2

Recently, a moderator deleted two low quality answers. The explanation given does a good job of explaining why:

We're looking for long answers that provide some explanation and context. Don't just give a one-line answer; explain why your answer is right, ideally with citations. Answers that don't include explanations may be removed.

This site has a lot of these types of answers cluttering up the site: answers that don't really provide an explanation for why their claims and conclusions are correct. These answers don't really have much, if any redeeming value. Despite this, these answers sometimes attract upvotes, particularly answers posted during the private beta.

Similar to how we have a process for closing questions, would it be useful to have a process for deleting unexplained answers. One potential process could be that answers flagged using a custom flag would be reviewed by moderators, who, if they agree, would apply a post notice to answers, and delete after a specified time period. Disagreements about when to apply this process could, like disagreements over close votes, be taken to meta.

Here are the two deleted answers if you don't have the points to see deleted answers:

I think it means, next time he will be given a long jail sentence, and the judge won't be foolish enough to give him a short sentence or exhibit any clemency.

and

I think (correct me if I'm wrong, I haven't read it yet) he means that next time he won't hesitate next time or foolishly give mercy.

Here's another example of an answer that a moderator put a post notice on but didn't delete:

It means a man must give up all of his self-importance and pride in order to be a servant of God, that as long as he (emotionally) hangs on to what he was before finding God, God cannot (or will not) transform him into what he should be with God. It is a demand for surrendering yourself to God's will completely without reservations as to what God may do with you.

  • Could you link to some examples of answers which would be deleted under the proposed policy? It might help people to get a clearer idea of what's being proposed here. – Rand al'Thor Jan 25 '18 at 15:51
  • @Randal'Thor I actually already link to two examples of answers: the answers that were deleted. – user111 Jan 28 '18 at 19:35
  • Oh, that's true. But is this the only type of answer that you're advocating deleting? It seems to me that "unsupported answers" covers a lot more than just VLQ one-liners, and I'm trying to get an idea of how far you're proposing to go. – Rand al'Thor Jan 28 '18 at 19:47
  • @Randal'Thor the post notice relevant to this post ("We're looking for long answers that provide some explanation and context...") is about the same length as the close reasons. People are able to identify which questions should be closed, despite the fact that close reasons are not several pages long with detailed instructions for every eventuality. This post notice/deletion process would work very similarly: people would flag things to be deleted, moderators would use their judgement, and disagreements could be hammered out on meta. – user111 Jan 29 '18 at 0:57
  • @Randal'Thor however, I find it strange that you're advancing the argument that this post notice is so vague that it could mean anything. "explain why your answer is right" seems like a perfectly clear criteria for removing answers to me. I suppose you could get more technical with something "the majority of the claims that directly answer the question must have a justification, and that justification should be at the very least a reasonable attempt at justifying the claims (e.g. not jibberish)" but that makes things more confusing, not less. – user111 Jan 29 '18 at 1:03
  • 2
    Matt Thrower's answer here is a decent demonstration of why your proposed policy needs to be clarified. One of the site's top users seemed to interpret this as a call for deleting answers without citations, including those backed up only by close reading. – Rand al'Thor Jan 29 '18 at 18:05
  • 1
    @Randal'Thor That was not quite my interpretation. Sorry, it seems I have been very unclear in my answer. The point I'm trying to make it simply that on this SE it's possible to post a good answer without support. Any policy on deletion has to either take that into account. – Matt Thrower Jan 29 '18 at 18:57
  • @Randal'Thor I would like to do a better job explaining this but I'm not sure how I can do that. "Explain why your answer is correct" seems perfectly clear to me. Given that you understood the proposal well enough to write an answer arguing that it's a bad idea, I'm not entirely sure why I'm getting these questions now. I would appreciate specific questions. – user111 Jan 30 '18 at 1:47
  • You know what, since you haven't given me any specific questions, I'll make up some questions and respond to them. Q: What if an answer makes two claims but only offers an explanation for one of them? A: a good litmus test to use here is if you edit out the unexplained portions of the answer, is there anything worth keeping? – user111 Jan 30 '18 at 1:53
  • @Hamlet I understand the proposal well enough to write an answer - I've been one of the most active users here since the beginning, so I've seen which answers you comment on and what you talk about in chat. What I'm trying to do is ensure that everyone is on the same page here. – Rand al'Thor Jan 31 '18 at 10:38
4

