# Should answers be ranked by reputation? [duplicate]

Possible Duplicate:
Feature Request: Ability to sort answers by responder's reputation.

I have notice that sometime detailed answers by experience mathematicians are not getting the number of up-votes that it deserves. Take this question for example, prove that $(1 + x)^\frac{1}{b}$ is a formal power series . Bill answered last and was at 0 for a while until I chose it as the best answer. So should there be a feature request to rank the answers based on reputation points? (also notice that second highest answer was the first answer).

-

## marked as duplicate by Mark Trapp, randomMay 19 '11 at 22:08

Yeah. I never get the up-votes that I deserve. –  Cody Gray May 20 '11 at 0:18

For lots of reasons, the answerer's site-wide reputation should not determine the order of display in the answers. For one, just having a lot of reputation does not mean you have the best answer (having the accepted answer is the best way to measure this, followed by the most upvotes). You may have gained all your rep from entirely different tags, or from asking questions and not answering, or from bounties.

On the flip side, not only is it bad to assume that high site rep correlates to high quality answers but it's bad to assume the opposite - that low-rep users do not have high quality answers. Everyone starts out with 1 rep at some point and deserve an even playing field to earn more reputation to ascend the ranks. Harming them by listing their answers lower in the answer display listing just furthers the divide between the haves and have-nots based on the faulty assumption that "low rep = bad answers"

-
I can understand your point, but how would you propose to solve this phenomenon that fastest answers are (sometimes) getting the most up votes? Should we just look past this flaw? (note: that I am raising a problem and is just proposing a possible solution) –  Mark May 19 '11 at 18:55
@Mark: Fastest answers getting the most upvotes is (to a certain extent) a feature, not a flaw. In the old days, when I had to rely on forum-style venues, getting answers to questions could sometimes take several days. On StackOverflow, I am surprised if I don't get two or three good answers to my questions in less than 30 minutes. –  Robert Harvey May 19 '11 at 19:20
@Robert Harvey Perhaps so on question that ask for clarification of syntax but usually not so for mathematical questions (the last and most detailed answer is most likely the best one). –  Mark May 19 '11 at 19:30
@Mark: Any questions, not just the simple ones. For Math.SE the velocity might be different, but the principle is the same. And the amount of detail in the answer doesn't necessarily make it the best answer for the OP. Sometimes, you just need to be pointed in the right direction. –  Robert Harvey May 19 '11 at 19:32
@Robert Harvey what makes you think that? I can understand a syntax problem would be best solved asap however, usually the fastest answer for a mathematical question is very likely to be sloppy (and lead to further misunderstandings i.e. the op is worst off if s/he had not ask it). There are many examples of this. –  Mark May 19 '11 at 19:35
@Mark: Then you need to make that case to the user community at Meta.Math.SE. But ultimately it is up to the users to decide how they want to use their votes. –  Robert Harvey May 19 '11 at 19:38
@Mark: The most worthwhile and lasting questions on SO are not "syntax problems" (the fact that the majority are is a separate problem); they too require reasonably detailed explanation and potentially suffer from difficulties with FGITW. –  Josh Caswell May 19 '11 at 20:07
@Rob On Math.SE, often the fastest answers are the shallowest. It's especially pronounced in a general math forum because the subject matter is quite deep. As such, some questions take days if not weeks to provide excellent answers. Often the best answers are the lowest-voted answers only because they came after the question was off the front page (so not as many people see it, either because they only read the newest question list, or they don't revisit threads they've already seen). The software needs better ways to expose later answers. Fastest is not best in mathematics. Not even close. –  Bill Dubuque May 19 '11 at 20:46
@Bill: What about using bounties to incent those excellent questions and answers? Bounties are well-suited for this sort of thing: long, well-thought-out, contemplative answers. –  Robert Harvey May 19 '11 at 20:51
@Rob The primary problem is getting slower excellent answers exposed, so that they can get deserved votes. I don't see how bounties would help. –  Bill Dubuque May 19 '11 at 20:56
@Bill: Bounties help because they increase exposure of the question. Changing the sort order of answers won't solve the problem because people vote based on what they know. The better an answer is, the less likely it is to get upvotes because it takes longer to write a good answer, and because the more detailed and technical an answer is, the less likely a potential voter will take the time to try and understand it so they can cast an upvote. –  Robert Harvey May 19 '11 at 21:00
@Bill: It's not about the reputation of the answerer, it's about time. If you really want to try and solve the problem with a sort order, you'd have a better result with a newest answer on top sort. –  Robert Harvey May 19 '11 at 21:02
@Rob I never claimed that sort order was a solution. For my take see my answer to the same question on Math.SE. Bounties are for questions, not answers. –  Bill Dubuque May 19 '11 at 21:10
@Rob fyi: Perhaps you mistakenly addressed your latest comment above? It does not to apply to anything I've said. –  Bill Dubuque May 19 '11 at 21:14
@Bill: You're right, it's a shame that the excellent answers on Math.SE don't get their deserved upvotes. But I don't really see how that problem can be solved - Robert explains why "The better an answer is, the less likely it is to get upvotes", and I think he's got a very strong point. Do you have a good idea what could be done? (I read your answer on Meta.Math.SE, and your suggestions on user behaviour are great, but how could one improve the system?) –  Hendrik Vogt May 20 '11 at 7:45

I think that it would be simple to add a reputation tab as an answer sort option. Options are good.

-
possibly the most diplomatic answer. although i agree with @Robert Harvey though, reputation doesn't necessarily correlate with good answers. –  DForck42 May 19 '11 at 20:06
@DForck42, yeh, I probably wouldn't use it, I like the 'votes' and 'oldest' orders. –  Lance Roberts May 19 '11 at 20:07
oops, put wrong person, meant @Daniel DiPaolo –  DForck42 May 19 '11 at 20:16

What about using total rep in an aggregate of the tags as a tie breaker? Or at least add is a weighting when determining the random order of tied answers.

-
While considering reputation in the relevant tags is at least better than site-wide reputation, it's still far too broad to make it immune from the same problems that the site-wide rep sorting falls prey to –  Daniel DiPaolo May 19 '11 at 22:10