44

We currently have just over 19,000 unanswered questions.

After reviewing about 500 of them in a totally unsystematic manner I conclude that <2% are truly reasonable questions awaiting a reasonable response. That is a few hundred at most.

Maybe we could work through these questions and:

  • answer the ones that deserve an answer,
  • do whatever we do with duplicates, and
  • downvote the poor questions.

I made a start this week answering a handful of ancient questions I thought I could tackle. In quite a few I think a nascent answer may be in the comments and others have a solid answer that has not been accepted, so we could ask the question asker to accept an answer.

I do not know which other sites are comparable to ours but Academic SE has <150 questions without an accepted answer.

Do we have a higher percentage than other comparable exchanges?

  • 6
    Closely related: questions answered by comments and Does GIS have a low voting problem?. The latter provides partial answers to your initial questions. – whuber Nov 6 '12 at 17:16
  • This is now featured. Could you either provide an update in the question or post an answer with new statistics to see how whether it's going the right way or not? – Mast Jun 19 '17 at 13:44
  • 1
    @Mast it has been updated already. If you click the link at the top it will take you to the unanswered questions page which says there are (as at time of writing this comment) 18,500 questions. – Midavalo Jun 20 '17 at 3:43
  • 2
    @Mast 18 months ago we were at 15-16K unanswered and now we are at 18.5K so, unless users want to volunteer more time towards answering and voting, we may need to get stricter on question quality to help line them up for the volunteer resources which we are able to muster. On the positive side 18 months ago we were at 73% answered whereas we are now at 78%. – PolyGeo Jun 20 '17 at 6:15
30

At 81% answered we are better on that particular statistic than Stack Overflow (everyone's measuring stick) at 71% and Cross Validated (which launched at the same time as GIS SE) at 61% - see http://stackexchange.com/sites#percentanswered - but I am all for trying to get us up into the 90s - which is what Beta sites strive for. I would tell you our precise ranking if I did not have to count to find it: Numbering the List View of the SE Sites site?

Only two sites out of 173 receive more questions per day than we do AND have a higher percentage of questions answered than we do. The top four sites on the Stack Exchange network by this measure are:

In many cases to move a question off our unanswered list, all it needs is for someone to upvote at least one answer to > 0 votes.

Improving wording of "non-reproducible" custom close reason? was posted here, and then implemented, to give us a slightly stronger tool to use to close some of the questions that are in no shape to be answered.

If you are wondering which tags result in the most unanswered questions then this SEDE query can be used to list them. The snapshots below are ordered first by the number of unanswered questions, and then by the percentage of unanswered questions.

enter image description here

enter image description here

I found the query to do the above via asking a Meta SE question: Identifying tags with poorest answer rates on a site?