Firstly, what do we mean by "low quality answers"? Obviously this needs to be clarified, because "low quality" without context can be a highly subjective evaluation. In the context of this meta post, you seem to mean unexplained answers. I'll call them that rather than "low quality" answers below, just for clarity and to make sure we're all on the same page.


Standard guidelines

The help centre has some guidance on when answers should be delete-voted:

When should I vote to delete an answer?

You may vote to delete answers in the following cases:

  • The answer is extremely low quality: There is little to no scope for improvement
  • The answer doesn't attempt to answer the question; it may be a comment or a separate question altogether.

The type of answers you're talking about here certainly do attempt to answer the question, just without explanation. And there's certainly scope for improvement: any unexplained answer can be turned into an explained one simply by, well, editing in an explanation. So according to the standard guidelines for answer deletion, unexplained answers don't need to be deleted.

However, that help centre page is the same on every site in the SE network, so it doesn't take account of any special problems we may have here due to subject material or site history. I know other sites have specific rules on deleting certain types of answer which aren't the same everywhere. As noted in this main meta FAQ:

What are the criteria for deletion?

[...] For answers, any post that is not an answer (should be a comment, doesn't answer the question, etc.) should be deleted. Answers that are wrong or that dispense poor advice should be downvoted, not deleted.

These are general guidelines; some communities in the network may uphold more specific reasons to delete posts or not. For example, on Puzzling.SE, answers to a puzzle without explanation are subject to deletion and some technical sites will delete answers which are not only wrong but also harmful when tried.

So ...

Is this enough of an issue that we need new guidelines?

  • Are we having a problem with too many unexplained answers?

    Well, for one thing, we're not actually getting many answers at all on this site! Our question count and answer count are roughly equal, and have been so for a long time; on every other SE site I know about, there are way more answers than questions.

    Of the answers we are getting, most of them are explained (to varying levels of detail), but there are also occasional one-liners and unsupported speculation. I don't have any exact statistics, but it must be well less than half or even a third of our total answer content. Evidently your campaign of downvoting and commenting on such answers has been somewhat effective.

  • To the extent that there is a problem, are votes failing to deal with it?

    The usual solution to an answer which is bad but not NaA is to downvote it, reserving deletion for extreme cases such as gibberish or non-answers. A negative score sends the desired message "this isn't good content", while allowing the post to be improved so that votes can be reversed. This only fails when people aren't downvoting enough: if harmful content is getting upvoted, that's when it's time to implement more drastic solutions such as new deletion policies.

    When someone posts a new answer without explaining it, it tends to attract downvotes pretty quickly. I can't think of any examples of recent unexplained answers which have been particularly well-received or highly-voted. The current community votes responsibly, for the most part.

    That still leaves the old answers from around the private beta period, some of which are unsupported and yet higher-voted than they would be if posted today. However, when these answers are bumped to the front page by edits, they do tend to accrue the downvotes that they should have had originally.

So the main problem here is unexplained answers from private beta which were upvoted then and haven't had enough downvotes since. This is a finite set, which (in comparison to the site's total volume) is only going to shrink over time. I wouldn't necessarily be averse to mods slapping post notices on some of those old answers, but the problem isn't big enough to merit outright deletion of swathes of our content. Especially since the policy you propose would also mean deleting a lot of new answers as they come in, which we're already dealing with satisfactorily using votes.

Just be liberal with downvotes and comments, and hope that people edit to improve. That does happen - I've seen it - and it wouldn't have if we'd deleted those answers instead of commenting to request improvement. And if these answers aren't improved, and remain downvoted instead, then it sends a visible signal that unexplained answers aren't well-received here, which deletion wouldn't.

