4

Unless I screwed up this query somehow, there are a lot of missing check marks here.

Is this normal for SE these days? If not, why is the Writing site different? I tried this query on a handful of other sites and saw more "missing checkmarks" than I expected. English and Worldbuilding have the same issue, maybe for similar reasons?

Other sites like Space Exploration and Code Review are closer to one third unaccepted. These are just randomly selected examples, not sure how to run this query against every site on the network.

Anyway, what gives? Some of these questions have tons of good answers, but without the checkmark I'm unsure how they should be handled. Answer them if I haven't already? Try to make my answer more persuasive if I've already answered? Or just assume OP isn't gonna accept one no matter what?


There's an interesting paper here on unresolved questions at Stack Overflow. It concludes with the following sentence:

Since the unresolved questions are less helpful for problem solving and are increasing rapidly, our models can assist in automatically identifying them for necessary quality management.

Some might argue that unresolved questions are not a problem on Writing or other sites on the network as they are on Stack Overflow, because the material is more subjective. But I would argue to the contrary: because the material is more subjective, it is even more important to know which answer was the most acceptable to the author of each question. Because of the subjective nature of both the questions and the answers, it is much more difficult to tell if asker and answerers of unresolved questions are all on the same page.

On Stack Overflow, anyone finding an unresolved question that they need an answer to can independently verify whether an existing answer addresses the question in an acceptable way in a matter of minutes, with minimal effort. On Writing, it could take months. This suggests that proper resolutions to questions are much more valuable here.

| |
  • 9
    Speculation: the more subjective a question is, the more likely it is that there are several good answers that together make a great answer, but choosing just one of them is hard so people punt. Maybe compare rates on science sites to those on, say, Workplace, Parenting, IPS, and some others to see if that's actually a pattern? (I'm not saying we're as subjective as all of them; I'm just wondering if there's a relevant spectrum there.) – Monica Cellio Aug 15 '19 at 22:01
  • 1
    Does the rep of the asker or their longevity on writing indicate any trend? – NofP Aug 15 '19 at 22:19
  • data.stackexchange.com/writers/query/edit/1091738 . hard to say... – NofP Aug 15 '19 at 22:46
  • @MonicaCellio Workplace is similar, Parenting is worse with more than half with no checkmark, and IPS is slightly better; around 2/5 with no checkmark. Definitely looks like some correlation with subjectivity, but if people are just punting, that's too bad. Knowing what helped the asker can help with future answers, and could improve quality of other Qs over time (i.e. identifying misuse of tags, frame challenges and so on; things in Q that didn't quite jive with accepted answer). – Thing-um-a-jig Aug 16 '19 at 5:54
  • I wonder if people should be encouraged to self-answer instead of punting, indicating parts of other answers that were useful to them. These could always be marked as CW if self-answering feels too tacky. – Thing-um-a-jig Aug 16 '19 at 6:07
  • 1
    I frequently ask on the main site and It's always very difficult to accept just one answer when there are multiple great answers, all with valid facts and content. And so I end up keeping all answers upvoted but no answer ticked with green. – Karan Desai Aug 17 '19 at 9:45
  • @KaranDesai if you go to buy chicken and you see chicken legs, chicken thighs and chicken breast, and you like them all equally, do you come back with no chicken? – Thing-um-a-jig Aug 17 '19 at 22:13
  • 1
    @Thing-um-a-jig I see chicken legs, chicken thighs and chicken breast; maybe I'm having guests, so I buy all three to create a variety of dishes. Or maybe what I need is actually chicken livers, so I might like the legs, thighs and breast, but they're not useful to me. – Galastel supports GoFundMonica Aug 18 '19 at 8:45
  • 3
    Yes I come back with no chicken if the shop has a strict policy of buying only one item instead of all (There can only be one accepted answer and I don't want that visibility of other answers significantly reduce due to one green tick) – Karan Desai Aug 19 '19 at 5:21
  • @KaranDesai and how are we gonna have a cookout with no chicken? Check this out... 8 answers, none of them say "just punt." Have times really changed so much here at SE? – Thing-um-a-jig Aug 19 '19 at 23:45
  • (In case the cookout analogy wasn't clear, what I'm saying is while some of the regulars might not have an issue with this now, because they can just read everything, I don't see how this could possibly scale well if the the amount of questions ever picks up significantly.) – Thing-um-a-jig Aug 19 '19 at 23:53
  • Hey, Ironically that meta-question itself doesn't have accepted answer :p However, I got your point @Thing-um-a-jig – Karan Desai Aug 20 '19 at 4:49
  • @KaranDesai it is tagged discussion, but I'll write one up if it helps add to the discussion. – Thing-um-a-jig Aug 20 '19 at 20:39
7

