I understand the reputation system, it works, among other things, as a reward to those who contribute.

I also tend to think of reputation also as a measure of value created by M.SE; when I gain a reputation, it is because I provided a 'service' to someone who values that service. With this, the sum of the reputation created on M.SE measures (approximates) the value created by M.SE.

Is there a statistics about the the sum of the reputation? How it varies with time? What are the peak hours/days/month? Is there a statistics across different SE sites?

  • $\begingroup$ None that I'm aware of but I've only been here about a year and haven't thought about such things that much. But as someone who uses SQL daily I'm sure there's a way to get that info using the data explorer. $\endgroup$ – user307169 Jan 31 '17 at 20:31
  • 5
    $\begingroup$ 25k+ users and mods have data on the number of votes which will by and large allow to derive the reputation generated. (Deletions could destroy reputation, but it would not destroy the service to somebody that upvoted earlier. There is also CW.) There are something like 2.5k upvotes a day, and 300 accepts and maybe 500 downvotes (not sure how you want to count those). So gain about 30k points a day, like a tad less. This is rather stable since a while. There is a noticeable weekly rhythm. And the academic year(s) and holiday(s) especially the US one seem to play a role. $\endgroup$ – quid Jan 31 '17 at 22:49
  • $\begingroup$ @quid: All these votes are on answers? $\endgroup$ – Asaf Karagila Feb 1 '17 at 6:30
  • $\begingroup$ @Asaf ugh, what an oversight. The "recent" change that question votes only give +5 did not yet really sink in. $\endgroup$ – quid Feb 1 '17 at 9:07
  • $\begingroup$ @quid Thank you, that satisfies my curiosity for a while. $\endgroup$ – Jan Feb 1 '17 at 10:58
  • $\begingroup$ Reminds me of a book I read a very, very long time ago: How to lie with statistics. Quantitative values is a proper subset of the realm of value. And just because a quantitative answer can be assigned to x, says little in the realm of value. $\endgroup$ – amWhy Feb 6 '17 at 18:13
  • $\begingroup$ Anyway, why the fascinations with statistics that measure only things that are quantitatively measurable ? $\endgroup$ – amWhy Feb 6 '17 at 18:17

Actually, all of such data is available on the http://data.stackexchange.com . It requires a little bit of ms-sql skills to query it.

Generally non-available data (for example, the voter accounts) isn't available even there, but the votes yes.

I've made this query for you:

SELECT COUNT(*) FROM Votes WHERE VoteTypeId = 2;

The number of the upvotes only on answers is:

FROM Votes LEFT JOIN Posts ON Votes.PostId = Posts.Id
WHERE PostTypeId = 2 AND VoteTypeId = 2;

Currently there are 4091275 upvotes on the MathSE existing, 2438174 is for answers from them. The total sum of the reputation of the MathSE users (query: SELECT SUM(Reputation) FROM Users;) is 42113423.

The difference is due there are various other cases which modify reputation (downvote, bounty, synergy bonus, initial 1 rep for new users, question upvotes get only 5 repu, etc).

  • 1
    $\begingroup$ @arjafi Well it seems the SEDE could have a little bit of improvement for the task. In my opinion, querys should be named, linked, deleted like the posts on the other SE sites. I replaced it with the query text, rather to find a way to publish the query. $\endgroup$ – peterh Feb 1 '17 at 12:09
  • $\begingroup$ It's not entirely clear (for me) what the comment referred to: The queries can be stored, named and linked, as in data.stackexchange.com/math/queries (but this is somewhat dynamic, and repeating the query code here should be fine) $\endgroup$ – Marco13 Feb 1 '17 at 12:47
  • 1
    $\begingroup$ I'm surprised to learn that I have accumulated 0.03% of all reputation ever awarded on this site. I was expecting that number to have more 0's in it. $\endgroup$ – Stella Biderman Feb 1 '17 at 21:12
  • $\begingroup$ So the top two users each have a little bit over 1% of the total reputation. $\endgroup$ – Gerry Myerson Feb 1 '17 at 22:00
  • $\begingroup$ @Marco13 You are right, unfortunately his comment is now deleted and I can't remember already the context. :-( $\endgroup$ – peterh Dec 22 '17 at 20:07

A low-tech answer comes from this table from the "user reputation side" http://stackexchange.com/leagues/21/alltime/math/2010-07-20/ from today:

enter image description here

Since the right column is "cumulative", one can calculate the number of users in the respective reputations intervals, take the lower end of the interval as lower bound and sum up the products "#users*minimal rep" and obtains the following lower bound


You also get lower bounds on the total reputation by "user groups":

 Rep. interval    Total rep.
 >100,000         2,800,000
 50,000 - 99,000  2,800,000
 25,000 - 49,000  3,125,000
 10,000 - 24,999  3,630,000
  5,000 - 9,999   2,220,000
  3,000 - 4,999   1,833,000
  2,000 - 2,999   1,358,000
  1,000 - 1,999   1,779,000
    500 - 999     1,637,000
    200 - 499     1,507,800
      1 - 199       291,167

You must log in to answer this question.

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