  • Deletion doesnt need to be immediate. We could put a post notice on these answers and delete after, say, one month if there hasn't been any improvement. – user111 Jan 16 '18 at 14:13
  • @Hamlet Hmm, I could maybe get behind that. But would you be proposing this for all unexplained answers or only the old upvoted ones? Do you agree that downvotes are doing a reasonable job of dealing with new ones that come in? See, part of me is worried that by killing off so much content we'd sort of sterilise the site. I do recognise that no answer at all is often preferable to a bad answer, but OTOH nothing is perfect, and trying to make it so can be more harmful than the imperfections. (There's a literary quote which is just perfect for this, but I'm not remembering it ...) – Rand al'Thor Jan 16 '18 at 14:21
  • "part of me is worried that by killing off so much content we'd sort of sterilise the site" what on earth does that even mean? There would still be plenty of incorrect or otherwise bad answers. All anyone is proposing is removing unsupported answers. – user111 Jan 16 '18 at 14:37
  • 1
    This answer's approach fundamentally misunderstands the purpose of the Stack Exchange deletion process. This isn't reddit. Not all content gets to stay by default. Content that isn't suitable for being an adequate answer to be voted on is removed, not left to downvotes and discussion. This is a major part of what it means to be a curated resource. The only way to arrive at the conclusions in this answer is to presuppose that deletion and voting are attempting to accomplish the same thing. Really, the post notice I put on that question is the barest minimum that can reasonably be asked. – Aza Jan 16 '18 at 15:59
  • 1
    I don't normally post multiple comments, but... re: sterilization: infrequency of good answers doesn't mean we allow low quality answers to make up the difference. That's building verisimilitude of activity rather than the kind of activity the site wants and needs. The response to low activity shouldn't be to permit anything in the hopes of bolstering activity - or fearing sterilization if we try - it should be to develop our otherwise nonexistent site guidelines, and to try to become something that can be meaningfully contributed to. – Aza Jan 16 '18 at 16:01
  • 1
    @Zyerah I've literally linked directly to the canonical description of the Stack Exchange deletion process. Deletion is for NaAs and answers with "little to no scope for improvement", and for other categories of answers if there's a site-specific consensus that such should be deleted. When you say "[c]ontent that isn't suitable for being an adequate answer", who decides what "adequate" means for posts that do answer the question and do have scope for improvement? – Rand al'Thor Jan 17 '18 at 12:47
  • 1
    I do get what you're saying re sterilisation, and perhaps I'm stupid to even worry about this, but I can see both sides of the argument here. Like I said above, "I do recognise that no answer at all is often preferable to a bad answer, but OTOH nothing is perfect, and trying to make it so can be more harmful than the imperfections." We need quality standards, yes, but at the other (hypothetical) extreme, we shouldn't delete everything that falls below some imagined perfect ideal. – Rand al'Thor Jan 17 '18 at 12:50
  • 1
    @Randal'Thor "but at the other (hypothetical) extreme, we shouldn't delete everything that falls below some imagined perfect ideal" that is just... asking posts to provide some sort of justification is about the lowest bar possible. Providing a justification does not in any way guarantee that an answer is correct or good. (That's what votes should be for imo). – user111 Jan 17 '18 at 16:57
  • 1
    @Rand "we shouldn't delete everything that falls below some imagined perfect ideal" - no one is doing this, or even suggesting this. For the rest, I've commented here. – Aza Jan 17 '18 at 19:21
  • Votes are for answering the subjective question of which answers is the most helpful and should rise to the top. Votes are not for enforcing rules that are the only thing allowing a site about literature to work properly. We don't downvote questions as a substitute for closing them, and we don't leave "be nice" up to votes. The same thing holds for a rule as important as "back it up". – user111 Jan 31 '18 at 3:37
4

As Rand al'Thor opened his answer, the real question here is "what is a bad answer"? Of course we should be deleting bad answers: Zyerah makes an excellent point that "this isn't reddit". So: what do we delete?

