"Of course, it would be a little tricky to set up a meaningful experiment." Easy enough, actually. You just need some high profile user to agree to, after writing a response, flip a coin to decide whether to submit that response while logged in or logged out. Sure, you can't get the same answer submitted twice, but you can get a random sample.

I guess I'd be willing to do this, although I'd feel pretty silly about it.

Just two cents: There will always be competition in some form. If you suppress it, it may manifest itself in ugly and unhealthy ways.

Just two cents: There will always be competition in some form. If you suppress it, it may manifest itself in ugly and unhealthy ways.

I'm mostly staying out of this discussion, but I agree with this. Obviously there's such a thing as too much competitiveness, but I don't see evidence of it on MO. People actually are competitive to some degree, and I think there's something to the argument that channeling that competitiveness into mostly meaningless reputation points frees people up to actually be more pleasant to each other.

Qiaochu, can you expand on what you mean by "good user interface design"? And how would those design principles imply that one should display the user's reputation but not the other information available from their user page --- e.g. age, location, and how long they've been a member for?

My understanding is that one of the main "points" of reputation is to give a very rough measure of how much you should trust answers from a person. If that's the case, it's information which is relevant to determining whether you should, say, put the effort into trying to understand an answer better. If I see an answer by Terence Tao use terminology I'm unfamiliar with, I'm more likely to try to learn about it and see if it's relevant to my question. If I see the same kind of answer from a user with reputation 1, I'm less inclined to believe that the unfamiliar terminology he uses is relevant to my question. Other information on the user page doesn't affect this decision as directly.

Given that reputation is important information for the basic function of the SE software - asking specific questions and getting specific answers - it's information that should be readily available when you look at an answer. Of course, the argument can be made that these kinds of concerns apply more to SO, for example, than to MO.

It's becoming apparent that there are certain types of people who handle MO better than others. Looking at the responses here, some are naturally competitive and enjoy this aspect of MO, some are oblivious to the competition and enjoy MO for the math. I suspect everyone mixes a little of both but those are the major trends. It seems that MO caters best to those who enjoy at least some competition and are comfortable enough to ignore any unpleasant side effects. Maybe that's the bottom line: some people are MO-types and others aren't?

Mark, my last statement was intended to mean exactly what you ask. I think that the mere mention of Romain Gary accidentally released some of my awkward French rhetorical style.

I'd like to respond to Emerton's comment just above.

I think that I am speaking not just for myself when I say that I don't use MO reputation at all as a tool for judging the correctness of an answer.

I absolutely agree with this, and I hope that everyone does as well. The key word here is "correctness". Reputation, good or bad, should have no relevance what-so-ever on whether or not an answer is correct.

I am much more likely to use the actual (i.e. professional, not MO) reputation of someone

Here's where I find that I do the complete opposite. I have never (to my, admittedly poor, memory) looked up someone from MO on MathSciNet, nor even on the arXiv (which might be a bit fairer). On occasion, I've done an internet search but generally only when I've already formed an opinion on that person (good or bad) and am looking for some background to see whether or not I'm jumping to conclusions. In fact, I suspect that I have a bit of reverse snobbery. If I see someone here who has a high professional reputation but a low MO reputation (proportional to how long they've been active) then I'm more likely to ignore what they say - I get highly irritated by the sort of people who sit listening, slightly smugly, to a discussion and interject "wise" comments now and then "for the education of the others"; if they care about the discussion, they should get involved, if not then they should go away. (To be clear: I'm not accusing anyone of this attitude, just that that's the image that gets triggered in my mind when I see a mismatch).

(As an irrelevant side issue, why MathSciNet? How does that help you? If that's how you judge me, then I'm sunk! As is every other young mathematician.)

I don't see why [MO reputation] should play much of a role in evaluating the actual mathematics.

Again, I completely agree. But I don't see why IRL reputation should play much of a role in evaluating it either!

I use MO reputation to determine whether or not to invest time in reading an answer or question that I wouldn't have normally read. It says that, over time, this person has shown that they make positive contributions to the community. As an actual reward, I will show them the courtesy of reading something that they have written to see that I would normally simply dismiss.

MO reputation is like money: by itself, it is completely worthless. What it is worth, is what it can be exchanged for (only unlike money, it doesn't actually need exchanging). What it can be exchanged for is attention, and that really is worth having.

There is a finite resource here: which one could call "attention of experts". I only have a certain amount of time to devote to answering questions. If Joe Bloggs with reputation 1 asks a vague question, then I'm unlikely to waste much time answering because I see no guarantee that my investment of time is going to be made use of. If Theo JF asks a vague question, then I probably will take time to read it carefully and see if I can find some nugget of nougat there because I already know that he will read my answers carefully. Those two are easy cases. What is a little trickier is if, say, 'algori' starts asking questions that I can answer. Then I look at that person's reputation (and user page) and see that they are a "player" and deserve a little attention.