  • On Spanish SE, the answered count is 100%, but it only gets 2.9 questions a day, which makes is somewhat less impressive. I suspect that when a site is in beta it attracts a higher proportion of experts/enthusiasts for the subject, which explains the high answer ratio. As a site gets established, the ratio of awful, unanswerable questions inevitably increases . I spent some time on the triage review queue on SO last week, as I simply had to get the El Dorado hat, a truly depressing experience. So, actually, a lower number might be indicative of success. – John Powell Dec 22 '15 at 9:39
  • This makes Sense. – If you do not know- just GIS Dec 22 '15 at 12:48
  • @polygeo Nice! Thanks for doing that, and thanks for providing a link to explore the tags!!! – DPSSpatial Dec 22 '15 at 23:40
  • Most of such tags are related somehow with 'serving GIS data' (interesting). – Andre Silva Dec 23 '15 at 13:01
  • Yes as a percent the serving GIS amount is high but in raw count the two major softwares are ~6k. I am up to about 10! Only 59,990 to go. – If you do not know- just GIS Dec 23 '15 at 14:49
  • @JohnBarça Coming from a beta site, I would say that it's the lower question rate. The lower question rate allows us to focus and improve each question individually, while providing it many quality answers. On Open Source SE, I've actually noticed that when we get a bounce in questions, the number of answers actually goes down. But I highly doubt that it would have anything to do with an "expert/enthusiast" ratio. – Zizouz212 Jul 8 '16 at 1:45
  • 1
    I thought I would start making a dent into the unanswered questions, starting with Postgis, which currently stands at 21% based on above query, mostly because there are some quite interesting questions in there and it helps clean up the site. I answered this question. The OP hasn't visited for over a year, so I'm certainly not expecting it to be accepted. I didn't get an upvote, even though I think it is a pretty acceptable answer. – John Powell Nov 5 '16 at 15:22
  • 1
    Do we have some means to monitor this situation beyond the late answers review queue. I am really way beyond the time when I cared much about rep, but it would be nice for it to be acknowledged, and more to the point, come off the unanswered stat count, or if it is not a good answer, perhaps someone could let me know why not. – John Powell Nov 5 '16 at 15:24
  • Thankyou for trying to help @JohnBarça and I agree that it can be a bit disheartening to go to the effort to provide answers to questions that no one else is attempting. As far as I know the system currently relies on the late answers queue or the active list to spot the questions that I would like to see upvotes on (when they deserve it). – PolyGeo Nov 5 '16 at 22:20
  • Alternatively, users could find yours using gis.stackexchange.com/search?q=user%3A429+is%3Aanswer+score%3A0 or to see that there are over 11,000 that potentially could come off our list with a single vote see gis.stackexchange.com/… – PolyGeo Nov 5 '16 at 22:23
  • but I am all for trying to get us up into the 90s - Is this still your goal @PolyGeo? Or do you think it is unrealistic/unachievable? If you think it is reachable, how do we do it? – Midavalo Nov 27 '17 at 4:21
  • @Midavalo I think 90+ would be sensational, and 80-90 eminently achievable, because we are currently at 78%. To achieve it I think we just need to be a bit tougher on old unanswered questions, while holding a minimum standard on new questions about where we have it. If a question is old and unanswered, then I start to think that it is as much the responsibility of its asker, as it is the potential answerers, to get it self-answered, closed or deleted. If the asker has lost interest in getting their question answered, or has left the site, then a close reason should be looked for. – PolyGeo Nov 27 '17 at 4:39
  • @Midavalo Also, if closure sometimes seems too harsh then a downvote can always nudge an unanswered question towards the Roomba. Downvote criteria include "unclear" and "not useful", and I think old unanswered meandering questions can often easily qualify as either. – PolyGeo Nov 27 '17 at 4:56
29

My opinion is that there are too many questions that get answered in comments and this has been discussed previously at:

I think that we should try to encourage people to write answers.

To qualify as a long enough answer, just 2-3 sentences is needed.

21

Recently I have been using a different search strategy to try and find those Unanswered questions where I am most likely to be able to make a difference.

  1. First I click the Unanswered option which is next to Ask Question (and not the one on the next row)

enter image description here

  1. Then I click my tags

enter image description here

  1. I then look along the bottom for the lowest voted Unanswered questions with my tags

enter image description here

Amongst these there are many that have an answer, but not one that has been upvoted, which makes them count as Unanswered. If there is a reasonable answer present I upvote it.

Reviewing some of them makes it apparent that answering them was either harder in the past or no one understanding the question, with time to answer it, had done so which makes some easy to answer.

There are often others there that match the criteria seen in the Deleting/Closing pointless questions? Q&A so they make good candidates for close voting (often as "non-reproducible" or "unclear" or "too broad" or "needs code snippet", whichever is appropriate).

If anyone is keen to make a dent in the current 20,000+ Unanswered questions you may like to try the same or a similar search strategy.

18

I've been looking into the open questions (only the ones related to tools I know).

One part of them contains potential solutions or pointers to solution in comments. Someone would have to migrate the main points from comment to answer but even then if the orignal author doesn't frequent the site anymore, they will never be accepted.

Some questions turn into feature requests. They ask if something is possible, the answer is "No, file a feature request". Usually nobody writes such an answer and/or accepts it.

Other questions require extensive knowledge about data and use case which the author can/does not provide. I guess those could be closed.

