15

It is not infrequent to see controversial reviews of specific kinds of answers in the Low Quality Posts review queue. Those answers are often short, made at most of a few commands or links with few to none words of explanation; they usually fail to provide a solution to the asker's problem (e.g. propose a tool not suitable for the job) and/or are not just slightly wrong (e.g. cannot be corrected by changing an option to a command).

To be clear, I'm not referring to the (many) answers reviewed in the LQP queue that are easily identified as variants of "not an answer": gibberish, link-only, "thank you", comments, new questions, "I have this problem too", details from the asker that should be added to the question, question bumping1.

These categories are collectively defined and addressed, among other places, in the Help Center, in the guidelines for reviewing on Meta Stack Exchange, on Meta Stack Overflow (with nice examples) and as the "recommend deletion" canned comments2. The prescribed action is quite unambiguously to delete or recommend deletion, there is not much disagreement around them and their review usually doesn't pose much of a problem.

Now, apparently there is strong consensus that we should avoid deleting poor and wrong answers, as stated on Meta Stack Exchange (Guidelines for reviewing low-quality posts, How to handle low quality answers in LQP queue?, "What are the criteria for deletion?" in the FAQ article about deletion, When should I delete an answer?) and Meta Stack Overflow (You're doing it wrong: A plea for sanity in the Low Quality Posts queue).

Along the same lines is also the accepted answer to When to vote to delete an answer? on Meta U&L, which this question looks tightly related to.

On the other hand, many posts seem to somehow advocate deleting "low quality answers", and some users in our LQP review queue show a similar sentiment.
From the guidelines for reviewing low quality posts:

[A] Answers that fail to address the question: If you evaluate the answer such, first check carefully whether there is a lack of clarity in the question that you and the answer’s author may have interpreted differently. Otherwise recommend deletion. Leave an explaining comment in both cases.

In Shog9's words:

Does that mean these answers should forever hang around the site? No, not necessarily - if it turns out they're just not that useful, they should probably still be removed - or at very least, down-voted so that they rank below other answers.

In Undo's words:

Only if you can't plausibly imagine anyone putting in the work to fix the post should you opt to delete these kinds of answers.

Also, the answers to that question outline some reasons for deleting certain kinds of answers that are "low quality" without being "not an answer".

In Adam Lear's words:

"Wrong" as in "completely unrelated to the question"? Go ahead and delete those.

All this outlines a grey area, which stretches between deleting only those answers that cannot be read as an attempt to provide a solution, no matter how useful, and deleting anything that unsalvageably fails to convey useful information.

What criteria do you follow when reviewing low quality answers? Are there answers you think should be deleted despite not being "not an answer"? Is the LQP review queue a suitable place for such a cleanup work?

Being able to agree on some specific guidelines for our site would be beneficial to reviewers (less work for them in the queue), to answerers (a more clear message on what they did right/wrong) and to all users trying to learn the art of reviewing.


1 As an aside, Meta U&L has discussed code-only answers, e.g. here and here. They are not "not an answer", nor are problematic in the scope of this question.

2 Link-only answers have been addressed on Meta U&L, too.

  • 1
    I think Manishearth had a point. I.e. the LQP review system as it exists, nudges some people towards deleting low-quality posts that are nevertheless technically answers. (Clue in the name :-). I don't want to have too high expectations of (re-)discussing this in Meta. Maybe U&L is smaller enough, that reviewers & flaggers can smooth off some more sharp edges than SO does. (Or choose to file the edges sharper, if that is the consensus :-). – sourcejedi Jun 26 at 12:47
  • "Common principles for the site" might not be quite the right phrase. I.e. at best it would gather consensus among the reviewers (and flaggers, for the manually-flagged posts in LQP) who look in Meta, but there will still be some users who do not agree, and not very tight enforcement of it. – sourcejedi Jun 26 at 12:50
  • 1
    @sourcejedi In my opinion the proposal in that Q/A makes a good point: let reviewers classify posts instead of deciding what to do with them. Though it seems to take a different perspective than that of my question, focusing on how the LQPQ can be made less confusing given that only answers that are not functionally answers should be deleted. About my phrasing: surely it can be improved, I'm open to suggestions (especially considering that I'm not a native English speaker and nuances may be important here). – fra-san Jun 26 at 20:05
  • You called me out, I guess I am being too nit-picky about the phrase specifically :-). But, the "guidelines" you linked to look like they are "just" a community post on the main StackExchange Meta. Maybe you could steal that word, i.e. "guidelines for our site". IMO "guidelines" sometimes has that connotation, that not everyone is forced to follow them. – sourcejedi Jun 26 at 20:45
  • @sourcejedi I'm sorry, it wasn't my intention. I see your point about "principles" VS "guidelines" and I edited my question. – fra-san Jun 26 at 22:37
  • LQP doesn't offer the option to downvote. Could this be added? To do this rather than vote to delete requires me to open the Q/A in a new window and downvote it there - possibly with a comment - before returning to the original review queue. Where it's a wrong answer but "plausible" I would prefer to downvote/comment than delete – roaima Jun 28 at 15:54
  • 2
    @roaima This is a good point, especially if we consider that downvoting is suggested by the guidelines for reviewing. But I see that (down)voting in the LQP queue has been status-declined. Also, that queue seems not to be designed to make it easy to engage with poor but salvageable posts (unfortunately, IMO). – fra-san Jun 28 at 19:11
7