This is where we run into trouble compared with other SE sites. For computing or maths based sites there is often a definitive "correct" answer which can be proved by a chain of logic. For science, it is often straightforward to provide a reference, even if it's just a page in a textbook. Same goes for humanities and social science, except here there's more debate over what's actually "correct". However, while that makes the Q&A format messy, it's not hard to provide an actual reference.

There are some answers though, especially in History SE, that don't. Yet I don't feel they're bad answers. Why? Because they make statements which are relatively self-evident.

Here, we face a double issue. Not only can we provide a lot of "self evident" answers by close reading and providing examples, but it's much harder to provide an actual reference. Most people don't have books of literary criticism littering their homes. Even in cases where you can find relevant, essays, papers and thesis on the web, it's often hard to provide a definitive reference, because Literature is subjective, and opinions differ.

A couple of simple examples from my own answers. There two:

Provide no references, because they are based on my own close reading of the text. Whereas these two:

Provide a single reference, but contain additional opinions which I've either arrived at myself of gleaned from forgotten, unreferenced sources?

Does that make those answers of no value? I would argue that it does not. In fact, looking at a number of my answers while I tried to find examples, I see that most of them are not referenced and, some, indeed are not based on close reading. Here, for example, is an accepted answer that probably got that way because it's pretty obvious to anyone who stops and thinks for a while about the question and the content of the text: What was the irony of the end of Lord of The Flies?

The subjective nature of literature and the objective demands of the site have lead us into some nasty blind alleys. I once composed an answer based on an amount of commonly-accepted historical practices, which got a negative reception because it was "not referenced". Yet if you're asking - to make a frivolous example - a question where the answer depends partly on the Catholicism of the Pope, you can't really expect someone to find a reference stating that the Pope is Catholic.

This little rant isn't really much of a concrete answer, because I don't know what the answer is to reconciling the message and the medium here. All I know is that I enjoy participating on the site, and I would like to continue doing so.

In summary:
- We need a better definition of what constitutes "a good/bad answer"
- That has to go beyond referencing/supporting the claims in the answer

