Improving on 36,000 unanswered questions

Question

We currently have just over 36,000 unanswered questions.

After reviewing about 500 of them a while ago, in a totally unsystematic manner, I concluded that <2% are truly reasonable questions awaiting a reasonable response.

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.

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.

Do we have a higher percentage than other comparable exchanges?

Answer 1

At 77% answered we are better on that particular statistic than Stack Overflow (everyone's measuring stick) at 69% and Cross Validated (which launched at the same time as GIS SE) at 65% - see http://stackexchange.com/sites#percentanswered - but I am all for trying to get us up into the 80s (or even 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 four sites out of 181 receive more questions per day than we do AND have a higher percentage of questions answered than we do. The top six sites on the Stack Exchange network by this measure are:

• TeX-LaTeX 38 questions per day, 90% answered
• Home Improvement 32 questions per day, 86% answered
• Electrical Engineering 48 questions per day, 85% answered
• Physics 54 questions per day, 80% answered
• GIS 28 questions per day, 77% answered
• Mathematics 267 questions per day, 77% answered

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.

I found the query to do the above via asking a Meta SE question: https://meta.stackexchange.com/questions/270413/identifying-tags-with-poorest-answer-rates-on-a-site

Answer 2

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

• Too many questions left unanswered when answer inside comments? and
• Answering questions with a comment?

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.

Answer 3

There is a search strategy I sometimes use to try and find the Unanswered questions where I am most likely to be able to make a difference.

1. First I click the Unanswered option in the left sidebar:

2. Then I click My Tags

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

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 that 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 36,000+ Unanswered questions you may like to try the same or a similar search strategy.

Answer 4

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.

Answer 5

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.

Answer 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.

Answer 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).

Answer 8

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?

Answer 9

(Note that it is now more than 30k unanswered, 77% answered out of all questions)

I see at least these reasons:

1. Some questions are answerable but hard, and it may take years for someone with the right knowledge to answer

2. It is not as lucrative (for the typical SO user who might be able to answer those that are programmatic in nature) in terms of effort-to-reputation ratio. If they have time to answer a GIS.se question, that time may be better spent on SO or picking up new tech or just working (for my work, GIS is an important but only a very small aspect, making money is far more impt).

3. GIS support channels are fragmented across SE, Reddit, ESRI, mailing lists, github issues etc. Some users got put off by SE culture so they are driven towards other channels after negative experience. They may still come back through search engine but active participation is not their goal, they only want the information they came for and move on.