Finally,

the (pure) mathematical community is fairly small, in the end

but very insular and very hard to figure out all the connections and rules and specialities. I've been waiting ages to ask some functional analysis questions because I've never figured out where to ask them before!

The senior mathematicians more or less know everyone in real life and attach more value to publications.

Grad students etc on the other hand, are newer to the scene and might take the MO reputation more seriously.

For an undergrad for grad school applications, or for a grad student for postdoc hunting, MO reputation might actually matter.

I had taken many avatars in MO asking and answering questions; to be precise this is my fourth avatar. I am no major guy here; but I can tell you that life in MO becomes much much easier when you have a decent reputation. Each little bit counts, I would say. Maybe of course when reputation goes over 3000, it may not matter much. That is, there is not much difference between 5000 reputation or 6000 reputation. But, there is some significant difference in the site experience when your reputation is at the bottom levels. In fact it was precisely to distract myself from reputation hunting that I went through many avatars.

I didn't want to rat myself out; therefore I had kept silent on this aspect so far. But it seems this bit of info might be useful to some people in this discussion without them actually doing a new experiment.

Not to mention extremely poor netiquette. If you have concerns that two avatars are the same person, and that this is causing mischief, email the moderators.

Whew! Lot's to respond to. And almost none on competitiveness!

MathSciNet is my "search of last resort". Stuff there is old! Pete lays the foundation of my argument when he says it's a "snapshot of 2 years ago". Except that it's worse than that, some papers take two years, some take longer, some take shorter, it's fairly random. When I do find an article I like the look of, it's generally not available or only via some convoluted set-up that generally makes it not worth the bother. If someone really wants to make it possible for me to use their work, they should make it freely available. If they choose to lock it up, I'll look for some other way to find the information.

I didn't mean it to be a "MathSciNet vs arXiv" debate; I was surprised that someone would use MathSciNet as a way of finding out about people's interests and reacted to it. I don't search either to find out about a person. I'll look for the author's webpage and that's about as far as I'll go. If you look at what I actually said, you'll see that I didn't claim the arXiv to be a better place to judge someone's reputation.

(As for Andy's point about MathSciNet vs arXiv version, all I can say is "huh? Which papers have you been looking at?".)

But back on track. I would want to see real hard evidence that MO is "too competitive" before trying to make substantive efforts to remove the reputation element of it because I think that reputation is extremely useful for the reasons I tried to give above.

I certainly read papers 10-20 years old, older than that even (I think my record is early 1900s), but I regard papers that old slightly differently to papers that are "current". In particular, I'm unlikely to email an author of a paper from a few years back to ask them about details but I would consider it for a paper that's just appeared on the arXiv. So I don't think I'm in danger of turning into a physicist! (I hope.) But I read "old" and "new" articles for different reasons (which is also why I'm much more forgiving with regard to mistakes in arXiv papers than journal papers, not that I'm complaining overmuch about the latter since at least one of my articles on MathSciNet came about from finding such a mistake ...).

I guess we should just end the conversation about competitiveness, since at the moment, there appears to not be any hard evidence for it being an important factor, and nobody is willing to change the rules without such evidence.

@Shevek, mathematicians are humans!

And as a solution I used to put the numbers inside dollars, in MO. Here in meta, a change from markdown to plain text should do the trick.

Yes, but you failed to support any of your claims with anything more than vague nonarguments based on your assumptions about human nature, which I pointed out earlier. I, for one, have no intention of "taking your word for it". If it really is such a problem, you shouldn't have trouble finding concrete evidence.

Yes, they are worthless.

Edit: And I've just noticed that you've tried to argue the same way, with your "I'm sure..." statement. This is a website for mathematicians, so I don't understand why you find it so hard to believe that we'd like some rigour in an argument about site policy.

As I think you know, many of the examples of "extreme competitiveness" on MO can be traced to you. I don't want to rehash all the arguments on meta about this that have gone before, and thankfully much of the offending material has been deleted.

This is not true. Whatever you want to say about the problems I've had, competitiveness has not been a problem.

Shevek-

Examples are different from proof. Most of us aren't looking a peer-reviewed study, just actual examples of this happening. Not many have been forthcoming.

Everyone-

As far as I can tell this thread is degenerating into mostly consisting of tangents and the same people saying the same things multiple times. It's clear at this point that different people have different perceptions of the situation, which is probably just not going to change. Can we perhaps move toward actual suggestions for action (probably small action, but often small action can make a difference. Some people, at least, seemed to think that removing the reputations from the front page was a positive move).