  • It's important to note that the type of "support" being talked about here isn't just limited to citations and references. Things like close reading can be perfectly valid "support" for an answer - I don't think anyone will seriously disagree with that. As I understand it, Hamlet's proposal is not to delete answers without citations to books or papers, but to delete those with no explanation at all (e.g. "I think X means Y", but not "X could mean Y because Z"). – Rand al'Thor Jan 29 '18 at 18:04
  • 4
    @Randal'Thor I understand that: but as I tried to explain in my answer, there are still several examples of good answers that provide neither references nor support, because they're based on self-evident principles. – Matt Thrower Jan 29 '18 at 18:33
  • There really isn't a such thing as an obvious answer or obvious conclusion. If someone is asking a question about it the answer can't be obvious to them. – user111 Jan 29 '18 at 21:51
  • If people want to build a "common knowledge" exception into deleting answers and that's the only way we'll get people to agree, then I suppose I could come around. But Stack Exchange's track record with common knowledge has not been good. For example, it was common knowledge on this site (and still is common knowledge on Science Fiction) that Tolkien hated allegory and therefore any allegorical reading of LOTR is incorrect. – user111 Jan 29 '18 at 22:21
  • @hamlet quite: I do agree it's not a great solution and prone to abuse. But I fear the opposite may be worse in terms of excluding useful, valid content. Some feel we already have an issue with seeming an approachable SE & a hard line on answers might make it worse. In terms of "obvious" questions you are, of course, right. But that's not the issue: it's whether answers to such "obvious" questions can be given useful references or other supporting information. I suspect in many cases that could be difficult. – Matt Thrower Jan 29 '18 at 22:39
  • "whether answers to such "obvious" questions can be given useful references or other supporting information" My experience is that the more obvious something is, the easier it is to find a reference supporting it. (This makes sense). To use your example of the pope being catholic: a biography of the pope would explain when/where the pope "became" catholic and so on. (Having more detailed information like that would make the answer better). – user111 Jan 29 '18 at 22:55
  • RE: unfriendliness or turning people away: Rand al'Thor had the same line of thought and Zyera had an excellent response: "The response to low activity shouldn't be to permit anything in the hopes of bolstering activity - or fearing sterilization if we try - it should be to develop our otherwise nonexistent site guidelines, and to try to become something that can be meaningfully contributed to." – user111 Jan 29 '18 at 22:58
  • @Hamlet predictably, I disagree on the Pope thing. Even textbooks tend to assume a degree of "common knowledge", otherwise every sentence would need a reference and the references section would be longer than the text itself. As regards turning people away, yes, a great point: a hard balancing act, but an important one. – Matt Thrower Jan 30 '18 at 9:26
  • If you have an example of something that is so obvious there is no source for it I would love to hear it. My experience is that there hasn't been a claim that's so "obvious" that there isn't a source for it. If you want to put in an common knowledge exception then fine I guess; I'm not looking forward to seeing what nonsense gets excepted as obvious knowledge but that's just me. – user111 Jan 30 '18 at 13:56
  • @Hamlet I gave one: my answer to the question on Lord of the Flies. – Matt Thrower Jan 30 '18 at 14:21
  • @MattThrower I'm not sure why that answer would violate any rule about requiring answers to have an explanation: that answer cites and explains specific moments in a book to make its point. – user111 Jan 30 '18 at 21:40
  • @Hamlet Ah, well, in that case, Randal'Thor was right and I don't entirely understand your point. I guess it illustrates the issue at the heart of this, which is that it difficult to draw a clear red line between "bad" and other content. – Matt Thrower Jan 31 '18 at 9:01
2

Opinion, by itself, is noise.

I think it's worth remembering that Literature is a subjective topic, and therefore by extension the Literature Stack Exchange site is a subjective site. There are a lot of questions on this site, such as Do Guildenstern and Rosencrantz deserve to die?, where the correct answer is often a matter of opinion. For such questions, the only way to make them work in Stack Exchange's Q&A structure is if answers back their opinions up with arguments. If these questions are seen as a place for anyone to post their unsupported opinions, then the question just turns into a poll where people vote for the answers they personally agree with. Polls do not work on Stack Exchange. But if people write answers where they explain why they believe in their opinion, then answers can be evaluated on the basis of who has the most compelling explanation, and a genuine sharing of perspectives can occur.

Our experience in the private beta shows that in practice, our site does not work unless answers provide some sort of justification. Without that rule, questions such as "what is the meaning of the pickle jar in Ethan Frome" become a free for all where anyone can post anything. Our experience also shows that requiring justification actually results in a diverse range of responses and answers: we're requiring justification, but there's freedom about what justification actually means. As BESW writes:

The Stack's basic "Back it up!" principle should, in theory, be sufficient for lit.se as well.

Unfortunately, in my experience a significant number of Stack users are members of that part of the wider population which has not been trained in the difference between assertion and support.

This means most attempts to ask that support be added to an answer are interpreted as a challenge to the user's expertise, and that creates a defensive attitude which opposes the support-based culture underlying the best Stack content.

I've seen it on multiple Stacks, where opinion-based subjects are considered either free-for-all open game for unsupported speculation, or are considered impossible to answer within the Stack structure.

Literature can't succumb to that false dichotomy or it will crash and burn where some other Stacks can still survive with limited function.

But we don't actually need to invent extra policies for answering questions here, at least not yet. The simple notion that claims need explicit support--ANY kind of support--should be enough to get us off the ground, with multiple answers providing multiple critical lenses.

This attitude (that claims need support, but not specifying the kind of support) creates a space for responses ranging from reader response (experience-based support is a major part of several Stacks already! see rpg.se, parenting.se, gardening.se, etc) to authorial intent.

