13

In an effort to improve myself in LaTeX and communication thereof, one metric I have used to assess the quality of my answers on this site is to take my total reputation and divide it by the total number of answers I've given (this is because I don't know how to reckon the total number of upvotes I've received). It helps measure how well I can convey useful information to others, which is not quite the same as measuring how proficient I am (or not) at LaTeX.

Sadly, in my case, this ratio is quite low compared to the major contributors. In some cases, it may be because I try to answer very particular questions that may garner little general interest, outside of the OP. But I also am cognizant of the lesser quality I often provide relative to others.

But, regardless of reason, this ratio gives me a metric I can use to measure and work on my improvement over time. But I found one glitch: on days where you max out at 200 points, there will be upvotes not reflected in the reputation number.

So my question is this: is there a convenient way on the site to see a tally of total upvotes for a user, so as to calculate a more relevant metric (upvotes per answer) rather than reputation per answer. And, if not, I guess that would be a feature request.

13

You could parse the output obtained by visiting the text-only reputation breakdown. The output resembles:

total votes: 18989
 16    22477 (2)
-- 2011-07-06 rep +2    = 3         
 16    22829 (2)
 2     22831 (10)
 2     22915 (10)
-- 2011-07-12 rep +22   = 25        
 2     23026 (5)
 2     23085 (10)
 2     23096 (10)
 2     23096 (10)
 2     23096 (10)
-- 2011-07-14 rep +45   = 70        
 2     23096 (10)
 2     23096 (10)
 2     23026 (5)
 2     23096 (10)
 2     23096 (10)
 2     23082 (10)
-- 2011-07-15 rep +55   = 125       
 16    23161 (2)
 1     23032 (2)
 2     23162 (10)
 2     23162 (10)
 2     23237 (10)
 2     23237 (10)
 2     23237 (10)
-- 2011-07-16 rep +54   = 179       
 2     23237 (10)
 2     23237 (10)
 2     23237 (10)
 1     23237 (15)
 2     23162 (10)
 1     23162 (15)
 16    23332 (2)
-- 2011-07-17 rep +72   = 251       
 2     23162 (10)
 1     23406 (15)
 2     23406 (10)
 2     23420 (10)
...

which gives a daily tally of votes cast toward you (up or down, for a question or an answer). The first column contains the type of vote you received: 1 = accept; 2 = Upvote; 3 = Downvote, ..., 16 = Approved suggested edit. The middle column represents the post ID and the last (bracketed) column represents the reputation gained. Days where you reach reputation cap have some of these entries listed as (0).

This approach will give you an up-to-date reflection of your current voting status. If you're not really concerned with an in-time calculation, the Stack Exchange Data Explorer allows an interface to creating your own queries to extract the data, summarize and display them. However, this has a lag (which could be months). It's also possible to download the entire data set via torrent and play with it, if you can't go to sleep easily...

For a list of vote identifiers, see the structure of the voting database in Database schema documentation for the public data dump and Data Explorer.

  • Very informative. The computer knows all...it's just a question of cajoling the information from it. You do it so well. – Steven B. Segletes Aug 30 '13 at 19:15
  • @StevenB.Segletes: This is fairly easy to do in something like Excel. – Werner Aug 30 '13 at 20:47
  • I already wrote the FORTRAN code 8^) to process it. (no jokes, please). – Steven B. Segletes Aug 30 '13 at 21:01
  • @StevenB.Segletes I'm biting my fingers here, not even a subtle one? :P – percusse Aug 30 '13 at 21:03
  • @percusse You have to understand, in my world I am modern in a nostalgic kind of way. I work a hundred yards from where the ENIAC was housed and worked with several (now retired) who programmed the ENIAC. Von Neumann was a frequent visitor to our lab, along with his buddy von Karman. Of course, that was before I was born, which is why I'm so modern. After all, it's not like I use FORTRAN 66... I'll stick with the '77 release. – Steven B. Segletes Aug 31 '13 at 2:58
  • @StevenB.Segletes Hahah, no I was just teasing. I'm an engineer so you don't need to justify Fortran to me. My whole life is around LAPACK libraries – percusse Aug 31 '13 at 8:48
  • @percusse A kindred spirit, I see. – Steven B. Segletes Aug 31 '13 at 12:31
17

The reputation system is a bit of fun but I think it's dangerous to think there is any correlation between quality of TeX information and points earned. Looking at my top three answers ordered by rep

234 How can I explain the meaning of LaTeX to my grandma?

75 What are penalties and which ones are defined?

68 How was TeX output visualised on screen, back in the day?

Only the second one has any real TeX information and the first one scored about as many as the next four put together, and is essentially content free.

  • 9
    Of course it's also dangerous to draw conclusions about lack of correlation by looking at outliers. ;-) – Alan Munn Aug 31 '13 at 17:19
  • @AlanMunn quite true:-) – David Carlisle Aug 31 '13 at 17:28
  • Thank you, David. Your answer is, as always, excellent and should certainly make me think about my metric. However, I'm accepting Werner's answer, because he more precisely answered the technical part of my question. – Steven B. Segletes Sep 3 '13 at 15:15
  • 1
    Sometimes the best answers come in very compact sizes. – LordStryker Oct 24 '13 at 16:39

You must log in to answer this question.

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