-18

Based on some disappointing early experiences with bounties, I asked this question. I'm new to Meta and didn't structure the question well for my goals, but with some help I've come away with a new suggestion for discussion which I hope better fits the mold.

Pretext

In many responses to questions and feature requests relating to bounties, we hear that bounties are like advertisements — that you are paying for exposure, and that there is no guarantee of ROI. What follows is a distillation of my thought process upon being introduced to that comparison.

Glossary

For clarity and brevity, a couple terms, some of which I may have made up:

  • Benefactor: a person offering/paying a bounty (from here)
  • Bountee: a person seeking a specific bounty by answering a question
  • Publisher: a StackExchange site listing a question
  • Consumer: a reader/user of the site.

Comparison to real world counterparts

In the real world, bounties and advertisements are very different things.

A bounty:

  • exists for the purpose of incentive
  • can be offered for as long as the benefactor wants to do so
  • can have its terms or amount changed at any time
  • costs the benefactor nothing until someone wins the bounty
  • pays the bountee, not the publisher

An advertisement:

  • exists for the purpose of visibility
  • buys a fixed amount of exposure
  • costs up front with no promise of return on investment
  • pays to the publisher, not the consumer

People come to SE with an existing knowledge of the world. It would be good if that knowledge carried over smoothly into SE.

StackExchange bounties are a confusing hybrid of both

Given the way a bounty works today, this makes some sense: you have to "buy" being listed at the top, but you also need to incentivize potential bountees.

I find this conflation of the two topics to be both confusing and less than ideal: if I'm paying for exposure, why am I paying the bountee? Shouldn't I be paying the publisher? If I'm providing an incentive, why must I pay even if I never get a satisfactory answer? Those are very valid questions, and the only answer I've seen today boils down to "because that's how it works."

From the help:

Part of what you’re “paying for” with the bounty is to get additional attention for your question, over and beyond what a normal question gets. In that regard, a bounty does not guarantee a response, and reputation refunds are not available if no answers are received as a result of the bounty.

I disagree; it's not part of what you're paying for. It's all of what you're paying for. You might also get some upvotes or helpful answers, and the latter is almost certainly why you paid, but it is quite clear based on spending a couple days reading everything I can find on Meta about bounties that it is not what you paid for.

To be clear: I am not saying that the existing system is hard to understand. It's quite easy and clear. What it isn't is obvious or intuitive.

I propose that bounties not be treated like advertisements

I am seeking possible paths to that end, as well as discussion on the merits of my thinking.

I have a potential solution in mind. To clarify that my idea is just one of several possibilities, I'm going to post it as an answer rather than part of the question, so that people can agree/disagree independently with the problem I've outlined and the details of a proposed solution, and so that others can post their own solutions.

I understand that making any change along these lines would likely be nontrivial, and that the current system has been in place and working for several years. I do not feel that either of those factors is a worthwhile reason to dismiss this suggestion without discussion.

What do you think of this idea?

  • 2
    Don't forget to read Wikipedia's page: Bounty (Reward) - Some outlaws enjoyed the advertising, some not so much. Sometimes the amount was low to insult, or high to encourage the risk-takers. I'll just leave the link as an improvement and not further comment on your new question; unless a review issue surfaces. --- Better luck this time. – Rob Aug 18 at 20:19
  • @Rob another noteworthy excerpt: "By definition, they can be retracted at any time by whoever issued them." – JakeRobb Aug 18 at 20:42
  • 2
    You should edit your question instead of posting answers to answer someone. – Ollie Aug 18 at 20:50
  • JakeRobb, the Bounty Hunter bringing in the bandit might be a good shot, and unhappy that the bounty was cancelled during the time it took to arrive at the collection point. @Ollie it's "The Dichotomy of the Bounty", and as Jon puts it: a polling question; not prohibited, just uncommon. – Rob Aug 18 at 21:28
  • 1
    @Ollie which answer are you referring to that should have been a question-edit? – JakeRobb Aug 19 at 2:53
6

I disagree that it would be worth spending developer time for such a big reconstruction of the currently existing bounty system.

