I was wondering if there is some place that has voting statistics / general usage statistics on Serverfault.
Possibly in relation to its closest sister sites: StackOverflow and Super User.

I'm mostly curious about quantifying the "decline of Serverfault" I saw some posts about.
I cobbled together some statistics using the data.stackexchange (the queries need a lot of work) interface, but the results are so much in line with the results from the main StackOverflow website, it's hard to really get a pulse on it.

In the lengthy thread started by Shane Madden there was a lot of statistics / graphs about the decline of Serverfault. Mainly based on traffic vs posts. Related to the google results that lead to ServerFault.
I was trying to find out if I can also see some indications that Serverfault has become more negative in the last 6 months. But all I can see is a more negative voting trend across most of StackExchange.

Whether I look at StackOverflow / SuperUser or ServerFault... Downvotes are up, up votes are down.
And honestly, isn't that kind of normal? As more and more questions have a direct answer on the sites. More and more questions will get a down vote with a "Duplicate" stamp on them.
Newer questions about simpler stuff will get down voted or ignored by the more senior members, and will be a lot less likely to receive attention / foster newer members.

The attitude that you leave simple questions to newer members might not be the right one.
A person who asks a question here, and does not receive an answer is a lot less likely to become a community member. However most people who ask questions here could be valuable community members. Given some time to garner some more experience.

Maybe the right response is to have more open discussions on newer questions, that further push the newer members towards the answers instead of rehashing old answers. Coupled with having those questions closed after a period of time, some-how, so that the current repository of high quality questions and answers does not get diluted. Not immediately closing duplicate questions might help in this. Perhaps having them removed from the "active" board, but allowing further discussion.

Regardless, I would like more statistics. The problem is quite interesting and I've seen suggestions that indicate the correct solution would be voting.
Yet on the flip side we have people such as Michael Hamtpon who correctly believe we shouldn't vote on what we don't have a thorough understanding of (which makes loosening the scope seem like a bad idea).
So maybe voting isn't the answer, maybe limiting the scope is. As a larger percentage of the users understands the subject matter, they will be able to voice more targeted opinions.

I've also seem a few posts that compare ServerFault to StackOverflow, that seem to be more about the way a few users perceived the two, without really much data behind it.

So... in the spirit of minimal understanding, does anyone have more data?
Have similar inquiries been made before? (In which case my apologies, I am horrible at finding stuff on StackExchange at times)

Apologies if the post is somewhat all over the place, I sometimes have issues with sticking to one train of though.

2 Answers 2


I was wondering if there is some place that has voting statistics / general usage statistics on Serverfault.

It's not publicly accessible. Shane's screenshots were from the page that Moderators can see. Moderators are only allowed to disclose the general trends and such, not much of the specific data. Shane is a Stack Exchange employee (they get diamonds like the Moderators, confusing I know, but the issue has been brought up and apparently SE wants users to be confused) which is why he can post screen shots and actual data.

Downvotes are up, up votes are down. And honestly, isn't that kind of normal?

Yep, pretty normal. We have sharply less people providing answers however, sharply less answers, and a relatively steady rate of incoming questions. This means questions either go unanswered, or Experts providing answers burn-out quicker than they should (assuming you're in the "it's normal for people to burn-out" camp).

The attitude that you leave simple questions to newer members might not be the right one.

Great idea, now find the small army of veterans to answer every question and we'll be set. Sorry, I know that's snarky. People new to this problem all have the same simple answer. If the problem was that easy to fix, we would. Getting people to commit vast swaths of their personal time in exchange for fake Internet points and a sense of helping a community is insanely difficult (SE has done a fantastic job of this overall, but it's not working well here on SF).

