16
$\begingroup$

I've seen a spate of questions of a form more or less along the lines of "Is this working correct?" just lately.

While this is good from the point of view that the poster is clearly making a proper attempt, the difficulty is that if the answer is "yes", there's often nothing else to say, which means (unless you're creative enough to find something worthwhile to add) to post an answer (basically amounting to the comment "yes" as an answer) is to break SE's rules about very short answers.

Further, it's a Q&A that seems less likely to benefit other readers than a question with a "real" answer.

I have been trying to encourage OPs to rephrase into a more answerable question, but I am not sure what a good approach to this really is:

  • do we tolerate "yes" answers to these questions? I have leaned this way in the past but because of what seems to be a narrower utility of these questions I wonder if that's a good idea.

  • do we try harder to push OPs to ask questions that have longer-than-one-line answers? (This might be good if there's a concrete suggestion we can offer them. While I can sometimes suggest something, I'd love to have a more generally applicable approach to changing the question that could be offered.)

  • do we come up with a different line altogether, anything from closing them to putting up with having a large set of questions that can't be properly answered (but hopefully avoiding either of those)?

I really don't know a good way to go with these so I don't have a preferred option of my own. Any suggestions or discussion would be welcome

A couple of recent examples that (at the time they were posted at least) were more-or-less of the type that concern me.

Changing Sigma Limits for Proof

https://stats.stackexchange.com/questions/171054/is-my-work-acceptable-fitting-a-normal-distribution

Here's an older question in somewhat of the same vein:

Which distribution do I get?


Many more questions at least have the potential to encourage such answers, but (because the answer wasn't 'yes') avoid the problem, but they, too, highlight the potential problem, since we can imagine a slightly different attempt at the question which would have the 'yes' answer, such as this:

Implication of convergence in rth mean

which asks "My question is whether the following steps are sufficient or not"

to which sort of question "yes" is certainly a potential answer.

I know I've also answered quite a few (maybe half a dozen, maybe more) where I was able to say "yes, but"... but on several of those it took quite a good bit of thinking to come up with a "but". I don't think that's always possible.

$\endgroup$
13
  • $\begingroup$ Can you include a link to one such question please? $\endgroup$
    – Dawny33
    Sep 7, 2015 at 5:20
  • $\begingroup$ @Dawny33 There's 3 there now. $\endgroup$
    – Glen_b
    Sep 7, 2015 at 5:54
  • $\begingroup$ Or maybe four (.... ish) $\endgroup$
    – Glen_b
    Sep 7, 2015 at 6:08
  • 7
    $\begingroup$ Encourage them to point out where they've got doubts about their reasoning & focus on that in the question? After all, there must be something they don't quite understand - unless they're worried about a careless mistake, which arguably isn't something worthwhile for CV. $\endgroup$ Sep 7, 2015 at 8:28
  • 2
    $\begingroup$ On Computer Science, we tend to close such questions as "Unclear what you're asking." and leave a comment along the lines of this, which links to a couple of CS Meta discussions about this very topic 1 2. $\endgroup$ Sep 18, 2015 at 9:48
  • 1
    $\begingroup$ I think this is a very clear example. $\endgroup$ Sep 23, 2015 at 12:06
  • $\begingroup$ @Scortchi yes, thanks, that's just the sort of problem. After XI'an's one line comment, is there much more one can say? $\endgroup$
    – Glen_b
    Sep 23, 2015 at 23:17
  • 1
    $\begingroup$ A good answer, or good answers, might (1) make explicit the assumptions behind the calculations, (2) explain the connection to the distribution of minima & maxima, (3) illustrate the use of series/parallel diagrams (like circuit diagrams for resistors) as an aid to solving this kind of problem, (4) give useful references, (5) suggest extensions of the approach e.g. to stand-by components, (6) show how to simulate failure times as a check. $\endgroup$ Sep 25, 2015 at 9:37
  • $\begingroup$ AndreSilva pointed out in a thread on another question that this meta.SE question is relevant: Question closed because yes/no answer $\endgroup$
    – Glen_b
    Nov 25, 2015 at 15:50
  • $\begingroup$ @Glen_b I know this is an old thread, but I had a thought. Could one option be to answer the "Yes" instances of these questions in the comments, but then encourage the poster to split their question into a Q & A where they answer their own question? (Assuming these questions are not just from one-time users with no interest in broader participation.) $\endgroup$
    – GeoMatt22
    Sep 7, 2016 at 2:34
  • $\begingroup$ @Geomatt If there's a way to reword the question to make the answer more than just "yes" then it's great if the poster answers their own question. But then again, if they can reword the question to do that, why not just reword the question in the first place and let anybody answer it (including themselves)? It's not clear to me that the OP answering with "Yes" is any better than if someone else (you, or myself say) doing the same. $\endgroup$
    – Glen_b
    Sep 7, 2016 at 2:43
  • $\begingroup$ @Glen_b I was thinking that an authoritative "yes" in the comments might give them enough confidence to reword and then answer their question. (As in they change the posted question to be the one from their teacher/book, and then post their solution as the answer.) $\endgroup$
    – GeoMatt22
    Sep 7, 2016 at 2:48
  • $\begingroup$ @GeoMatt In a few cases, perhaps, and with encouragement to do so, perhaps a few more. More generally, they see a 'yes' and leave. $\endgroup$
    – Glen_b
    Sep 7, 2016 at 2:51