The point I'm trying to make here is that the principle that answers need to be backed up is something so important that it shouldn't be left up to votes. It's a fundamental requirement for this site to work, and answers that don't back claims up stick around, clutter up the site, and because people post content based on what has already been posted, encourage more poor answers as well.

I'm flexible about how this should be implemented. I would be OK with putting a post notice on answers that don't provide support, leaving them up for a month to give the writer a chance to update the answer, and then deleting the answer.

But we really shouldn't be leaving these answers up: they aren't just noise, they're antithetical to everything that allows a Stack Exchange about literature to exist. Votes should be for when an answer is incorrect, or bad. But an answer not providing justification isn't a bad answer, rather, it's a failure to participate in the norms that make this site possible. These answers should be removed.

  • 5
    I flat-out agree with this. In other fields, an answer without support is still marginally useful. On our site, the explanation is the answer, and without one, you have no answer. No answer on our site can actually answer the question, unless it's a purely factual question, without touching on why it is that way. How it's handled is up to the community, but we can't pretend these even come close to answering what was asked. – Aza Jan 17 '18 at 19:28
  • 2
    Would your proposed deletion policy apply only to answers on lit-analysis questions? You mention "There are a lot of questions [...] where the correct answer is often a matter of opinion", but what about the "purely factual question[s]" that @Zyerah mentions? We mustn't make the mistake of assuming Literature = literary analysis. E.g. if someone answers "what does [word] mean in [passage]?" with a definition of the word, that might be enough to answer the question to the OP's and every other reader's satisfaction, even if they don't cite a dictionary. Would you advocate deleting such answers? – Rand al'Thor Jan 18 '18 at 2:16
  • 3
    @Randal'Thor in my mind, backing up should apply to every question on this site: backing up claims is important for other reasons as well. With regards to the dictionary example: if someone is asking a question about it, then it's not obvious to them, so you should quote the dictionary anyway. However, with that said, I am very flexible about this: I would be fine with any sort of rule as long as there is a rule. Perhaps the details could be hammered out in a different meta post. – user111 Jan 18 '18 at 13:36
  • Since @Zyerah and Mithrandir would be the ones putting this into practice, perhaps they could elaborate on your question. – user111 Jan 18 '18 at 13:42
  • 2
    @Rand No one was talking about policy for purely factual questions until you brought it up just now. No one here has made that mistake. Please stop extending statements and suggestions past what they were meant to mean. We can also have a separate discussion on those answers - but that's not what this answer was intended to address. – Aza Jan 18 '18 at 19:04
  • 2
    @Zyerah This meta question doesn't specify that it's about lit-analysis posts only. Without an explicit statement of scope, surely the default interpretation of "let's delete unsupported answers" is "let's delete all unsupported answers", not "let's delete unsupported answers to a specific class of questions". Care needs to be taken over the wording of policies, especially those which will result in mass-deletion of content - hence why I asked for clarification. I wasn't accusing anyone of anything, so please don't put words into my mouth. – Rand al'Thor Jan 18 '18 at 19:54
  • @Randal'Thor I agree that I was perhaps too broad in this question. My intent was to propose a broad idea, with the hope of narrowing it down further. Perhaps a separate meta question would be in order where we can hammer out the specifics. – user111 Jan 18 '18 at 20:04
  • @Randal'Thor It's also worth noting that many purely factual questions on this site are subjective (mostly because we often lack the information needed to answer the question objectively). A good example might be any question regarding the identity of Shakespeare. So I'm not convinced that a subjective/objective divide exists on this site. – user111 Jan 21 '18 at 14:22
  • @Zyerah this answer doesn't seem to be working. Any suggestions? – user111 Jan 21 '18 at 14:25
  • 2
    I believe that this is attracting some negative attention because the original question is not phrased as a question asking for discussion. If you open it up as a "Should we start deleting unsupported answers on sight?" discussion instead, you may have better luck. – Mithical Jan 22 '18 at 12:27

You must log in to answer this question.