6

When looking at the "Top Users" page of any particular tag, (http://{site}/tags/{tag}/topusers) both the "Top Answerers" and "Top Askers" sections show the total score and the count of posts that use the tag, for each user. These numbers are sorted by total score, and there is no mechanism for sorting by count.

IMHO - this isn't very useful, nor is it fair. The score is based on up/down votes, which is impacted by the age and popularity of the question or answer. It doesn't help determine who is most active on the tag.

For answers, sorting solely by score artificially increases the rank of answerers that have made only minor contributions. As an example, here are some of the top answerers of the datetime tag on StackOverflow:

Top Answerers

Daveb is in second place with only one question answered. That answer is correct, but it is also fairly trivial. It doesn't demonstrate expertise in the topic of the tag, it just describes something that's easily found in reference documentation. Does daveb really deserve to be ranked higher than Martijn Peters, Oded, or myself?

The problem is compounded by the fact that the top 5 all-time answerers are shown in the sidebar of the tag wiki:

Tag Wiki Sidebar

So when someone is reading the tag wiki, they may be led to believe that these are the proven subject-matter experts.

Similarly, the top askers of the same tag only have asked one or two questions. How is that useful to anyone? If I'm looking for people that are interested in a tag, I'd rather find an asker of 50 questions that have only earned a few upvotes each than to find someone with one good (or lucky) question.

I've noticed this on many tags across the StackExchange network, so I don't think it's limited to StackOverflow.

I propose one of the following:

  • Rank by count instead of score.

  • Rank by some formula where the score and the count are both considered.

  • Allow tag badges to influence the rank. (Or at least, show which users have earned the tag badge for this tag.)

  • Provide a link for the viewer to re-sort the list by score or count as desired. Though this would also need to carry over to the Top Answerers sidebar that appears on the tag wiki page.

Am I correct in this assessment? Or is there some other reason why the score is the only factor taken into account?

  • The top 20 for the past 30 days shows score only for answers posted in the past 30 days. Old answers don't count for that ranking. – Martijn Pieters Jul 10 '14 at 21:22
  • Both the "All Time" and the "Last 30 Days" lists sort the same way and have the same issue. And I don't see how the separation helps anything. What I'm saying is that IMHO, "top users" should include the users with the most activity, not just the users with the most upvoted posts over that period. – Matt Johnson-Pint Jul 10 '14 at 21:26
  • There are probably users with lots of activity that don't make the cut despite being productive and active. Because they are usurped by people with one or two posts, they don't get any recognition here. – Matt Johnson-Pint Jul 10 '14 at 21:28
  • Just for the record, since I am in that list: I couldn't care less about daveb being in that listing. They got in a popular answer, good on them! – Martijn Pieters Jul 10 '14 at 21:30
  • 3
    Wasn't trying to pick on daveb, it's just an example. The point is that if I want to know who the experts are on a tag, I look at the top answers of all time - both here and in the tag wiki sidebar. But that list isn't accurate. – Matt Johnson-Pint Jul 10 '14 at 21:33
  • And the 30-day top answerers clearly show Jon, you, then me listed as the top 3. Really, I don't see what the big deal is, myself. – Martijn Pieters Jul 10 '14 at 21:34
  • But it is a top answer of all time. I have similar answers, freak posts that have gone viral. They happen, and you are lucky if you have one. I like playing around with the stats from time to time, any regular in a tag knows who the experts are in what ways. That's fine too. – Martijn Pieters Jul 10 '14 at 21:36
  • Perhaps I should have gave an example where I wasn't in the list. I actually use the top users pages in many other tags all the time. I don't really care that my rank is high, I just want to know that when I look for users who are the experts on a tag that I can find them. – Matt Johnson-Pint Jul 10 '14 at 21:38
  • As for reasons why: the top scores are only calculated once a day; the process is intensive and for a while, failed more days than not. Providing sortable tables is probably too much for the amount of data in the system now. – Martijn Pieters Jul 10 '14 at 22:29
  • Ranking by count would just introduce a similar problem: the people with the most questions or answers may not necessarily be experts. Perhaps someone is just asking 1000 DateTime questions or answering simple questions that can be found in the documentation. At least score has some measure of quality. – Troyen Jul 10 '14 at 23:30
  • @Troyen - I think those cases would be handled by closing questions or flagging non-answers. Or perhaps the count should exclude zero or negative vote answers. – Matt Johnson-Pint Jul 11 '14 at 2:43
  • @MattJohnson counting the score means exactly that you ignore zero-score answers. For instance, users who go through the backlog of unanswered questions tend to get zero upvotes and hence will never enter the list of top users even if they were the main "maintainers" of the tag. – Nemo Nov 29 '16 at 7:33

You must log in to answer this question.

Browse other questions tagged .