  • 3
    Good assessment. Regarding questions in which the OP will most likely never return, it's my understanding that an answer only needs an upvote to count, not an actual marking as accepted. When you look at the "unanswered" tab, it states "with no upvoted answer." So would upvoting one of the provided answers help? – Brett White Nov 6 '12 at 19:35
  • 6
    Brett, that gets back to the previous thread on low voting, where it has been pointed out that many questions (and answers) are so specialized that few community members feel qualified to vote on them. I don't think we want arbitrarily (and perhaps ignorantly) to upvote answers just to increase the answer ratio on our site. Assuming you agree, just who exactly is going to do the upvoting? (I periodically look through unanswered questions and have voted wherever I can, so I'm confident that I personally would be of no help in adding votes at this point.) – whuber Nov 7 '12 at 17:11
7

The others have identified the reasons why, but one possible solution:

Back when I first joined the site a few months ago, I went through a period where after checking the days questions, I'd use the rest of my 40 daily votes to try and vote up answers to questions that appeared to be correct (as well as under-rated questions) but hadn't been yet. Even managed to answer one or two as well.

I did this by going to "Unanswered", sorting by "newest" and then going to the very last page. Currently page 43 when using 50 a page and then working forward. It doesn't take that long though of course with only 40 votes a day I only managed to get through probably 2-3 thousand before I gave up.

Other folks may therefore want to try similar (although not everyone should start at the very "end" of time).

7

Just to bring this question up to date and back into the fold, here is some updated data on our site's answer rate. The Stack Exchange Data Explorer has a wealth of information that can be extracted with queries such as this one: Is the rate of answered questions declining?

The rate of questions with answers has slipped from 90% to 77% since Jan of 2013:

date        answered  total   ratio
-----------------------------------
5/1/2011    2459      2515    98%
6/1/2011    2836      2913    97%
7/1/2011    3203      3292    97%
8/1/2011    3542      3659    97%
9/1/2011    3893      4033    97%
10/1/2011   4213      4376    96%
11/1/2011   4564      4767    96%
12/1/2011   4899      5130    95%
1/1/2012    5173      5434    95%
2/1/2012    5538      5843    95%
3/1/2012    5995      6339    95%
4/1/2012    6479      6884    94%
5/1/2012    7062      7554    93%
6/1/2012    7725      8294    93%
7/1/2012    8355      9022    93%
8/1/2012    8999      9775    92%
9/1/2012    9604      10480   92%
10/1/2012   10176     11157   91%
11/1/2012   10918     12017   91%
12/1/2012   11710     12958   90%
1/1/2013    12354     13722   90%
2/1/2013    13137     14686   89%
3/1/2013    13819     15551   89%
4/1/2013    14585     16623   88%
5/1/2013    15317     17664   87%
6/1/2013    16032     18720   86%
7/1/2013    16675     19685   85%
8/1/2013    17364     20779   84%
9/1/2013    18057     21844   83%
10/1/2013   18757     22907   82%
11/1/2013   19525     24140   81%
12/1/2013   20189     25247   80%
1/1/2014    20758     26230   79%
2/1/2014    21518     27477   78%
3/1/2014    22151     28744   77%

If you compare this with other mature sites (by switching the site selector at the bottom of the query page) you'll see a similar trend. This obviously does not answer why this is happening, but I would assume it's a sign of maturity and probably closely follows the site's user population.

I found another very telling query here: Number of users answering or questioning.

Year Quarter Questioning Answering 
---- ------- ----------- --------- 
2009 3       1           1         
2010 1       2           2         
2010 2       2           8         
2010 3       215         253       
2010 4       231         226       
2011 1       382         330       
2011 2       497         396       
2011 3       533         404       
2011 4       554         424       
2012 1       796         550       
2012 2       1209        670       
2012 3       1259        772       
2012 4       1475        822       
2013 1       1743        974       
2013 2       1918        1007      
2013 3       1962        995       
2013 4       2058        1033      
2014 1       1693        819

This seems to confirm something that I've suspected has been going on, which is that there are a whole lot more people asking questions than answering questions than there used to be.

This is probably a natural progression with site maturity. When a site is young it is filled with enthusiastic, knowledgeable users who have been around since the dawn of (the site's) time and are both knowledgeable with the site (they know, for example, that answers yield twice the reputation that questions do) and the subject matter. As time goes on and word gets out about the site being a great resource to ask questions, the demographics change and you get a whole lot more people looking for information rather than offering it.

