(Note though that the help centre doesn’t present this as an exhaustive list...)

In this instance, the answer given was

Maybe this will work, curl URL > /path || tar -xvf /path/file.tar.bz or whatever is the extension

Importantly in my assessment, it had a comment,

The || is a logical 'or' operator. It will cause the tar command to only be executed if the curl command returns a failure exit code. I asw your answer only because I offered to help new contributors to this site by offering advice, so I guess the advice would be to try out your answer before posting it. You may have meant to use the logical 'and' operator &&.

The answer is really just a comment in my book: it’s tentatively suggesting a solution, without much effort. However deleting it would drop the comment, and the writer would be unlikely to learn anything from the exercise...

To answer your question, I think the correct reaction to incorrect answers which are actually answers (unlike the one we’re discussing here) is either to edit them and salvage them, or to downvote them, perhaps with a comment, and move on. Trying to delete incorrect answers is tempting, but it’s a slippery slope.

  • Thanks. I'd written most of my question here before you picked up on my incorrect approval of the broken edit. It (the deletion puzzle, not your comment) is something that's been puzzling me for a while now. – roaima Jan 30 '18 at 23:43
  • Good point about losing the comment. That comment is valuable advice. – Wildcard Feb 2 '18 at 5:09
  • Upvoted because I take the same approach. I just thought I'd mention that a poster can still see their own posts (and any comments on them) even after they've been deleted. That's why I usually comment when voting to Recommend Deletion in the Low Quality Posts review queue. – Anthony Geoghegan Feb 2 '18 at 9:55
  • Good point @Anthony, thanks. Would the writer still get notified? What happens if the answer is deleted before they see the notification? – Stephen Kitt Feb 2 '18 at 10:16
  • Searching on Meta Stack Exchange indicates that users get notified when a moderator comments on their deleted posts. I presume that question refers to diamond moderators but it shows it's possible that users also get notified of comments by non-moderators: meta.stackexchange.com/a/301722/229493. Update: This answer clarifies that they won't get notifications of non-moderator comments: meta.stackexchange.com/a/231990/229493 – Anthony Geoghegan Feb 2 '18 at 10:59
  • 1
    Another update: Because I could still see my own self-deleted answer, I presumed that all users would see all of the posts that they “own”, regardless of whether they’ve been deleted or not. Further research shows that they can’t view posts that are deleted by moderators: meta.stackexchange.com/q/135362/229493 I’d advocate that this Meta feature request be upvoted as it seems like common sense and I’m disappointed that this isn’t already the case: there’s little point in us spending time and effort educating new users if they may not ever see our feedback. Sorry for the confusion. – Anthony Geoghegan Feb 2 '18 at 11:24
  • Users will receive notifications of canned comments left by reviewers in the Low Quality Posts review queue – even if the post is deleted. However, if you take the time to leave a more detailed comment that is tailored to that specific post, the user will not get a notification: meta.stackexchange.com/a/143640/229493 – Anthony Geoghegan Feb 2 '18 at 11:38
3

I think an answer that provides some code or command as a solution should also provide either of:

  1. An explanation of what the code does, or
  2. An explanation of why the author thinks that code will work

Without either of these, an inexperienced visitor to that answer has little hope of correcting or even spotting mistakes in the code. Experienced people could correct the code, but they may not be able to tell the author where they went wrong, which could be in multiple places.

We have enough sites on the internet filling up with posts that go "maybe try this [one-liner/100-line monstrosity without comments]" without any explanation as to why I should try it.

  • 1
    So, are you suggesting a delete vote for "maybe this..." or comment-less code is acceptable, and possibly even recommended? – roaima Jan 30 '18 at 23:33
  • 2
    At the risk of being pedantic, this is a comment that does not answer the question.     :-)    ⁠ – G-Man Jan 30 '18 at 23:35
  • @roaima (in the case of "maybe this...", if there's no indication of what this is or why maybe this). I routinely downvote and vote to delete answers that give code without explanation. – muru Jan 30 '18 at 23:35
  • 1
    @G :) not directly. When I say that an answer should do X, the implication is that answers that don't do X aren't worthy of being called answers and should be nuked as such. – muru Jan 30 '18 at 23:36
  • Why not do the same for answers that are wrong but have a persuasive (but wrong) explanation? – G-Man Jan 30 '18 at 23:37
  • @G-Man then you have an indication that the poster has some understanding of what's going on and an opportunity to improve that. In that case, I downvote with a comment, and VTD only if OP shows no intention of fixing it after a couple of days. – muru Jan 30 '18 at 23:50
  • 1
    @G-Man, in my opinion, a wrong answer with a persuasive wrong explanation will ideally wind up with many downvotes, a clear comment explaining why it is wrong, and many many upvotes on that comment. NOT deleted. In that form, it's a lesson for future readers. But I would agree with muru that a monstrous 100-liner without even the slightest attempt at explanation that is also wrong, should be deleted if the author isn't willing or responsive about fixing it. – Wildcard Feb 2 '18 at 5:08
2

I'll vote to delete a bad answer that comes up in the review queue if it's dangerously wrong and either unfixable or if whoever answered it shows no inclination to fix it. I probably wouldn't flag it for deletion myself unless it also met one of the other deletion criteria.

I'm unlikely to edit the answer to fix it unless it was a very trivial error, a typo or similar.

I'm even less likely to edit a command-only answer (whether correct or not) to add an explanation. The boundary between "helping" and "subverting authorial intent" is far too blurred.

If the question or topic was interesting to me (and if none of the existing answers covered ground I thought was important), I'd be more inclined to write my own answer where I could write whatever I liked without having to be concerned about what the original author intended. If it wasn't interesting, I'd either vote to delete or click on "Skip".

BTW, I don't think that even a great comment on a really bad answer is a good enough reason to not delete the answer. Comments are ephemeral, even highly-upvoted comments, and some readers don't read comments (or explanations) anyway - they just see something easy to cargo-cult into their shell.

  • I like your "dangerously wrong" – roaima Feb 11 '18 at 13:11

You must log in to answer this question.

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