I cannot answer for others, but I can answer for myself. Sometimes, I am answered, I can accept an answer quickly, great. But as often as not, I go and try one approach, which takes time to write, see if it works, play with it, figure out if I need to improve something with this approach or try another...

Sometimes it takes several months before I can figure out what "worked". This doesn't mean that the other answers didn't help. As often as not, what "worked" would be an amalgam of several answers + elements of my own. And even if not, exploring paths that ended up not working is a learning experience too.

Once in a while, I go over my old questions, and mark answers which I found helped. Or I don't - maybe I'm still struggling with the passage for some reason, or maybe I've decided to remove it entirely, making the question no longer relevant, or maybe I used such an amalgam of everything that I cannot mark one answer as "best".

This is different from upvoting: I upvote straight away every answer that looks like it might give me something useful.

| |
  • 2
    That makes perfect sense. If you haven't gotten feedback yourself yet, there's no way to give feedback on whether the answers were helpful. It's reassuring to know that you eventually come back to it. I'm still of the opinion that if you come back to it and none of the answers have helped enough to be acceptable, but you have made a decision about what to do, then self-answering is a good course of action, but that's just my opinion. I don't know if SE staff feels any particular way about answered-to-accepted ratio as a site health indicator (that's really what this question is about). – Thing-um-a-jig Aug 16 '19 at 23:00
  • Here's something interesting. I whipped up this query to see what happens when users stick around for a number of months after asking their questions. That is, what does unresolved rate look like for users who were last seen at least N months after asking their question. Over 12 months, the unresolved rate drops from around the halfway mark to around the one-third mark. Got similar trends on other "subjective" sites, better unresolved rates on less subjective sites. I think you're on to something here, check it out. – Thing-um-a-jig Aug 17 '19 at 1:53
  • 3
    @Thing-um-a-jig SE kind-of prefers not to care about "accepted answer" (ref: Let's stop displaying a user's accept rate) because it's the OP's right, not a necessity... and also any upvoted answer is considered "answered" by SE's system. Also, based on my experience as a reader, sometimes the accepted answer is useful, but not the right solution for my specific case (which is handled by another unaccepted answer). And another experience was that I dropped the whole project related to my question, thus I can't evaluate/try any answers at all. – Andrew T. Aug 18 '19 at 10:07
  • I don’t interpret that move as not caring about accepting answers. I remember when that all went down, people were getting singled out for having low accept rates and staff decided the value of not having them feel ganged up on outweighed the value of question resolution. This doesn’t mean question resolution has no value; it does. – Thing-um-a-jig Aug 18 '19 at 11:51
  • 1) Resolved questions are clearly marked with a big green square so we can see from the list whether the asker found an acceptable answer. I always look at these first when looking for answers. 2) Resolved questions help answerers write better answers, because they see which answers the people asking the questions like best (as opposed to a bunch of random votes from HNQ). 3) Once the thing is properly resolved it signals to other potential answerers that they can leave it alone and focus on an unresolved question where someone actually still needs help. – Thing-um-a-jig Aug 18 '19 at 11:52
  • 3 reasons why question resolution is helpful. In cases where an answer solved your problem, is there any reason why resolution is harmful? Sure, better answers might come along some day, but if someone has something significantly better, they can still answer. You can game the system and farm some rep by leaving it open, but that’s selfish. It can be hard to decide which one is best if they’re all good, but it’s not gonna kill you to pick one. If you don’t have a favorite, come up with a tiebreaker system you like (earliest post, most votes, most fancy words, alphabetical order, w/e). – Thing-um-a-jig Aug 18 '19 at 11:53
6

Not every question needs to have an accepted answer.