For me the line in the grey area is drawn depending on the effort a user puts into their answer. If the answer is (1) clearly factually wrong or misinterprets the question and (2) it's a one-liner, I don't feel there's a point in keeping it.

If the one-line answer looks wrong but you're not 100% certain, comment, downvote, but click on "Looks OK". Give people who know the topic a chance to see it.

If the answer is clearly wrong but the user has put some effort into it, comment, downvote, but click on "Looks OK". Give the user a chance to learn.

I can understand the position of people who insist that everything that qualifies as an answer should stay, no matter how bad or wrong it is. But I respectfully disagree: reviewers who choose to downvote obvious garbage are punished by losing reputation points, so if we insist that garbage answers should not be deleted, many of those answers will simply stay there with zero score forever.

  • "one-liner" meaning only that it is short? Or that it is short because it only has part of an answer? "You can ensure the rsync data is not corrupted in the network by using the rsync --checksum option" could be an one-line answer to a question, which is sufficient except for being wrong. – sourcejedi Jun 27 at 8:26
  • The effort put into answers is an interesting criterion and I feel it somehow has a role in shaping my opinion about keeping/deleting too. – fra-san Jun 27 at 9:54
  • 1
    @sourcejedi It's about effort, which indeed correlates with length. If in the example you provided rsync is made a hyperlink to the man page, and the command is properly formatted, I'd be less inclined to vote to delete. If the very same answer comes up with no formatting whatsoever and a couple of typos, it wouldn't qualify for nice treatment. – Dmitry Grigoryev Jun 29 at 18:45
  • My only adjustment here is that the LQP review privilege comes alongside the editing privilege, so if the only thing "wrong" with an Answer is some typos or missing formatting, I would encourage you to edit the post accordingly. Note that the queue's subtitle says "Identify, then improve or delete low-quality posts". – Jeff Schaller Jul 17 at 19:46
  • I would also encourage you to think about treating long posts and one-liners the same in regards to addressing the question. Why would you delete a short & wrong answer but not delete a long & wrong answer? I would recommend keeping both and leaving a comment on both that points out the shortcomings or misinterpretation. Give the author a chance to correct the post or to delete it. – Jeff Schaller Jul 17 at 19:47
  • 1
    Perhaps the part that we're missing here is the decision about when to delete a post, separate from the LQP queue. Answers that have been commented on and possibly downvoted from the queue may eventually be deleted. I think the point of the LQP is to be something like a triage gate where NAA's are weeded out. – Jeff Schaller Jul 17 at 19:51
3

apparently there is strong consensus that we should avoid deleting poor and wrong answers, as stated on Meta Stack Exchange

I want to address this point specifically. The "consensus" doesn't exist. The network, as it works (and you can ask users that use several sites) will delete those answers. That specific post had been heavily edited by Stack Overflow users. And only a small portion if them to boot.

I believe that it's more useful for our purposes to have our version of that post and establish explicitly that we don't fully agree with all the recommendations.

  • I think I see your point. Independently of the accuracy of "strong consensus", which is to some extent instrumental in my question, I agree that our aim should be to define this site's version of the guidelines. – fra-san Jul 9 at 8:57

You must log in to answer this question.

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