Maybe the right response is to have more open discussions on newer questions

Perhaps, but SE has a strong anti-discussion slant. The SE Sites really only work well for the Q&A Model, not for discussions. To that end some of the original founds have now moved on to creating a great "discussion" website: http://discourse.org

Sorry I'm skipping the rest of your questions as their either redundant or more along the lines of observations and rhetoric that requires no direct response within the context of asking for data

  • Thank you for your answers. They help a lot with understanding the current community. Is the data on data.stackexchange in any way limited? As you say they wish to foster confusion yet release data dumps.
    – Reaces
    Commented Dec 12, 2014 at 16:55
  • 1
    Er, not "foster confusion" in general, just as to who is a Moderator and who is an Employee... It's a long story involving an employee and a moderator both acting like children, but it's also ancient history at this point too. The data in the dumps has quite a lot of information removed. Most is related to personal information privacy or voting privacy, but other information get caught too (they use a whitelist approach so data doesn't accidentally get let loose).
    – Chris S
    Commented Dec 12, 2014 at 19:30
  • My wording was perhaps poor... Implying some kind of malevolence. What you say makes sense. My apologies if my questions are basic for long-time community members, the information is hard to puzzle together from old posts and asking directly seemed like an easier approach.
    – Reaces
    Commented Dec 12, 2014 at 19:32
  • 5
    I can't imagine trying to really "catch up" with the story at this point. The slow and steady "decline" has been going on basically since the site started, and people have been talking about it for years. But it basically boils down to less people contributing (good) answers.
    – Chris S
    Commented Dec 12, 2014 at 19:36

Sure, the downvotes data is in the post you linked, but here's the data for both for the last year:


The upvotes data looks pretty healthy, but we can probably attribute some of that to the increase in traffic coming from Google.

And here's the percentage of questions that are closed or deleted:


Chris addressed most of your questions well, but just to add something on the topic of duplicates:

As more and more questions have a direct answer on the sites. More and more questions will get a down vote with a "Duplicate" stamp on them.

Closing as a duplicate is one of our least used close reasons. I don't think its use has had much of a negative impact on the amount of engagement or discussion taking place.

Partly, this is because our search system is currently pretty bad at actually finding duplicates, which is being worked on.

Theoretically, when a question is closed as a duplicate, it should be a very healthy close - it's saying to the asker, "This other question is the exact problem you're having, your solution is over there, enjoy!" - it's for the kind of question where an astute answerer would just copy-paste the answer from the existing question, so there's no reason to re-hash it.

In practice, a lot of our dupe closes are against our canonical questions, and not just random copies of the exact same question.

Running out of questions remaining to be asked is definitely something that could make it very hard to get involved in participating, but I think we're a long way from there - and our field is changing quickly enough that there should always be some fresh topics, even if we manage to reach the point that we've answered every possible Apache question under the sun.

The problems that we're having currently are a combination of poor quality questions and a decreasing number of people contributing answers. I'd love to be at the point where we're running out of problems to solve, and Stack Overflow might be there on some topics, but I think trying to solve that problem is just theoretical at this point.

  • Thank you for the explanation. Could you tell me how I can see these deletion stats on data.stackexchange.com? As I tried to query these statistics, but if I look for posts where ClosedDate is not null I seem to only end up with 13009 posts, which does not correlate with your graph.
    – Reaces
    Commented Dec 12, 2014 at 19:01
  • @Reaces I'd need to go to the developers to explain the exact mechanics of what's in SEDE, but I suspect that's because closed questions are eventually deleted and would then stop being exposed in the data dump. Commented Dec 12, 2014 at 19:03
  • @ShaneMadden I wouldn't expect such a JUMP around Sept and Dec. Doesn't seem like we have that many students here (which one might typically associate with jumps in low quality posts at the beginning and ends of semesters).
    – Chris S
    Commented Dec 12, 2014 at 19:37
  • 1
    @ChrisS Add a certain level of frustration after closing a lot of student questions. Also I imagine a lot of companies have freezes and reports near the end of the year that end up causing inexperienced personnel to do things they're less than comfortable with. Also a lot of personnel are on holidays, currently being substituted by people with less experience. The holiday periods have historically had a large impact on productivity in the western world.
    – Reaces
    Commented Dec 12, 2014 at 19:45
  • @ChrisS Yeah, I don't have any good explanation for that - The same jump exists in Oct - Dec of 2012, but in 2013 the Jul-Sep period was where there was a jump, then Oct through Dec was flat. Commented Dec 12, 2014 at 19:58
  • 3
    I suppose you could call it a state of Extreme September, a phenomenon that occurs when a calendar September falls within the Eternal September.
    – Andrew B
    Commented Dec 12, 2014 at 22:55

You must log in to answer this question.

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