# An indicator of how good a user's answers generally are

Users A and B both have 2000 reputation. However, user A has attempted in answering 10,000 questions and only 1000 of his answers have been upvoted. User B, on the other hand, only has 400 answers, but all his geese are swans, and he has even grabbed some bounties on the way.

If they both answer your question, who would you take as a more authoritative source?

I would choose B. But what indicator can sum up this difference between them?

I've been thinking about average and median, but both have their own disadvantages. If we take the average rating of a user's answers, it can skyrocket because of one answer with 200 upvotes, even if most other answers suck. The median will give integer numbers, and I guess for most users the median would be either 0, 1 or 2, so there's not much information in this rating.

What do you think? Or is there already such an indicator and I missed it?

-
Too many variables for this to be useful I think. How popular are the tags that the users have answered questions in, how fast are they to write their answers, how many answers have they given in the tags relevant to your question, how many of their questions have been linked to from Digg or some other popular site etc. – ho1 Mar 13 '11 at 10:20
I think the question is interesting but the context is terrible : "If they both answer your question, who would you take as a more authoritative source?". You can't determine the quality of an answer through the user rep. Even a new user with no rep can be far more qualified to give a good answer if it's his domain. I think every answer should be tried. Then if you notice difference between two valid answers, try to find out which one is the best. Hope you will consider that ;) – alain.janinm Feb 20 '12 at 19:23

I would choose the one that works, regardless of rep, or rep per post.

But I understand where you are going — I might want to try the answer the from B before I try the answer from A. However, even in this case the way it's supposed to work is that others who have some knowledge in the problem domain will come in and likely vote B higher than A, at least if B's answer really is better. So just try the one with higher votes.

The only time this is ever a factor is when both answers are also voted the same score (or not voted yet at all), and that's relatively rare.

-
Of course the answer itself is what counts, but then, why do we have reputations scores at all? – Ilya Kogan Mar 13 '11 at 4:33
@IIya - rep is about 3 things, in order: Trust -- how much the system thinks it can trust you with certain privileges like editing or voting. Ego -- a fun score that gives users a reason to participate, as well as promote themselves. Competency -- I say "competency" rather than "skill" because as you noted it's possible (easy?) to offset low skill with high effort, but high effort won't help you without some basic level of competency; it's very hard to earn a high rep score without it. However, this is useful more for areas like the careers site than judging answers. – Joel Coehoorn Mar 13 '11 at 4:41
Trust: That's the point, isn't it? I'd trust user B, but not user A. – Hendrik Vogt Mar 13 '11 at 9:23