# Can I “raise a bounty” for a contest?

I would like to ask a contest-style problem on puzzling SE and I came up ith an interesting idea. Could I start a contest in a post and offer all reputation that I recieve as upvotes on that question as a bounty prize for the winner?

If not, can you give other suggestions as to how I might conduct a contest? Is this even allowed?

• Could you clarify a little what you mean by "a contest-style problem"? Thanks. – Gareth McCaughan Mod Jun 15 '17 at 10:14

So far as I know, there's no mechanism that will automatically have that effect.

You can certainly make a promise that you will "donate" reputation from upvotes to a particular question as a bounty. However:

1. Bounties only come in multiples of 50 points. Upvotes on questions are only worth 5 points each. So in order to get enough upvotes to make a minimal bounty, your question would have to get at least +10. That's not terribly unlikely if the question is well-received, but it's by no means guarantee.

2. If you do manage to post a question that gets enough favourable attention to earn you enough reputation to equal a bounty payment, then very likely the best answer(s) will already be getting plenty of upvotes. And upvotes on answers are worth 10 points each, and good answers tend to attract more upvotes than good questions. So your bounty will probably not, in practice, be a very large incentive.

3. If you say "rep I gain from this question will be handed out in bounty" then, especially as you actually can't do that exactly unless the number of upvotes you get happens to be a multiple of 10, some users may suspect that this is really a ploy to get rep for yourself. (Or to gain upvotes, which look good even if you then sacrifice all the resulting rep as a bounty.) Even if they're 100% wrong, they may believe it, and the result may be hard feelings on one side or the other or both.

4. It's not at all clear why a good solution should be better rewarded if the question is higher-voted. What do those two things have to do with one another?

In view of all this, my feeling is that if you want to do something like this then you should probably decouple the bounty from the upvotes on your question. Just say "The best answer will receive a 50-point bounty" or whatever. You may end up losing Imaginary Internet Points overall in this case, but that doesn't seem like a terribly scary risk; they're only Imaginary Internet Points. And if your challenge is well received you may come out net positive, which actually seems only fair.

• I agree with everything you've said, but you've focused only in the logistics of awards and don't address the issue of effectively breaking the current "too broad" policies. – Alconja Jun 15 '17 at 2:58
• True. That would be another problem with the whole scheme. It feels like there are really two quite separate questions here: (1) "can I ask a contest-style question looking for the best answer rather than the unique right answer?" and (2) "can I reward it with a bounty in such-and-such a way?". My answer is all about #2. – Gareth McCaughan Mod Jun 15 '17 at 10:11
• (#1) sounds like just an open-ended question to me – dcfyj Jun 15 '17 at 15:57
• @dcfyj - only if there's an objective measure for determining the winner. If it's just a popularity contest, or the one that op likes the most then it moves into "too broad" territory. – Alconja Jun 15 '17 at 21:53

What follows is my personal opinion and not policy... (i.e. before doing anything see how the votes on this or any other answer plays out, to gauge general community consensus)

I have seen questions asked on the main site in the past that effectively constitute "contest" questions and they've been closed as too broad. This is strictly true under the current policies and I don't think you'd have much luck taking that approach.

That being said, we also have precedent for "open ended" bounties here on meta, and we similarly have precedent for topic challenges and other community "events", so I think there's still wiggle room to work with.

So, if this were to work, I think the way to go about it would be to create your "challenge" post here in meta, stating your contest restrictions and proposed deadlines/prizes, so that participants can post their entries as unique puzzle posts (each individually following normal site rules). People could then post on the meta question ...i.e. basically make it like a custom version of the normal fortnightly topic challenge.

Two challenges/problems I foresee though:

1. You may get some community resistance because it arguably competes with the existing FTC process, which is more democratic
2. You'd receive no bounty from your question to pass on to the winner, so you'd have to find alternative ways to "fund" it*

My personal opinion is that we should absolutely support this sort of community activity, so I'm all for it (under the FTC-style structure described above).

* If the general idea gets community consensus, and I personally find the contest topic interesting, I'd be happy to contribute to the prize pool.

• This answer assumes, unless I'm misunderstanding, that by "contest-style problem" Nilknarf means a question inviting the making of other questions. I interpreted it more broadly: to mean a question with lots of possible answers where the goal is to get as good an answer as one can rather than the provably best answer or the unique correct answer. I wonder which was intended. – Gareth McCaughan Mod Jun 15 '17 at 10:14
• @GarethMcCaughan - you're right. I interpreted "contest" to mean "make a puzzle" (similar to FTC), but now that you point it out, that's very much not clear from the question... If the intention was a contest looking for the best answer, then I think there's clearer policy at least (either it's ok, as per existing open ended puzzles assuming there's a measurable/objective "best", or it's not ok/too broad if it's just a popularity contest). – Alconja Jun 15 '17 at 10:50
• I interpreted different than both of you. I took this to be, can I make a question or puzzle that should be difficult, and see how can solve it fastest, most efficiently, most creatively etc etc. Rewards going to those answers. So while it may be broad, it'd still have an answer, but the means of getting it may vary. - granted if that's the case, it sounds like that'll be more of a math problem or lateral thinking which may not fair well. – n_plum Jun 15 '17 at 12:44

To me, this sounds like a great idea!

As to how it could be done, you could list the bounty conditions at the bottom of the puzzle, and when done you could award a bounty corresponding to $\left\lfloor\dfrac{question\_upvotes}{10}\right\rfloor\cdot50$ reputation (what you receive, rounded down to highest multiple of 25 below that).

I don't see any problems with this, though I haven't been on this site for very long, so I'll defer to some more experienced site members for this.

• Except that bounties can only be awarded in discrete amounts, as mentioned by @Gareth. You would need at least 10 upvotes to even be able to provide a bounty from the reputation, and you would have to either over-bounty or under-bounty unless your rep from upvotes happened to exactly match an existing bounty amount. – GentlePurpleRain Mod Jun 15 '17 at 14:27
• @GentlePurpleRain That's what the formula does (or at least I think it does). – boboquack Jun 15 '17 at 21:54