I am giving a course which occasionally features some basic arguments from functional analysis. I am giving a lecture today, and when I was looking through my notes last night I noticed a hole in one of the arguments I was going to present. I was at home, far from the friendly analysts down the corridor (and I am no analyst myself). "In desperation" I thought I could ask MO instead, so I did:

http://mathoverflow.net/questions/27521/injection-between-non-isomorphic-irreducible-hilbert-space-reps

What was going on in practice was that I ideally needed an answer within 14 hours. Now the nice thing about MO is that it's not at all unreasonable to expect that a sufficiently basic graduate level question might well get answered within 14 hours, and probably a lot more quickly. But I have reputation to spare and I wanted to make sure, so my plan was to offer a reasonable bounty, not because I thought the question was hard, but because I wanted an answer ASAP.

I then found I couldn't give out a bounty :-( because the question had to be open for 2 days.

So here's a generic statement/question---occasionally I might want to reward a quick answer. How can this possibly be built in to the system?

The most naive idea would be to argue that I should be able to offer a bounty immediately. I imagine that this is not possible because of potential abuses of the system (either "let's both log on at 10:00, I'll ask a question, you answer it, I'll accept and you can get that 500 rep you need", or perhaps "I'll log on at 10:00 with both my accounts and then transfer 500 rep from one to the other"). It's not clear to me whether these things could or would actually occur, but I am guessing that is the logic which inspired the no early bounty feature.

But there are more complex ideas, of the form "I can guarantee that there will be a bounty on this question within 2 days, and you can have it if you answer quick", which could be implemented.

Note also that any user who wants to transfer 500 rep from one of his accounts to the other can still probably do this, by asking a generic question, waiting two days, and then accepting an answer from another account...aah, I see the problem. It's hard for a non-mathematician to ask a generic question which won't be closed within 2 days, and presumably you can't offer a bounty on a closed question.

Anyway, that was the situation I found myself in. I wanted to reward the guy who answered quickly, but I can't. I could have waited 2 days, offered a bounty, and then instantly accepted his answer, but I thought this was a bit ridiculous. In particular I could see no reason not to instantly accept the answer the moment I realised it answered the question perfectly. The speed was just a bonus.

Perhaps the "speed" issue is something that comes up sufficiently rarely that this is not worth thinking about? As you can see, I can't formulate an explicit proposal for a feature which would have helped me out here. On the other hand I felt that MO wouldn't let me do what I wanted to do, so I thought I'd mention it here before I forgot.

I remember that I had a similar situation a while back. For much of what the bounty would do, I think this is a case where one can't expect MO to do everything and so one has to rely on the fact that the reputation stuff is an expression of your (or the communities) gratitude, but does not necessarily reflect accurately the degree of the gratitude.

There is one aspect of the bounty system that you didn't touch on that might have been useful to you: the fact that bounty questions are marked out from the rest so your question would have attracted more attention simply by virtue of that fact.

Maybe having a tag "I'm lecturing on this tomorrow so please help!" would adequately cover this facet.

As I think I've said elsewhere, I use all the reputation, voting, and so forth to help me figure out what to devote time to on MO. But it's a guide, not a be-all-and-end-all. If I see a question that needs a speedy answer, I'll take a look and see if I can think of something that might help, even if I can't think of an actual answer on the basis that a small idea early on might be more useful than a full answer after the lecture! (Note that this is exactly what I did in the case in question)

Reading the above paragraph again, I think that it summarises my view of MO quite well:

I want to be helpful to the "mathematical community" beyond my own immediate surroundings (both physically and thematically). But that's quite hard to do in an effective manner. The infrastructure of MO helps me focus my efforts so that I can maximise the help that I can give. However, if I see an opportunity to be helpful outside of the framework of MO (which includes being helpful on MO but outside the reputation system) then I'm not going to pass it up!

(I think that this has application to the "vague question" debate currently raging elsewhere in these hallowed halls: to be helpful, I need to be able to find easily the questions where I can help. An excess of vague questions makes this very difficult and so would, ultimately, lead to me looking for other ways to contribute.)

However, my response is not to worry about this at all, because MO is already perfectly structured to reward quick answers. I have asked 30 questions on MO, and I cannot remember one which was answered at all but not answered within 12 hours of my asking it. In your particular case you got an answer one hour later, right? Problem solved.

This is because most of your questions are in highly MO-populated subject areas. Even simple questions about homotopical category theory sometimes take a good deal of time to receive an answer (also commutative algebra, for some strange reason), and that is certainly an area that is more populated than functional analysis.

Actually, the right place to post the request would be meta.SO since the SE framework is migrating to the SO codebase.

I think this is a case where adding a comment to the question to the effect of "I would really appreciate any speedy answers, as I ought to know this for a lecture tomorrow!" would pretty much have the desired effect. It's not just "reputation" people are after -- many mathoverflow participants like to be helpful!

The bounty system is generally agreed to not have been thought out too well, but it appears to be a genuinely hard problem. To my knowledge, no one has a coherent idea of an alternative system.

See also the meta.SO post Why can’t I just offer a bounty for a question right off the bat?

I think Scott's solution is pretty close to optimal. In general, if it's hard to imagine a good software solution to a problem, it's probably best to just communicate the problem in the body of the post (or in a comment).

Somewhat related on the SO blog: Improvements to Bounty System