Basically, the checkmark says "this works for me" or "this answer was the most useful one to me", where "I" is the user posting the question.

If you feel that you can post an answer that adds something not already said in an answer, then by all means go ahead, whether or not the question has an accepted answer.

I wouldn't try to make an answer "more persuasive" just for the sake of that. However, if you can add something more which is relevant to the question to an answer of yours, then by all means do it.

Community wiki self-answers are a tricky beast for a number of reasons. If their only purpose is to say "I used part X from User A's answer, and part Y from User B's answer, and part Z from User C's answer", then I don't think they add much to the corpus of knowledge, and thus don't benefit anyone (other than possibly giving the OP the ability to accept the self-answer). If OP has points that on their own are an answer to the question, then a non-CW self-answer would be a reasonable choice.

Community wiki answers were introduced before suggested edits were a thing, and live on as sort of a misunderstood historical artefact in the Stack Exchange ecosystem.

| |
  • 1
    The thing is, when anybody else facing the same problem turns up one of these questions and finds no indication of which answer solved it, it's much less useful. With no checkmark, there's no way to know whether none of the answers solved it, or one of them did, or some did and some didn't. If people aren't going to leave a checkmark when some of the answers did solve their problem, self-answering would be another way to indicate that (it doesn't have to be checked, the message is clear). Votes used to kind of work for this, but with HNQ voting patterns often make no sense anymore. – Thing-um-a-jig Aug 16 '19 at 13:21
  • 1
    This is all based on experience with other sites on SE, but this is the first "subjective" site I've posted anything on, so maybe it's different here. But I can't imagine it's so different that signalling which answer(s) helped the asker is completely useless. Besides others who may come along with the same problem, I'm thinking of a few questions in particular where some indication of what worked and what didn't would help me understand how to make better contributions to the site. I'm assuming I'm not the only one in that position. – Thing-um-a-jig Aug 16 '19 at 13:29
  • @Thing-um There is an element you're missing, precisely because writing is so subjective. We make our questions broad, because "too specific" is bad. But what happens then is, answer A might be the one that worked best for my particular issue, but answer B would work best for you with a very similar issue. Or, even without the issue of "too specific", it might be that answers A and B are equally valid, though cardinally opposite, and I chose answer A because it works better with my sensibilities - I consider it more "beautiful". Does it imply that answer A is the one you should implement? – Galastel supports GoFundMonica Aug 16 '19 at 16:54
  • @Galastel I'm not really following. If A and B both solve the issue, and both say similar things, you can pick whichever one you like; this signals to both A and B that they were on the right track with their answers and that they properly understood the question. It seems even more important here than on the less subjective sites. We can know for sure on those sites that our answers are correct without any feedback from asker. The crucial thing here is that without any feedback at all, there's no distinction between "all these answers were garbage" and "all these answers were great." – Thing-um-a-jig Aug 16 '19 at 17:27
  • 1
    @Thing-um-a-jig But what if A and B say things that are very very different, yet both are valid solutions? – Galastel supports GoFundMonica Aug 16 '19 at 17:35
  • @Galastel personally I'd give one the checkmark and comment on the other (assuming one of them actually set me in the right direction). By the way, the question that's really nagging at me most might not be the one you think. I took this particular question in a different way than the other two answers. If any of the answers had a checkmark, I would know whether to proceed with suggesting a new tag. I think that question was improperly tagged and led other answers down the wrong path, but without any feedback, I'm not sure. – Thing-um-a-jig Aug 16 '19 at 17:43
  • @Thing-um-a-jig Ah, I see the problem with that particular question. Your answer did make me wonder if the question was mistagged, or if, at least, a tag was missing. You can suggest a new tag, it might be useful regardless of what the particular person intended. – Galastel supports GoFundMonica Aug 16 '19 at 18:22
  • 3
    @Thing I get the feeling that you are treating the accept checkmark as an indication that an answer is "correct". It's not. It's an indication that the answer in question was the most helpful to the person asking the question; nothing more, nothing less. I've seen examples on far less subjective sites in the network where the accepted answer is flat out wrong. It's unlikely that a garbage answer would end up voted highly by the community anyway, so there's still signal. And you could upvote and comment "this was helpful" on both answers without accepting either, if both were equally helpful. – user Aug 16 '19 at 18:33
  • @aCVn I'm not sure why you get that feeling. If I said anything to suggest that I think that, I misspoke. I know what it's for, and am only suggesting that it should be actually used (when there is at least one helpful answer), rather than not using it at all. – Thing-um-a-jig Aug 16 '19 at 18:44
  • If you're talking about "we can know for sure on those sites that our answers are correct without any feedback from asker," that is exactly what I meant to say. I can answer questions on those sites and know that the answer correctly addresses the problem. I don't need any feedback there, I can test it myself. I get that there are no "correct" answers here, and I can't test it myself, and so I think feedback is even more important here. I never meant to suggest that the checkmark had anything to do with correctness. – Thing-um-a-jig Aug 17 '19 at 2:10
  • 4
    @Thing-um-a-jig: But the nature of this site also means that the asker may not be the one best qualified to judge whether an answer really solves the problem. On Stack Overflow it's generally easy: If the code does what you want it to do, the solution was helpful. On this site, it might be that you got a terrible advice, but don't see that it was terrible, and chose to follow it, and possibly never learn how bad that was because for completely unrelated reasons you removed or replaced the scene or chapter where you used the advice, and no one but you ever got to see the result of that advice. – celtschk Aug 17 '19 at 10:35
  • @celtschk fair enough, but on the other hand if nobody ever got to see the result, we could say the answer was never really tested and the author of the question doesn't know if it was helpful, so no check mark is appropriate. Maybe that's the real answer here... less stuff gets tested, so there's a higher unresolved rate (which sounds like a different problem entirely; presumably the point of these answers is that they'll be applied to something). – Thing-um-a-jig Aug 17 '19 at 18:46
  • @Thing-um-a-jig I still think votes are the better signal. Even on technical sites, it's almost always worth reading the non-accepted highly voted answers as they can cover other facets of the problem. I accepted an answer on the question I asked here, but at the time it was the only answer posted and it was posted too late to actually be of help in the situation, and it bugs me to not see the green box filled in. – Troyen Aug 20 '19 at 8:11
  • 1
    @Thing-um-a-jig I think you may be underestimating how many times the asker just wants to give up on the question and move on, either because the issue is rendered moot or they don't feel they're going to get anything more of value. – Troyen Aug 20 '19 at 23:27
  • 1
    I've also asked a few questions over the years where, after considering and trying solutions proposed in answers, I've come up with a different solution. The latter is one time when a self-answer absolutely is appropriate. – user Aug 21 '19 at 6:35
4