Bounties are what they are, nothing more nothing less. And even if it's often used as a metaphor that you can think of them like Ads, they probably aren't, and the metaphor is a bit misleading in some points (which is lying in the nature of metaphors).

Before spending a bounty the benefactor can well inform themselves, to avoid the experiences the OP described here.

Also, there are other ways (e.g. editing questions for clarity and improvement) to bump them up, and receive attention (again).

| improve this answer | |
  • To be clear: I did not say that any such solution we might land on would be "worth spending developer time." Before we can decide that, we would have to determine what the preferred solution is, how significant a change it would be, and what the pros and cons are of the new vs the old. In my alternative answer, I suggest a much lighter-touch solution which still addresses my concerns. – JakeRobb Aug 18 at 20:16
  • The fact that many metaphors are flawed is not a reason to pass on an opportunity to improve one. – JakeRobb Aug 18 at 20:36
  • As for informing oneself prior to offering a bounty, yes, of course -- reading the manual first is always helpful, but an honest acknowledgement of human behavior says that making the software behave in ways that are a surprise unless you read the manual first is a violation of the Principle of Least Astonishment. – JakeRobb Aug 18 at 20:38
-1

Here's another potential solution:

Continue treating bounties like ads, but acknowledge the conflation.

Instead of "start a bounty", the link at the bottom would say "promote this question with a bounty".

In the UI that results, you would be told that promoting a question costs 50 rep, and that you can also choose 0-500 additional rep to be awarded to the answerer.

All of the existing timing and automation remains, but the "additional" rep you chose goes back to you if the bounty is not awarded. You don't get your 50 back -- that was payment for promoting the question. The rest was not paying for promotion, but rather increasing the incentive, and that comes back, having been held "in escrow" as described in my original answer, if the bounty is not awarded (and half of it comes back if the bounty is half-awarded).

I'm open to debate about whether the original 50 rep should be included in the award, or if it should just disappear into the ether. I'm leaning toward the former; if the latter, I think we'd need some UI considerations for promoted posts with no associated award.

| improve this answer | |
  • 1
    Why would I answer a bountied question if the additional rep is 0? If there's no benefit to the answerer then we'll just filter out and ignore them. In all of your answers you want to shift power from the answerer to the bounty offerer without acknowledging that if you do that the answerers will just spend less effort (or totally ignore) bounties. – Robert Longson Aug 18 at 19:58
  • Like I said, I'm leaning towards the former. But to respond to your question, do you only answer bountied questions? I am still very interested in a more thorough explanation of the power dynamic you describe here. You continue to shed light on it half a sentence at a time. I have no basis for the perspective that there is any "power" granted to the answerer in the existing bounty system. – JakeRobb Aug 18 at 20:29
  • 1
    My record of what I answer is public in my profile. I rarely answer bountied questions even now because they are usually the most poorly written and least well articulated. – Robert Longson Aug 18 at 20:40
-3

πάντα ῥεῖ said in the comments:

One point you completely seem to miss is that the main goal of Stack Exchange sites, which is to build a FAQ like repository and not primarily to help the questioner. Not all questions asked here can be answered in a reasonable manner, and time limits and urgency aren't important in the light of that.

Me:

isn't that an argument for eliminating bounties altogether? If so, that'd be a valid solution for not treating bounties like ads; feel free to post as an answer.

πάντα ῥεῖ:

Yes it is, I never was a fan of them, nor do I think it's worth the efforts to do such big, ground grinding changes in what they currently are.

Me:

then please post your opinion as an answer so we can see if the community as a whole agrees

I still think this opinion solves the issue and is worthy of consideration, so while it's not the solution I would prefer, I'm posting it as an answer myself.

| improve this answer | |
-3

I think that offering a bounty and getting visibility to your offer should be separate but connectable. Here's how I'd implement it:

Promoted Questions

  • Buying exposure is called "promoting" a question.
    • It has a fixed cost per duration, e.g. 50 rep / week.
      • Potentially, you could choose among multiple durations, with longer durations being slightly more cost-efficient (e.g. 200 rep / month).
    • This would result in the post showing up on the Promoted tab, which takes the place of the current Bountied tab.
  • You can pay for exposure for any question, independent of offering a bounty, provided the question is at least 24 hours old.
  • Answering a promoted question has no inherent benefit over answering a regular question.

