-1

I used to have a link, which I have lost, which would show me all of my questions which I have asked but not accepted an answer to.

I would use it regularly to make sure that I was being a good member of the community and would then either accept an answer, delete a question which no on had replied, to or accept my own answer "dear future searcher, there seems to be no answer for this".

I don't want to debate whether that is the optimal way to handle things (although we could start another question to discuss it (if there isn't one already?)).

What I do wonder is whether we should have such a link on the profile page.

Not only would it help users to tidy up their own unanswered questions, but, before answering a question from a low point account we could check if they have a history of asking questions and not awarding an answer.

This is not, as it might seem, a chasing after points. It is meant to allow us to point out to such users that, if questions are not awarded an answer, then future seekers with the same problem will be frustrated (the point, after all, it that SE are Q and A sites).

What do you think?

4

I see two aspects of this request. On one hand, you want to see this page for your account. On the other, you want to view it for someone else's.

Check your own

I don't think this is super useful. It wouldn't hurt (assuming resources are infinite, which they certainly are not), but for a significant majority of users, it wouldn't help.

I ran a query on SEDE to test how many users ask a lot of questions. Looking only at users with over (what I believe to be a very generous) 500 rep points who have asked at least one question, it appears to me that a feature such as this would be useful to a very small percentage of users. It should be noted that this rep filter does amend a certain bias towards answerers, since answerers receive more rep per upvote. You can remove that filter in my query, or just trust me to know that I actually added that to make your case look better (mostly by filtering out one-and-done, drive-by users). Also, note that this query was against Stack Overflow, just because it's (presumably) what you're talking about, and it's the most data. I suspect other sites run similar statistics, though, probably skewed even less.

Graph of users asking questions

(Count of questions asked on the Y-axis, count of users on the X-axis)

For those who find that data speaks louder than graphs, these are the data for users with one through twelve questions asked. The first column is how many questions were asked, the second column is how many users asked that number of questions, and the third column is, and this is important, the percentage of included users who asked that number or fewer.

In other words, the first row tells us that 9,678 users asked exactly one question, and that made up for 8.8% of the polled users. The second row tells us that 7,425 users asked exactly two questions, and 15.5% of polled users asked one or two questions. 21.1% of polled users asked one, two, or three, and so on.

 Q Count  User Count  Percentage
 1        9678         8.8%
 2        7425        15.5%
 3        6170        21.1%
 4        5325        25.9%
 5        4642        30.1%
 6        4170        33.9%
 7        3756        37.3%
 8        3277        40.2%
 9        3134        43.1%
10        2760        45.6%
11        2600        47.9%
12        2346        50.1%

You can't know this from only the data I've presented, but for 92.6% of users, three pages is more than enough to review all of the questions they've asked (assuming the current default of thirty per page). That's not very many pages for the vast majority of users. It's important to remember as well that this is limited only to users who have asked at least one question, and who have at least 500 rep. These numbers would be even more heavily skewed if I included all data.

You're definitely in the upper percentile (the upper 99.92nd percentile, from my data) of askers, which is awesome! So I can definitely see why this might be a problem for your case, but unfortunately it is an edge case, and I don't personally see it being worth the resources of the Stack Exchange dev team to work on a feature that only saves 8% of users anything more than two clicks, and only 3% any more than four clicks.

Beyond all of that, think about how most users use the question-asking system: to find help. In most cases, theoretically at least, a user will ask a question, get an answer, and know whether it helps them or not at that time. It becomes hard to keep track as they start piling up, or for those of us with some delay between their first couple questions and starting to really use Stack Overflow, but in many cases, acceptance doesn't need a second review.

Check someone else's

Displaying Acceptance Rate of an account is a pretty debated topic, that was overruled by the community and Stack Exchange team, for a reason similar to what I fear this would lead to.

It was around before I was, so I'm not really in any position to comment on it aside from hearsay, most of which is illustrated nicely by that link. Ultimately, in practice, displaying information that specifically highlights whether a user typically accepts answers or not isn't super useful.

On top of this, the same issue arises as was previously mentioned. It's such a small number of cases where you really need to see an aggregated or summarized view of this, that it doesn't help usability. And realistically, seeing that someone has twenty questions without accepted answers is only as useful as knowing the percentage of questions for which that's true. Maybe they've asked six hundred questions, and accepted 580. Maybe they've asked twenty and accepted zero. You really need the accept rate to be able to understand this data about someone else's profile, and the accept rate just causes other problems.

Conclusion

I just don't think this is useful in most cases. It'd be handy, and again if resources were infinite, I would support it. But even this generously filtered data doesn't back up its use for most users, especially experienced ones.

