10

Conjecture Questions

In Lit.SE's activity so far, I've noticed a type of posts I think of as Conjecture Questions -- posts that define a thesis or a proposition, and their question to the site is, "can you confirm or disprove this?".

Some common forms as examples:

  • Is [Character] gay?
  • Was [Author] inspired by [Thing]?
  • Was [Book] the first to use [Trope/Technique]?

--but conjecture questions can be pretty much anything, as long as some thesis is being proposed. For example, I think of my "Could the perpetrator in [Book] be other than what the book clearly implies?" as a conjecture question, since it offers a thesis that runs counter to a straightforward reading of the book.

The Problem With Conjecture

Conjecture is the bread and butter of literary analysis -- finding interesting readings and interpretations and connections. Literary conjecture is clearly on-topic.

However, I feel there's a problem with conjecture questions, with conjecture Q&A, as I see them on the site so far. It's something like this:

  • Poor conjecture questions are extremely easy to ask. Any supposition, no matter how groundless or uninformed, can be phrased as a conjecture question. Additionally, conjecture questions can be extremely unclear is what it is that they are proposing -- for example, what is actually being requested by an "Is [Character] gay?" question?
  • Poor conjecture questions can be unclear on what they're looking for, and have prompted answers explaining what the conjecture even is, or could be, or should be. This makes sense if, wanting to give a substantial answer beyond "your conjecture is wrong," the answerer needs to seek some version of the conjecture that can at least be debated.
  • Poor conjecture questions make the site look amateurish. In bulk, conjecture questions that seem self-evidently groundless and uninformed, make the site look very bad - like a site for literary conspiracy theorists. Raising wild suppositions, in the hopes that someone will do something with them. Even if the poor ones get excellent "No, there is no support for this proposition," the site is still littered with posts going "Is this random idea correct?" "No.".
  • Poor conjecture questions are hard to tell apart from good ones. Literary analyses can turn black into white, and back again. While it is easy to say "In my opinion, your conjecture is entirely unsupported by the text," it is very difficult to say "--and nobody else can possibly draw a connection either." It's always conceivable you're missing something. This makes it difficult to vote to close or downvote a conjecture question -- what if it's a great question, what if it might get an excellent answer, and you just aren't seeing it?

    And that works both ways. People can ask about conjecture that makes a lot of sense, but without explaining why -- after all, supporting or disproving the conjecture is what the answer is supposed to do. But bystanders who don't know the answer, can mistake these as being just as arbitrary and baseless as the poor type of conjecture question.

Based on the posts I've seen so far, I think this is a concern we should address. Happily, we have an excellent example in a sister site, which also had to deal with questions going "I have a theory, is it true?" -- at Skeptics.SE.

Skeptic.SE and Notability

Skeptics.SE's site intro explains:

One area that is tricky for new users who ask questions, is that we aren't that interested in spending effort researching answers to ideas that no-one actually believes. We want to confirm or disprove real claims that many people think are true. Therefore, we ask that questions address notable claims.

"Notable claims," on Skeptics.SE, means demonstrating that a claim is believed by many, and/or believed by somebody notable. (Note: this is not the method I'm proposing for us. :) )

I found this explanation extremely clear, and ringing some familiar bells:

There are several reasons for the requirement for notability.

  1. We don't want to waste our time on claims invented by a couple of drunk guys at a bar. "My buddy reckons a tiger could beat a hippopotamus in a fight."

    For that reason, we are looking for "notable" claims. Ideally, notable means believed by a significant number of people, but we also commonly use a short-cut. If you can find someone famous saying it, we assume that lots of people have heard it and a significant number believe it.

  2. We get many questions by people who haven't understood the claim properly. They have misunderstood it - perhaps failed to see that it was satire. We like to be able to confirm the claim is being seriously made, and it is as described, so we don't spend effort writing arguments against straw men.

  3. If we get bogged down in debates about definitions, having some context that we can use to root out what the original claimant meant by the claim can be helpful.

You can read their Notability FAQ for more details and related discussions.

I find Skeptics.SE's issues with notability to be extremely similar to our issues with unsupported conjecture. The assumption that something can be meaningfully discussed, on the grounds that somebody has asked to discuss it, can drag us towards overanalyzing trivialities -- or, just a bunch of "actually, that's pretty uninteresting" answers.

Proposal: Explain support explicitly

Skeptics.SE uses notability, but I don't think that's what we want -- people coming up with new, original suggestions is fantastic for literary discussions.