Bountied Questions

  • The benefactor can change the amount of the bounty at any time.
  • A bounty doesn't expire until it is awarded or withdrawn.
  • The bounty is not withdrawn from the benefactor's rep until they award it to someone. The bounty amount is withdrawn and held in escrow until the bounty is awarded or withdrawn. If a bounty is withdrawn, the amount is returned to the benefactor.
    • I changed this to avoid the complication of multiple outstanding bounties which, when combined, the benefactor could not afford to pay.
  • If the benefactor is also the asker, accepting an answer leads to a prompt to also award the bounty, and vice versa.
  • Anywhere a bountied question appears, there is a badge indicating that it has an open bounty and the size (I know this is already true in at least some places; not sure about all).
  • There is no expiry and therefore:
    • no grace period
    • no automatic awarding of the bounty
  • Other factors remain the same:
    • A question must still be 2 days old before a bounty can be offered.
    • Bounty awards are still limited to 50-500 rep.

Connecting the two

  • When you promote a question, the system prompts you to also offer a bounty; this is optional.
  • When you offer a bounty, the system prompts you to also promote the question; again this is optional.
    • If you don't, you won't get any listing priority.

Finding bountied and promoted questions

  • A new Promoted tab takes the place of the Bountied tab (to get promoters the visibility for which they have paid). Most recently promoted questions show first.
  • The Bountied tab gets moved to the More dropdown
  • The Filter by option for question gets a Promoted checkbox in addition to the existing Has bounty checkbox.
  • The Bounty ending soon sort option is removed.
  • Size of bounty and Age of bounty sort options are added. These are disabled unless the Has bounty checkbox is checked.

Justification

The existing system warrants improvement because:

  • It attempts to solve two problems (visibility and incentive) with one solution that is not ideally fitted to either.
  • No matter how large a bounty I offer, it doesn't buy me more exposure.
  • The entity providing the exposure (SE) gets nothing for doing so, and yet expects me to "pay" for it.
  • By offering a bounty, I am said to be "paying for exposure," but the entity receiving the award did not provide the exposure.
  • Note all of the confusion here.
    • Some things have changed since then, but notably, it is still not possible (without tomfoolery, e.g. posting an answer using a throwaway account and then accepting it) to prevent a bounty from being awarded to someone whose answer has two upvotes at the close of the grace period but who does not satisfy the offerer's reason for offering.
  • There are ~1400 questions tagged bounties on Meta.

The proposal is an improvement because:

  • As noted here, sometimes those hard questions need more than a week to catch the attention of the right person (and possibly also for that person to craft a worthy answer), and that person deserves an award commensurate with the obscurity of their knowledge and the magnitude of their effort.
  • Sometimes visibility is all you need — especially for a question that might be easy to answer for the rare person with the necessary knowledge.
  • Sometimes the bounty is all you need — maybe plenty of people are seeing your question, but their reaction is, "that would take forever to answer well; it's not worth it!"
  • And, of course, sometimes you need both.
  • It would obviate the need for a grace period and any debate or consternation about how long it is.
  • It would obviate the need for auto-awarding logic when a bounty expires or any discussion about the details of that logic.
  • There are also several bounty-related feature requests and suggestions which would be obviated by this change.