Realistically, it's more useful now to just use an SEDE query to find the same information. I've separated these answers out such that you can ignore these opinions should you see so fit, without compromising the workaround I've (hopefully) presented.

I think an important part of all of this is also how active the user is. It sounds like you're looking (honorably, if I do say so myself) to teach new users the importance of accepting answers that do answer their question. I'm not sure offering a page to do this will necessarily encourage them, because most users with this problem are likely relatively new. A majority of users who actively want to find these kinds of questions don't need an extra reminder about how to, as I've previously asserted. I definitely respect your intentions, but I think we need an alternative way to teach new users to accept, and once that's in-place, this will only help a small percentage of users.

  • Some nice work there. Quite convincing. Although I still think it might help if we can identify those who haven't accepted the culture that accepting an answer helps those who come afterwards. – Mawg Feb 11 '15 at 14:03
4

I already answered this with my opinion that a page isn't necessary, but I'd be remiss if I didn't offer you another option. This should be a viable workaround for not having the page you're looking for, all opinions aside.


Edit: I suppose I got a bit excited with all this, and failed to recognize the comment left by gnat, which appropriately suggests that using the built-in search functionality is completely sufficient to do what you're looking for.

To construct that search query, you'll want to filter by user and hasaccepted. For you on Stack Overflow, I believe that should look something like:

user:me hasaccepted:no

But, if you're like me and enjoy the over-complication of simple tasks, here's a nice step in that direction. That said, I think this searching option is actually more reliable for a couple reasons, including the effect of caching, and the overall complexity of my SEDE approach.

In case you lose or forget this again, you can reengineer it by visiting the How do I search? help article. There is also a dropdown under the "Advanced Search Tips" link on the Search page. I'd typically use the latter, but the first one is easier to link to. In the dropdown, look for "author" (because you want your own questions), and you'll find that "user:me" will work fine (you can use a user id if need be). Then, you want a question with no accepted answers, so you can look under the "answers" header to find "has accepted:no," which means that there is no accepted answer.


This is an SEDE query that I believe does what you're looking for. Someone else may make it a bit prettier, and perhaps more efficient (I was feeling a LEFT OUTER JOIN vibe, but a nested subquery might be cleaner), but this should tell you a few things. Specifically, it will tell you which questions a given user has asked which have no accepted answer, and for each of those, how many answers are present.

SELECT DISTINCT questions.Id AS [Post Link], SUM(IIF(answers.Id IS NULL, 0, 1)) AS AnswerCount
FROM Posts questions
  LEFT OUTER JOIN Posts answers ON questions.Id = answers.ParentId
WHERE questions.PostTypeId = 1
      AND (answers.PostTypeId = 2 OR answers.PostTypeId IS NULL)
      AND questions.OwnerUserId = '##UserId##'
      AND questions.AcceptedAnswerId IS NULL
GROUP BY questions.Id

It may be useful to add an ORDER BY clause, but considering you only have a few that show up, I didn't think it necessary to decide on what it made sense or order. It occurs to me while writing this that I'm also not sure whether including the question's score would be useful, but I'm too lazy (now I choose to be lazy) to add that. It'd be pretty straight-forward if you need it, though. Same with tags, or whatever you may need from each post.

One thing to note, although I suspect not relevant, SEDE is only updated weekly (I believe on Mondays, 0300 UTC jumps to mind). I figure that's not such a hindrance, since typically (I imagine) you'd be well-aware of any questions you've asked or answers you've received in the week prior, but it's something to remember when digesting these results--they can be subject to caching.

  • 1
    No need for SEDE. It's available via the built in search, as gnat posted in this comment – Shadow The Dragon Wizard Feb 11 '15 at 12:43
  • @ShadowWizard You say that as if over-complicating things is a bad thing! But yes, touché. I have prepended that to my answer. Thanks. – Matthew Haugen Feb 11 '15 at 12:52
  • 1
    Agreed. BUT, please see my comment to Gnat's comment – Mawg Feb 11 '15 at 13:56
  • 1
    @Mawg I'm not sure if what I said really helps, but I added a paragraph (the last one before the second horizontal line) that explains how you could build that search again if you lost it. Definitely not an alternative to actually linking to it, but at least as long as you remember that you have to search, that may help. – Matthew Haugen Feb 11 '15 at 21:40
  • Great stuff, but not everyone is a SEDE genius, alas. I would still like to have a one-click link to my own unaccepted questions on my profile page in each site. And one which covers all sites in my S.E profile – Mawg Jul 14 '15 at 7:42

You must log in to answer this question.

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