What I do want to see is explicit support. If you are offering a conjecture to be evaluated by the community, I want to understand why this conjecture should be taken seriously.

Support can be many things. It can be references from the text, or things others have said about it, or details about the author's life. I don't think we can place any limits on what might constitute support. But there needs to be something.

There's one immediate objection: If OP can support their conjecture, what do they need to ask for?

And my answer, as much as I have one, is: OP doesn't need to support that their conjecture is correct. But they should support that it's worth considering; that there's a question here with actual substance.

And this gives us a critical benefit: once support is made explicit, it becomes much easier for the community to assess. If I can tell what the conjecture is based on, I can tell whether it's a solidly-grounded conjecture. If it's based on statements by others, I can get a better sense of what they said and what they meant. If I know what the OP is keying on to, I'm much more able to address what he's actually asking about.


So that's what I would propose: Questions of conjecture, that do not explicitly offer support for their conjecture, should be closed. I would go so far as to suggest this be a site-specific close reason - something along the lines of:

Questions requesting the proof or disproof of a claim, are required to demonstrate what support for the claim is already known, or context for why the question is being asked.

Note that I am not proposing we define or limit what "support" should consist of. Insufficient support would not be cause to close a question - complete absence of support would be, or not being able to tell what OP considers the support for his question. Once we know what support OP is relying on, we can upvote or downvote intelligently.

This gives us the following clear guideline:

  • No support mentioned, or is it insufficiently clear what is meant to be the support for the conjecture? Vote to close.
  • Support is clearly outlined, and you think it's lousy support? On topic. Downvote, comment, complain in chat about how Lit.SE is going to the dogs.

Pros and Cons

This proposal comes with an immediate price tag: it makes it harder to write questions, and it gives the community a new guideline we need to moderate and enforce. It also puts the question of "What do you mean by 'support'? What constitutes 'good' support?" front and center, as something we need to tackle anew with practically every question of this type. This can make our site less intuitive and welcoming to newcomers, as there will be additional, non-obvious rules for them to learn.

All that being said, I suspect we're going to be paying that price anyway. My impression so far is the Conjecture Questions with no support are already met with mixed reactions -- some enthusiastic, some disapproving -- and that subsequent improvements to the question body, universally, have the OP adding context and support. In other words, it's a rule of thumb that these questions are improved by adding context/support - we're going to be suggesting people add those anyway. We're going to be dealing with these questions and these issues.

Making the requirement explicit, establishing this as community policy, will give us a solid structure to offer OPs, and make the community's guidance clearer.

  • 3
    This sounds like a great idea. But to clarify: Does this boil down to explain what made you think this? If so, maybe consider adding a TL;DR? :P – user58 Feb 27 '17 at 19:07
  • I think this gets at a larger problem: people just aren't doing their research before they ask questions. I personally tend to downvote questions when it's clear that the person asking it hasn't done even cursory research about the topic. I'm not convinced that we need guidelines specifically for conjecture questions, since they're part of a larger problem. I'm also not convinced that moderators should be closing these questions, again since they're part of a larger problem. – user111 Feb 27 '17 at 19:21
  • @Mithrandir : I'm a little wary a TL;DR will seem dismissive or arbitrary without the complete discussion. I've highlighted the core proposal to the best of my ability, and I think the title is fairly clear :) – Standback Feb 27 '17 at 19:24
  • @Hamlet : My issue with conjecture questions is (A) it's not clear what research is "expected" for a conjecture question, and (B) it's hard to tell whether a conjecture question is researched or not. That lack of clarity is also what makes downvotes the wrong tool here - a "what does he want from me; what is she asking for here" question should be closed, not (only) downvoted. – Standback Feb 27 '17 at 19:31
  • 4
    Essentially you're asking for something like Sceptics' policy on false claims. I think it's a good idea, given recent questions over at SFF along the lines of "I heard Draco raped Hermione, is dat true?" – Gallifreyan Feb 27 '17 at 19:33
  • 2
    @Standback I honestly would not be opposed to having a close reason along the lines of "no prior research" or "based on faulty assumptions". It's just that, as of right now, I'm not sure it makes sense to have a close reason specifically for conjecture questions when the problem exists for other questions. – user111 Feb 27 '17 at 21:46
  • 3
    I dislike Skeptics and don't use that site, and at least part of this aversion has to do with their notability policy for questions. -1 for now; I'll try to write up an answer at some point. – Rand al'Thor Feb 27 '17 at 23:15
  • 3
    @Gallifreyan Those questions are a special case; there's strong evidence suggesting they were all being asked by the same one or two people. You wouldn't call for closing all questions about time-travelling robots just because of Rondo, would you? Also (and importantly), those questions were downvoted into the cellar anyway. IMO, there isn't much of a need to establish a close reason for a particularly bad yet rare type of question if community votes are already doing a good job of marking them as bad posts. – Rand al'Thor Feb 27 '17 at 23:19
  • I'm with Rand on this, at least for now. In the meantime, I would encourage more people to downvote bad/poorly thought out content. What surprises me about this question is that there are a lot of poorly researched questions on this site that make incorrect assumptions, but they all seem to be upvoted. If people feel strongly about these questions, express yourself with your votes. – user111 Feb 27 '17 at 23:35
  • @Hamlet : I suspect a lot of the community doesn't really know how to spot the incorrect assumptions. And there's always the gap between enthusiasts who like a question and upvote it, and site-hygiene enthusiasts who dislike some question patterns. It's particularly a problem in a new site, where there aren't clear norms yet -- so upvoting happens a lot, and potential downvoters go "well I'm not sure this is a bad question; there's no site policy on this; it might get a great answer..." – Standback Feb 28 '17 at 5:00
  • 2
    I still think this question would benefit from more actual examples (not necessarily from questions which have already been asked on the site), just to give us an idea of what you'd consider a supported question and what not. – Rand al'Thor Feb 28 '17 at 14:08
  • Related: literature.meta.stackexchange.com/q/758/481 – Shokhet Jun 11 '17 at 18:09
1