I don't ask any questions on Writing. That said, I put very little stock in the checked answer. That's great, it helped one person, but it may not help the most people that have this question or a similar question. So I am more interested in the upvotes (and downvotes) than I am in which one was accepted.

Likewise, if I read a question and have an answer of my own, I will first check and see if others have given my answer. I will upvote those that capture some or all of my initial thoughts (lookin' at you, Galastel!), and if I can't think of anything else to add after reading the existing answers, I won't answer.

Sometimes I can think of another angle to talk about, and will add that, with a hat tip to the best of other answers to explain what I already agree is important.

I think more answers are better, even on old questions where the poster has moved on or deleted their account. Part of the philosophy of this site is we actually want users to read old answers. Any answer you write is intended to be read by more than just the original questioner; we are trying to build a body of expertise and ideas around every one of these questions.

And Writing is not Stack Overflow asking about computer language syntax issues or Linux commands, Writing is less constrained, more open to innovation, and you cannot "prove" one approach or way of thinking about a story is better than another.

I'm happy to know which answer a poster found most helpful, but that has nothing to do with whether I post an answer or not, and whether I upvote it or downvote it or leave it alone. I am usually talking to the wider audience, and trying to use the specific question as a lens through which we can discern a more general rule or technique in writing.

| |
2

Yes, it it appears to be normal for low-traffic, subjective SE sites to have around a 50% unresolved rate.

