19
$\begingroup$

I notice that some users are extremely active while many only post once. I wonder how much activity is done by the top users. How much of the answering activity was done last quarter by those 100 earning the most reputation?

$\endgroup$
  • 5
    $\begingroup$ Maybe the 20-80 law applies (20% of the users give 80% of the answers). en.wikipedia.org/wiki/Pareto_principle $\endgroup$ – Gerry Myerson Jan 11 '14 at 4:04
  • 8
    $\begingroup$ (I'm hoping for Post No Bulls coming to this thread and showing some magic — as it usually happens with any statistics question on Meta.MSE nowadays :-) $\endgroup$ – Grigory M Jan 11 '14 at 21:10
  • $\begingroup$ As a top poster on Money.SE, I've observed the ratio is far more skewed than 80/20. In our case the top 100 users are probably at 95% or more of the answers. $\endgroup$ – JoeTaxpayer Jan 12 '14 at 0:07
  • 1
    $\begingroup$ Currently on Data Explorer trying to get the query right. I'll post it with the results if I do. $\endgroup$ – Dennis Meng Jan 12 '14 at 2:27
  • 2
    $\begingroup$ It would also be interesting (though presumably harder to learn the answer to) what percentage of the total reputation acquired in some time period is acquired by the top $100$ users. $\endgroup$ – Pete L. Clark Jan 12 '14 at 18:07
  • $\begingroup$ @PeteL.Clark Well, depends on exactly how you're defining "top 100 users". As pointed out by Asaf, my answer gives the percentage for the top 100 users overall. You could probably drudge up a query to do "top 100 users by quarterly reputation gain", but offhand that particular query is actually harder. $\endgroup$ – Dennis Meng Jan 15 '14 at 8:44
  • 1
    $\begingroup$ @GrigoryM I always thought Post No Bills was Bill Dubuque posting with an appropriate pseudonym during his suspension. $\endgroup$ – Brian Rushton Jan 19 '14 at 21:57
21
$\begingroup$

Alrighty, looks like I've got the query down. I don't remember exactly where the quarter cutoffs are, but you can modify it easy enough.

This puppy just takes all answers from September 1 and on.

SELECT count(p.Id) FROM Posts p 
  WHERE p.creationDate > '2013-09-01 00:00:00'
  AND   p.PostTypeId = 2
  AND   p.OwnerUserId IN (SELECT Top 100 Id FROM Users ORDER BY Reputation DESC)

Running it (query link here) gets me

  1. 22924 answers from the top 100
  2. 73989 overall (just comment out the last line)

which gets me a percentage of ~30.98%


As pointed out by Asaf, this does "top 100 users by overall reputation". I might be able to post another query later that does "top 100 users by quarterly reputation gain", but skimming the options in Data Explorer, that won't be as simple of a query.

$\endgroup$
  • 2
    $\begingroup$ I really appreciate you posting the code; I'm interested in tinkering with the parameters. Thank you! $\endgroup$ – Brian Rushton Jan 12 '14 at 2:46
  • 3
    $\begingroup$ @BrianRushton Not a problem. It's actually a little addicting once you grasp enough of it to play around. $\endgroup$ – Dennis Meng Jan 12 '14 at 2:48
  • 6
    $\begingroup$ @Brian: This is not quite what you asked for. This is the top 100 reputation ever. It is certainly possible, if not true, that the top earning users are not the top overall 100. The deviation, however, shouldn't be large. $\endgroup$ – Asaf Karagila Jan 12 '14 at 7:27
  • $\begingroup$ True, I'm not fully sure how to adapt this to get the people with the most rep from the last quarter alone, at least without some ugly workarounds. $\endgroup$ – Dennis Meng Jan 12 '14 at 7:31
  • $\begingroup$ Surprising lower than expected. Money.SE hasn't graduated, yet, and has less activity. That may skew things. $\endgroup$ – JoeTaxpayer Jan 12 '14 at 18:38
  • 4
    $\begingroup$ @JoeTaxpayer Mathematics has roughly ten times as many users as Personal Finance & Money. There will be more inactive users here, but the top 100 on money make up a much larger percentage of active users than on mathematics. If you adjust for that, it may come out about equal. $\endgroup$ – Daniel Fischer Jan 12 '14 at 19:34

You must log in to answer this question.

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