2 Answers 2

10
$\begingroup$

I feel there usually ought to be more to say. If you ask someone to check your working you want to increase your confidence that you've got the right answer by having them agree with you; but you also want to to learn new ways to check similar problems—quick checks to show your results are roughly what they should be, an intuitive explanation of the answer, proofs or references for methods that you used just because they seemed reasonable, how to arrive at the same results by a different method, the making explicit of assumptions you'd made implicitly, &c. And your confidence increases more when they give such reasons for agreement than when they just say "looks all right"—which meshes nicely with our goal of providing not over-short answers that may be useful to more than one person.

  • do we tolerate "yes" answers to these questions?

I think not, for the same reasons we don't tolerate answers like "Use a t-test" or "Read this blog post". They should be comments, even if they appear to satisfy the OP.

  • do we try harder to push OPs to ask questions that have longer-than-one-line answers?

I think so; as I commented, there's often going to be something in particular they're unsure about. But we don't want to make people jump through hoops & insist on their editing their questions just because the wording allows "yes" as a potential answer. Wouldn't it be a little picky to insist on changing "Is this the correct way […] ?" in Changing Sigma Limits for Proof ?

  • do we come up with a different line altogether, anything from closing them to putting up with having a large set of questions that can't be properly answered

I think we should place questions equivalent to a big "code dump" on hold, as good answers would be impractically broad.

[https://stats.stackexchange.com/q/171054/17230 is one where the OP's approach is perhaps acceptable to whoever's marking their homework, & where perhaps they just hope someone will check the basic calculations. Nonetheless they'd surely benefit from being prompted to at least consider—it's a self-study question, so they shouldn't simply be told—under what conditions it'll be good enough for government work.]

$\endgroup$
1
  • 5
    $\begingroup$ +1 There is often an alternative method, a "shortcut", a sanity check, a verification-by-simulation, ... Somebody who is asking "is this right?" is in some ways implicitly asking and how can I know if it is? $\endgroup$
    – Silverfish
    Sep 8, 2015 at 18:27
6
$\begingroup$

To borrow from @cardinal, sometimes rules should be treated as 'guidelines'. In that vein, if the complete answer really were "Yes.", I think that should be allowed, although in general I think answers that could be elaborated at least a little should be. For example, answers that are very short and do not yet have any upvotes are automatically entered into the low quality review queue. I have seen answers that I would typically think are too short, but that do fully address the question upon inspection, and voted Looks OK numerous times.


On the other hand, there used to be a close vote reason called Too Localized:

This question is unlikely to help any future visitors; it is only relevant to a small geographic area, a specific moment in time, or an extraordinarily narrow situation that is not generally applicable to the worldwide audience of the internet.

This was removed from the list of possible close reasons I gather, in large part, because it was believed to be misused. But it was not considered to be necessarily an invalid close reason.

5(A). "Too Localized" is no longer necessary since the specific off-topic reasons now address its main use case. (new)

Too Localized was, by far, the most misused close reason in our surveys, with both Community Managers and Moderators deeming over 50% of randomly sampled TL closures to not have merited closure (including on SO).

Today, TL is used helpfully on code dump questions on SO, but the new OT reasons are where this is now correctly addressed. SO can use "Large blocks of code with requests for debugging without meaningful supporting info" as a specific OT reason.

I would argue that the questions you are describing are somewhat analogous (for us) to "code dump" questions on Stack Overflow.


@Scortchi's comment strikes me as the ideal solution. If the answer cannot be broadened into something that could be of wider utility, I think a simple "Yes." (as an official answer, or in comments if you are more comfortable that way) and closing the question seems the appropriate way to handle these.

$\endgroup$

You must log in to answer this question.

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