23
$\begingroup$

[2016]: I have the subjective impression that the rate at which interesting questions and answers are being posted on cstheory.stackexchange.com has decreased substantially over the last year or so. Does anybody else have this impression (maybe answer that in a comment)? Or even better, does anyone have access to cstheory.stackexchange.com metadata to check this trend objectively?

[August 2018]: Moved the data that had accumulated here down into a community-wiki answer, and posted updated data suggesting that, as of 2018, the downward trend seems to be continuing (see chart at end of post).

$\endgroup$
14
  • $\begingroup$ Most of metadata is online and can be queried: Data Explorer $\endgroup$
    – Kaveh
    May 4, 2016 at 2:27
  • 3
    $\begingroup$ A comparison with Math Overflow and Computer Science might be useful. There is also the issue of new more specialized sites like Cryptography. $\endgroup$
    – Kaveh
    May 4, 2016 at 5:32
  • 11
    $\begingroup$ I know several grad students who like cstheory but feel uncomfortable to ask questions,, they feel their questions are not good enough. I have tried to convince them to give it a try but haven't been very successful. They seem afraid to ask questions that might look silly or trivial and it is very difficult to convince them it is OK, you don't need to have an extraordinary good question to ask. We may need to do something more proactive about it. $\endgroup$
    – Kaveh
    May 4, 2016 at 5:36
  • 2
    $\begingroup$ Unfortunately it doesn't have data for earlier years. There seems to be a similar decline on MathOverflow as well. It would be interesting to see if there is a decline also in the amount of visits, I will write a query for that next week if someone doesn't do it by then. $\endgroup$
    – Kaveh
    May 5, 2016 at 1:51
  • 2
    $\begingroup$ It's too bad. I used to stop by the site every few months to look for interesting questions and answers and would usually find a few new ones. Now, not so much. $\endgroup$
    – Neal Young
    May 5, 2016 at 3:19
  • 2
    $\begingroup$ What happened in November 2014? $\endgroup$
    – Bruno
    May 15, 2016 at 10:01
  • 6
    $\begingroup$ I do think that "perceived reputational risk adversion" may be at work. Perhaps it would be good to encourage members to be more accepting of imperfection. E.g. rather than "this is a bad question" ... comment "did you mean to ask..." when its possible to discern. Also I notice that perfectly good answers are often put into comments, perhaps because author doesn't believe what they wrote is canonical. $\endgroup$
    – shaunc
    May 16, 2016 at 16:01
  • $\begingroup$ meta.mathoverflow.net/q/2876 $\endgroup$
    – Kaveh
    May 24, 2016 at 21:48
  • $\begingroup$ An interesting point: the older posts are more likely to have higher total vote count, one has to take that into account when doing the comparison. We have one of the highest number of votes per post, that might be contributing to the fact that we have a steeper slope. It would be interesting to count only votes on posts for a fixed time period after they are posted, e.g. 30 days. $\endgroup$
    – Kaveh
    May 24, 2016 at 21:50
  • $\begingroup$ excellent quantitative analysis, thx! think SE should provide these basic engagement stats across sites but they are not much available. mods can see site hits which is generally more what SE mgt cares about. there are some rumors of SE-wide slow declines in voting etc.; this may even be a general property of social media sites in general eg wikipedia, and facebook has also noted "status update" decline. however, have long advocated spicing up this site and broadening the charter in various ways to little avail, site mentality is conservative. suggest/ encourage further discussion in Theoretical Computer Science Chat $\endgroup$
    – vzn
    May 26, 2016 at 1:14
  • $\begingroup$ I also got the data for older posts: data.stackexchange.com/cstheory/query/490846/… $\endgroup$
    – Kaveh
    May 26, 2016 at 4:01
  • $\begingroup$ How about we make the EDIT 1-5 parts into an answer? $\endgroup$
    – Bjørn Kjos-Hanssen Mod
    Aug 23, 2018 at 4:56
  • $\begingroup$ @BjørnKjos-Hanssen: Ok, good idea. $\endgroup$
    – Neal Young
    Aug 23, 2018 at 12:10
  • 1
    $\begingroup$ @Kaveh Do you still experience people being reluctant to post on Theoretical Computer Science? Any luck pointing them to Computer Science? $\endgroup$
    – Raphael
    Sep 26, 2018 at 7:56

1 Answer 1

Reset to default
10
$\begingroup$

Both interesting and overall activity on cstheory are decreasing!

Let's measure "interesting" activity by the rate of posts that receive significant upvotes. By this measure, Figure 1 below suggests that both interesting activity and overall activity are trending downward:

enter image description here

Figure 1. Above are plots (over time, in 60-day blocks), of (a) number of questions, (b) number of answers, (c) # upvotes / 20, (d) number of questions and answers that received at least five (or at least ten, for (e)) upvotes within 30 days of creation, on cstheory.stackexchange.com.

All trend downward. (The query that generated the data is here.)

The same plots on other stackexchange sites, for comparison

cs.stackexchange.com

enter image description here

crypto.stackexchange.com

enter image description here

math.stackexchange.com

enter image description here

tex.stackexchange.com

enter image description here

$\endgroup$

You must log in to answer this question.

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