What problem would this be solving?

  • Are we getting inundated with bad "unsupported conjecture" questions? Until they become a significant problem, it's not really worth creating a special close reason for them.
  • If so, is the community failing to handle them? Are we getting a lot of these questions which are bad for the site (not just 'bad questions', which is a matter of opinion, but for some reason shouldn't be allowed to survive here) and yet are being upvoted? If a class of questions is already getting downvoted into the basement as soon as they're asked, it doesn't seem worth setting up a close reason for them too; just let them rot.

What exactly are these questions anyway?

  • You talk a lot about questions which "don't offer support", but what do you imagine "offering support" as entailing? Can you provide some example questions which do and don't offer adequate support, in your opinion? I assume your meta post was inspired by this question, which would meet Skeptics's notability requirement as it cites an article claiming Edmund is gay.
  • Is the criterion usefully definable? Can we define what "unsupported" means in a sufficiently objective way to be useful? With a close reason like this, I envisage people VTCing every conjecture they disagree with, on the grounds that "well, that evidence doesn't convince me - it can't be properly supported". Although some sites have made subjective close reasons work, I'm not a fan of them - I prefer "this question should be closed" to be, in the majority of cases, a very clearly provable or disprovable statement, with little room for argument.
  • 1
    I think you're misunderstanding my proposal; let me clarify. I'm proposing an objective criteria -- is support offered, not is it good. Bad support = downvote; no support = close. My issue isn't with Is Edmund gay?, which provided its support from the start. People doubted that support, but were able to assess it. Whereas compare the original vs. the final version of the Frog and Toad question -- originally unanswerable, final question is clear and great. – Standback Feb 28 '17 at 4:29
  • 1
    As for "what problem would this be solving," I devoted an entire section to that. Do you feel that's unclear, are you saying you don't see those as problems, or are you saying it could be a problem but presently aren't bothered by it? (For myself, well, I think it's already a problem. Partially in numbers, but partially in reception -- whether unanswerable/unclear questions are being closed/downvoted; whether we feel community norms are clear enough for a close vote to be appropriate. That's why I think we need a line - this merits a close vote, this doesn't.) – Standback Feb 28 '17 at 4:37
  • I see now that my phrasing "We'll judge each question by the way it's supported" may have been unclear. I've edited to clarify, and I think this might address some of your criticisms. – Standback Feb 28 '17 at 6:39
  • 1
    @Standback Unfortunately, there's no easy way to draw the line between support being offered and support being good. Would you take "I saw this somewhere on the internet" or "I read the book and just had the feeling this might be true" as support? (I mean exactly those sentences, not pointing to a particular place on the internet.) If so, then people could just write down any old nonsense as 'support' for anything, making the whole thing pointless. But if not, we'd be getting into disputes about what constitutes support and whose arguments are better. – Rand al'Thor Feb 28 '17 at 13:36
  • @Standback My "what problem would this be solving" section is more intended as a rhetorical question. Obviously you feel it's a problem, otherwise you wouldn't have started this meta post! But if the answer to all my subquestions is no, then I don't think it's a big enough problem to need a close reason. (Btw, I don't think the Frog/Toad question was originally unanswerable. In its final state it's easier for someone who knows nothing about the work to appreciate the question, but it's not any more answerable for an expert.) – Rand al'Thor Feb 28 '17 at 13:39
  • 2
    I think this answer could benefit from a more thorough reading of the question. While it's true that the question is somewhat lacking in the area of specific examples, it does identify common patterns between them, and this answer seems to ignore those patterns entirely. Also, the last two reasons (while important) aren't justifications for declining the proposal; they're justifications to think about exactly what it means in practice more thoroughly. – Aza Feb 28 '17 at 21:35