At least on Writing, some of the regulars accept and defend this, and actively participate in raising the unresolved rate by not accepting an answer on a question that was answered in an acceptable way (especially in cases where they found multiple answers to be acceptable), or not asking for more information or attempting to clarify the question on questions that had no acceptable answer.

Here is a list of unresolved rates by user, as evidence that it's not only new users who are contributing to this, but also experienced users.

The usual advice on SE is to accept the most helpful answer. If multiple answers are helpful, the suggested approach is that all of the helpful parts of those answers should be edited into one canonical answer. If no answers are acceptable, the suggested approach is that the asker should clarify their question or request further information so that an acceptable answer may be given. That advice does not seem to be followed on this site.

| |
1

Disclaimer: SQL and I are not close friends.

https://data.stackexchange.com/writers/query/edit/1091750#graph

If you run the query, you should get: two lines, with the same colour... a nightmare.

They represent the question counts divided by questions with and without accepted answers, grouped by buckets of reputation of the asker.

It seems that the odds are 8:2.5 for low reps (less than 30?) to not accept an answer, and 7:2 for high rep (10k?) to not accept an answer. Higher reps and medium reps users tend to have a 50:50 chance of accepting.

| |
  • Strange, not sure what to make of that. I could understand the feedback loop taking longer here than somewhere like SO, since programmers can test whether things work right away while authors need to wait for next group of beta readers to see if it "works." Length of time active on site could play into that; the lower rep users might never come back to accept an answer. Could test this by splitting question age into buckets, maybe. But the high rep users accepting less than the medium reps or very high reps, I can't think of any reason for. All of these groups should know how the site works. – Thing-um-a-jig Aug 16 '19 at 1:35
  • @Thing-um-a-jig regarding "low-rep users", everyone can ask a question without registering on Writing.SE, and unregistered (cookie-based) users losing their access to their account is a common occurrence, and thus they cannot interact with their own questions anymore (including commenting, upvoting, and of course, accepting an answer) – Andrew T. Aug 18 '19 at 10:15
  • 1
    Yeah, there are definitely a lot of those "fire and forget" users; check that SEDE query I posted on Galastel's answer, it's that sharp drop at the front. That does explain a good chunk of the unresolved stuff, but there's a lot more than that going on. I think the "punting" thing is definitely an issue. Encouraging people to punt is, IMO, A Bad Thing (TM). – Thing-um-a-jig Aug 18 '19 at 11:59
  • 1
    @NofP alright, here's my new theory, FWIW. The lower-rep users are fire-and-forget users or don't care enough to accept answers. The medium rep users are asking not-too-hard questions that get good answers, and they accept one. Both the high and very high reps ask hard questions, so some can't be answered in an acceptable way. The very high reps care a lot about the site and try not to punt, but some high-reps and mods have convinced themselves and others that resolution is unimportant, so they (and others) punt. data.stackexchange.com/writers/query/1093013 <- some surprises here. – Thing-um-a-jig Aug 19 '19 at 4:19
  • 1
    @thingumajig excellent insight. I'd strongly recommend adding it as an answer. – NofP Aug 19 '19 at 10:48
  • @NofP At this point I suspect a lot of people reading this have already taken a position that they know on some level runs counter to the way SE is supposed to operate, and are going to defend that position without addressing any of the usual reasons that resolution is useful. I don't know that anything else I can say will do any good. Anyway, I guess since this is tagged as discussion it's good enough to leave it as a comment. At least we can have a discussion here without thinly disguising it as a QA session. :p – Thing-um-a-jig Aug 20 '19 at 1:01
  • @Thing-um-a-jig I see your point. My suggestion was mostly to give it the visibility it deserves. Also, I don't buy into the answer that it takes months to see a result... the moment one tries a solution rather than another they have already decided which solution they liked more. – NofP Aug 20 '19 at 6:51
  • 1
    @NofP I tend to agree with that. If Q was explicitly (or implicitly, I guess) about beta readers not liking something, and what can be done to get them to like it, it's fair that it could take months. The rest of the time I think you're right; an answer can immediately solve a problem, be properly accepted, and if it turns out readers don't like it later, it can be unmarked or moved to a different answer. – Thing-um-a-jig Aug 20 '19 at 20:43

You must log in to answer this question.

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