  • 1
    +1, but I think we can do better with more pro voting attitude around here. – Andre Silva Mar 4 '14 at 19:55
  • 1
    +1 and it may be worth mentioning that you would expect Questions available to Answer for 3.25 years will have a higher probability of being answered (or closed) than those there for only 30 days. – PolyGeo Mar 4 '14 at 21:52
  • 1
    That's a good point, the query probably does not account for when the question was answered. That's probably also why the number apparently never goes up. – blah238 Mar 5 '14 at 0:01
  • It's a pity the Visits per day data is not accessible to graph - I suspect that would give the clearest picture of an SE site's health over time. – PolyGeo Mar 5 '14 at 12:11
  • 1
    @PolyGeo Maybe not, check out this site: quantcast.com/gis.stackexchange.com -- according to that our traffic has basically gone up by 5x since 2011. – blah238 Mar 5 '14 at 17:47
  • 1
    There is another factor that we cannot take into account with these kinds of queries. Some unanswered questions are deleted after 1 year: meta.stackexchange.com/a/92006 – blah238 Mar 5 '14 at 20:49
  • 1
    There is a good analysis on Meta SO here: meta.stackexchange.com/a/143137 – blah238 Mar 5 '14 at 20:57
  • 1
    Thanks for providing these numbers - all in all it looks like GIS SE is healthy enough. – PolyGeo Mar 6 '14 at 2:45
6

I also think that the lack of specificity to GIS in the FAQ has been part of the reason for the lower answer rate. GIS questions tend to be highly technical in nature and very often question askers do not provide enough information or structure their question in such a way that it can be reasonably answered. A better FAQ could help.

Related: (How) can we improve our FAQ?

  • 1
    A nice notion, but remember that not many people necessarily read the FAQ. – GIS-Jonathan Nov 12 '12 at 12:24
  • 1
    @GIS-Jonathan I think that's part of the problem. I don't think we want people who haven't read the FAQ to run around loose ;-) – R.K. Dec 19 '12 at 13:06
  • @R.K. - I can't say I've read the FAQ..... ;-) – GIS-Jonathan Jan 4 '13 at 11:53
  • @R.K. - Huh, we have a FAQ? The problem is not that people are running around loose, not reading a FAQ, it is that the site has a massive weakness when it comes to existence and access to a FAQ. – Martin F Jun 11 '14 at 20:04
  • @martinf do you mean the Help Center? What was previously known as the FAQ is now the Help Center. If you mean frequently asked GIS questions, we have the beginnings of one here: Do we need a list of canonical questions? – blah238 Jun 11 '14 at 21:16
  • I meant a GIS FAQ (and didn't realize about FAQ being the old name for Help). You have a good list at "Do we need..." but i don't know how we'd use it. – Martin F Jun 11 '14 at 23:33
6

I think a lot of the problem is that so many of the unanswered questions are basically unanswerable. We get a lot of people who come to the site with a single specific problem who do not or can not provide enough detail to generate an answer and then leave the site when an answer is not forthcoming.

This will also discourage people from answering older questions, as the person they are trying to help appears to be long gone.

A moderator flag of Unanswerable could be useful as a way of identifying such questions, allowing a distinction to be made between these questions and good questions which have not yet been answered for other reasons.

  • 4
    I like your idea. But what would "unanswerable" do that closing does not accomplish? After all, if there are questions that do not have enough detail to allow answering, they ought to be closed under existing guidelines. – whuber Nov 7 '12 at 17:08
  • That is a good point. Maybe unanswerable is the wrong term, I was trying to convey a state somewhere between an active, valid question and a closed, poor question. So that users can focus their efforts on answering questions that will actually benefit other users. – sgrieve Nov 8 '12 at 8:31
  • 2
    Maybe we should try something like cleanup day? But instead of cleaning up the titles, we flag questions that need to be closed. – R.K. Nov 10 '12 at 14:39
  • That sounds like a great idea. – sgrieve Nov 10 '12 at 16:58

You must log in to answer this question.

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