  • @Emrakul Did I say they were justifications for declining the proposal? The first two bullet points are rhetorical questions to which the answer is (IMO) "no, and therefore we don't need a close reason"; the last two are more pointing up a lack of clarity with the proposal. Admittedly, they might thus have been better suited as comments than an answer (especially the first one), but the whole question/answer system is looser on meta anyway. They're at least useful to consider, and necessary to answer if we're going to make a policy out of this. – Rand al'Thor Feb 28 '17 at 23:11
  • We had a somewhat related discussion on Stack Overflow Meta recently (in particular, if it's worth closing questions that are now off-topic retroactively). The consensus there seems to be that leaving this kind of question open can have a "broken windows effect" if people take it as evidence that they can ask that kind of question too. It's also possible for this kind of question to attract low-quality answers from new users. – EJoshuaS Mar 7 '17 at 20:29
  • @EJoshuaS Right, closing should be done retroactively if site scope changes. I don't see what that has to do with the issue being discussed here though ... – Rand al'Thor Mar 7 '17 at 21:07
  • @Randal'Thor Probably poorly phrased on my part - my point wasn't so much that we should retroactively close bad questions as much as that even if we aren't getting a lot of bad questions of this sort yet, that doesn't mean that we shouldn't deal with the issue now - if we allow them now then people in the future might assume that it's OK to do that. – EJoshuaS Mar 7 '17 at 21:14
  • @Randal'Thor I was (rather poorly, in retrospect) trying to make more of a "broken-windows" argument - even if there aren't many of these questions yet, we should still have an official policy on them; broken-windows theory argues that you shouldn't wait until there are lots of broken windows before you take action, you should start taking action as soon as there are any broken windows. – EJoshuaS Mar 7 '17 at 21:30
0

I agree that this kind of question should be closed, but it's possible that we don't need a custom close reason for it. IMHO, we could just close those questions as "unclear what you're asking" or "primarily opinion-based" (which they are if it's not possible to write an answer you can support with some kind of evidence).

I think that the following heuristic would be valuable here: is it possible to write an answer that can be supported with specific evidence? If not, it should be closed as primarily opinion-based. (Admittedly, though, some ridiculous theories could be refuted with specific facts, so it's not a perfect heuristic).

We had a somewhat related discussion on Stack Overflow Meta recently (in particular, if it's worth closing questions that are now off-topic retroactively). The consensus there seems to be that leaving this kind of question open can have a "broken windows effect" if people take it as evidence that they can ask that kind of question too (or it makes it look like the community doesn't care about low-quality questions). It's also possible for this kind of question to attract low-quality answers from new users.

The application here is that if we allow poor questions like that now (even if it's just a few for now) future visitors might assume that it's OK to ask questions like that in the future. As an analogy to "broken-windows theory," the best time to act is as soon as the first window gets broken - you don't want to wait until people break lots of windows before you take action.

  • 1
    I think you're misunderstanding the issue here. Nobody is suggesting closing all questions which can't be answered directly from the text - there's often surrounding material which can be equally useful. The proposal is to close questions which ask whether X is true without providing any evidence for why X may or may not be true (regardless of whether there is or isn't evidence for X in the text itself). Also, I don't understand what your last paragraph about retroactive closure has to do with the issue here. – Rand al'Thor Mar 7 '17 at 21:15
  • @Randal'Thor You're correct, I didn't intend to imply that evidence from the text itself is the only valid form of evidence to support a conclusion. I edited that point to read "is it possible to write an answer that can be supported with specific evidence?" The point I was making was that it should be possible to write a fact-based (rather than purely opinion-based) answer. I edited to clarify my last point too. – EJoshuaS Mar 7 '17 at 21:27

You must log in to answer this question.

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