Additional Thoughts

  • Paying the publisher for exposure feels a little weird, the publisher being a free-to-use, ad-supported website which has no use for the "currency" being exchanged. I think that's okay — it doesn't seem any worse than what happens in the current system if nobody is awarded the bounty. Since the amount lost this time is fixed rate, and behavior more consistent with expectations due to the change in terminology, I think it's an improvement.
  • I believe this obviates the need for an "urgency" computation as described here.
  • The block against closing a question with an open bounty becomes unnecessary.
  • A mechanism for the community to award abandoned bounties might be warranted, with the following considerations:
    • This could be a votable action, with such votes being a privilege requiring a fairly high reputation (3k, same as close and reopen votes?).
    • The benefactor should be notified of every instance of community-bounty-voting activity.
    • The bounty is awarded when:
      • the vote total (FORs minus AGAINSTs) exceeds a threshold, e.g. 10.
      • at least 30 days have elapsed since the benefactor was notified of the start of voting activity
      • at least 7 days have elapsed since the nominated answer reached the threshold
      • the benefactor has not:
        • withdrawn the bounty
        • awarded it to another answer
        • voted AGAINST (effectively a veto)
        • (all of these indicate that the benefactor has not abandoned the bounty — the idea here is to handle the case where the benefactor is absent and a bounty clearly deserved)
    • An answer on a bountied question must reach a minimum age before this is possible. (3 months? a year?)
  • Arguably, the Draw attention option when starting a bounty should be removed from the list, as that purpose is now served by promotion.
  • On a user's profile, it might make sense to show somewhere quantity of rep they have currently held in escrow for outstanding bounties.
  • I'd be interested to discuss the merits of, in this system, having very old outstanding bounties (1 year? 2? 5?) auto-return to the benefactor.

Edit: addressing @RobertLongson's concerns:

Bounties should not be withdrawable except by moderators i.e. exactly what we have now. Weaker: Bounty withdrawable, manual awards only or lots of criteria need to be met for bounty award so much so that in low traffic tags automatic awarding would never happen

I'm struggling with the incomplete sentences, but I think what Robert is saying here is that bountees risk not being rewarded for their answer. I'm not sure what low-traffic tags have to do with it, but to the rest, I respond: that's already a thing. A bountee is never guaranteed the bounty.

I've offered two bounties, and nobody got awarded for anything. Once largely due to timing (I was unable to accept the answer in time due to a rookie "don't ship on a Friday" sort of mistake), and once because nobody answered at all.

My takeaway from this as well as lots of the other feedback I've gotten here is that the bounty system is pretty broken, which leads me to wonder why there's so much resistance to changing it.

benefactor offers 500 rep then changes it to 50 as soon as I answer.

Reduction or withdrawal of bounty after receiving an answer while a bounty was offered would be worthy of a flag for moderator attention. With multiple offenses, I think the benefactor should be banned from offering bounties (temporarily? I lean towards no, but AFAIK SE doesn't permaban for anything but spam).

In all of your answers you want to shift power from the answerer to the bounty offerer without acknowledging that if you do that the answerers will just spend less effort (or totally ignore) bounties. You're saying that from the point of view of the offerer bounties don't give you what you want, I'm saying that from the point of someone that might be interested in answering a question with a bounty on it more than one that doesn't have one, your proposals don't give me what I want i.e. an incentive to do that.

Anything that can be abused, will be abused.

Unless I'm missing something, with appropriate moderation against abuse (which I think would be rare, given that one must have a good reputation in order to offer a large bounty of reputation), my proposals generally increase the incentive:

  • a bountee has unlimited time to provide a worthy answer
  • a benefactor has unlimited time to reward a worthy answer

What I've often seen is that bounties are offered on poorly written questions (otherwise they would have likely been answered anyway). The bounty then acts as an incentive to guess what the question asker really meant and the wisdom of crowds gives me a shot at half the bounty. The bounty offerer (usually the question asker) then realises that the answer is not helpful to their precise situation because they simply didn't understand what their problem was or have been unable to articulate it sufficiently for the answer to be what they needed.

By eliminating the time component, I think it would actually reduce the frequency with which bounties are offered on poorly worded questions, as well as the frequency with which bountees simply guess for a chance at the bounty. There's no time limit, so ask clarifying questions until the question is understood (just like we do on poorly worded non-bounty questions), and get the bounty by providing the real answer.

In all of your answers you want to shift power from the answerer to the bounty offerer without acknowledging that if you do that the answerers will just spend less effort (or totally ignore) bounties.

What I want to do is fix apparent problems with the bounty system. It seems that you agree that there are several problems. I am not interested in a transfer of power, nor do I see what about my proposal would actually do that.

I rarely answer bountied questions even now because they are usually the most poorly written and least well articulated.

Then why are you so adamantly against changing the bounty system? You clearly think it offers the wrong balance of incentives.

| improve this answer | |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .