14
\$\begingroup\$

I think that the Bounty system is not working correctly on our site and needs a solution for it.

The issue I am seeing is that questions with a bounty on it barely get any extra attention, while they do seem to get answers.
In order to fully monitor the situation we would need a SEDE query, which I myself am unable to give, but let's take a look at the posts with a bounty on them right now, after we clarify what a bounty means:

A bounty is a special reputation award given to answers. It is funded by the personal reputation of the user who offers it, and is non-refundable. If you see a question that has not gotten a satisfactory answer, a bounty may help attract more attention and more answers. Slice off anywhere from +50 to +500 of your own hard-earned reputation, and attach it to any question as a bounty. You do not need to be the asker of the question to offer a bounty on it.

(Emphasis mine)

So the goal of a bounty is to attract both more answers and more attention, the latter which can be measured with views and upvotes.

We also know that questions that go "hot", end up receiving 10+ votes and 500+ views a lot of the time. It is possible for a question to get 10+ votes quite reasonably without getting hot, but then usually the number of views is far under 500.

Currently we have the following questions with a bounty on them:

While it does seem that the bounty system gives a quite reasonable (around 50% possibly) chance that it gets an answer, it does in its current form on Code Review not live up to the expectation that a bounty will help a question attract more attention (in both views and upvotes).

So I'd like to start a discussion on the following points:

  • Are my observations correct? Can anyone back them up with SEDE data about the usefulness of bounties? Possibly compare the upvotes/views/answers of the question before the bounty started with the situation once the bounty has ended?
  • Is it really a problem that questions with a bounty do not get more attention?
  • Should the system be changed StackExchange-wide? For example, give the featured questions a more prominent place on the homepage, or let it appear in a bar similar to the hot network questions?

I would expect a bountied question to generate close to the amount of upvotes and views a hot question would. I also strongly believe that the questions in example here are excellent questions that would have received a lot of upvotes and views if they went hot.

\$\endgroup\$
  • 2
    \$\begingroup\$ The LogManager Tests answer, it's a selfie. \$\endgroup\$ – Mathieu Guindon Oct 6 '14 at 14:39
  • 1
    \$\begingroup\$ I've found myself once having to "pimp" a question I had put a bounty on in chat because the bounty was about to expire and I didn't want the rep to just go to waste. \$\endgroup\$ – Phrancis Oct 6 '14 at 15:45
9
\$\begingroup\$

Ranty answer, to try and get the "mindset" of someone out there why bounty questions get less attention.


I took a quick look at the linked questions.

My quick reactions:
The first one: - I don't know those languages. Question looks hard.
The second one: - I don't wanna review javascript. I don't like that language. Scripting languages seem to require a different mindset (more ad-hoc, less structure) because otherwise you wouldn't have gone with that language in the first place.
The third one: - Bleh. I don't like VBA either; It has funny syntax and it's something that serious yet I have no idea what you'd use it for.
The fourth one: - More bleh.
The fifth one: - Javascript again.

You can add bonuses to questions I don't like all you want, it's not gonna help the part where those questions don't interest me.

Now, the recent hot questions...

Celebratory fireworks animation
Meta-effect. It's about a meta event (CR graduation).

Regularity in the "Rusty Towel of Mutual understanding"
Java. I like Java. Also, the question was fun.

Proof that I was wrong about Random Number Generators
C#. Follow-up... and... meta?

The Term-inator: Pi edition
C++. Seems to have a fun title.

If she floats then she is not a witch like we thought
C++, fun title.

Adding a course to one of 8 periods
Java. Interesting title.


In an all-you-can-eat restaurant, you cannot tell me to "eat your greens, you'll get dessert after that" because I can get up and fetch dessert myself.

On places like Stack Overflow, reputation is "hard" to get. (Kinda?) So when someone offers you a huge chunk of reputation, that's worth a lot.

Here? 50 points?

Haha.

Most questions around here, you post an answer, you will get 20-30 points. Possibly 50-80 points. If you get 90 points (9 upvotes) you'll get another upvote from someone via the 1-short of a badge SEDE query. And those results are cached so you might get two upvotes.

There's plenty of questions to answer. There's many answers for a single question, so you need less overall knowledge to have an answer to a question.

With a 50-point bounty, all you're doing is adding a small bonus. A small bonus in points that are plenty. I don't care about that bonus. I give those questions a look from time to time, but generally they come with a ton of baggage: The question makes me spend a lot of effort to answer it. And I don't even like the question. So I'm not gonna answer it.

Then there's the whole thing surrounding bounties: Only one person gets the bounty. The rest gets nothing.

If upvotes gave +1 and accepted answers still gave +15 Code Review would be a mighty different place when it came to points.

I have 20 accepted answers. That makes for 300 reputation. I have accounts of other sites for +100. I once asked a question and got 11 upvotes (55 reputation). I made a couple edits before I got to 2K. Maybe 500 reputation of my 4.1k is non-upvotes. The rest is all upvotes.

It's far easier to get upvotes than accepted answers. 360 upvotes versus 20 accepted answers...

End rant, lets look at your sub-questions.


Are my observations correct? Can anyone back them up with SEDE data about the usefulness of bounties? Possibly compare the upvotes/views/answers of the question before the bounty started with the situation once the bounty has ended?

I don't have SEDE for you, but I feel bounties are useless. You can apply them once your question has started to fall into zombie territory. All you're doing is making it a special zombie. It's not gonna get your question answered like it would on SO.

Is it really a problem that questions with a bounty do not get more attention?

No? Yes? It's sad if you ask a question and it becomes a zombie. Not being able to get people to answer it... yeah, that's sad.

But TBH, the reason I'm not answering your question is because it's not interesting to me or because I don't have the knowledge.

You've either posted your obscure thingmajig that does something that's probably useful to someone (but not to me). In this case, you want a full review of your thingmajig; I guess it does what it needs to do and all we can do at this point is be an expert of the subject of thingmajiging to realize your thingmajig needs some extra helper functionality to be thingmajig-A3-B compliant. Or maybe find some nitpicks that you're happy to know, but weren't really interested in. Basically you've made something and are verifying its quality by a lack of answers. Most of us are not gonna write "Looks like you've done it all correctly" answers because those can be wrong. Maybe I would write an answer like that, but you state that you don't like the way you did X or Y in your thingmajig. And I agree, I don't like that either. I don't have a solution for you, so I can't tell you you did everything right either.

Maybe you have a design flaw that you'd want to get fixed ("I have a problem, please fix it in this review"). And since nobody answered it you added a bounty. I, perhaps incorrectly, assume said bounty requires me to solve your problem. I can't do that for you, thus I don't answer your question.

In these cases more attention is given, but none of this attention results in answers. Because we don't know. Or because your question looks like we won't know. Whether that's a problem... well... yeah, it is, if you relied on CR to get things done for you. I hope not.

Should the system be changed StackExchange-wide? For example, give the featured questions a more prominent place on the homepage, or let it appear in a bar similar to the hot network questions?

Hot questions are hot. Bounties are not. Bounties are difficult questions that you can answer for big points. Usually hot questions are questions that can be easily answered. They get answers for various angles. On a hot question, you get one style review, one algorithm optimization, a general optimization, a sidecase review, error handling and maybe some alternative solutions. And they all cover something different.

On bounty questions those things aren't available. You got your whole system already and it needs to do this cleaner or better or what have you. Usually, this means one answer is correct. The rest are nice to have. Also, the question is harder.


Questions go hot and get bounties for different, mostly even opposing reasons.

Hot questions are mostly easy. Everybody understands what's going on and everybody understands the solutions. That's why it's hot; it's understandable and relatable.

Bounty questions are mostly hard. Someone wants it to get answered but it hasn't yet. It's deep and only for those with the knowledge. It's a far more localized issue. That's not likely to go hot.

\$\endgroup\$
  • 1
    \$\begingroup\$ This answer would seem to suggest though that the primary motivation for answering is fake internet points (which I'm not sure it is). Good (well thought out, complete, working code that shows effort) questions are what will hopefully attract users to CR because these types of questions will attract people who enjoy solving problems (since on CR a quick Google search will most times not yield a useful answer) instead of those who just grind for points. Is my understanding of CR deficient? \$\endgroup\$ – xDaevax Oct 6 '14 at 14:15
  • \$\begingroup\$ @xDaevax Part of it is the points, part of it is puzzle solving, part of it is helping people. Bounties don't give enough points on CR, upvotes are plenty. Reviewing well written code is hard; usually solving the puzzle is both in detecting the code smell and solving it. In well written code, you have to search long and hard before finding a puzzle. Thus there's nothing to solve (boring). Lastly, regarding helping people... when I see well written code I get the feeling you don't need help, you're doing perfectly fine on your own. \$\endgroup\$ – Pimgd Oct 6 '14 at 14:22
  • \$\begingroup\$ Interesting point about finding the smell. Thanks for the clarification. \$\endgroup\$ – xDaevax Oct 6 '14 at 14:38
  • \$\begingroup\$ I thought +200 would somewhat counter the "bleh" effect ;) \$\endgroup\$ – Mathieu Guindon Oct 6 '14 at 14:49
  • 2
    \$\begingroup\$ @Mat'sMug It's VBA. I took one hard look at the question, realized I didn't understand crap about it, saw that you were asking a bigger-picture question (You're looking for flaws that affect the bigger picture, design flaws, algorithmic flaws), and realized I was not gonna be able to provide the answer. \$\endgroup\$ – Pimgd Oct 6 '14 at 14:53
  • 1
    \$\begingroup\$ I agree with most of this answer. There's just this, 'Most of us are not gonna write "Looks like you've done it all correctly" answers because those can be wrong.' You can avoid being wrong about it if you're specific about what's done right. I've said before that we all need to focus a little more on what was done right and I stand by that claim. \$\endgroup\$ – RubberDuck Oct 7 '14 at 8:18
7
\$\begingroup\$

Anecdote time!

I've personally found that for the 2 bounties I've applied, I've either gotten at least one answer (typically when the bounty first goes up, or as the bounty approaches expiry), and in the cases when I haven't received an answer (and often when I have), people have upvoted my question enough to cover the 50 rep loss.

I have noticed an attention increase as a direct result of applying a bounty, but some questions just aren't easily answerable.

One gripe I do have, is that the bounty period is too short for Code Review. Reviews are not always quick, particularly for large questions, and sometimes people need a bit longer to read through and see it.

\$\endgroup\$
  • 4
    \$\begingroup\$ Bounty period too short, yes! That might absolutely be an issue here. \$\endgroup\$ – Simon Forsberg Oct 18 '14 at 18:17
6
\$\begingroup\$

I, for one, have so far always gotten my money's worth for bounties that I have placed on questions.

Bounties serve as an additional incentive to answer, but they won't turn a poor question into a good question. If the question was poor to begin with, then putting a bounty on it would likely be a waste of points. And rightfully so. If bounties "worked" the way you described, we could equally lament that questions with bounties are overshadowing meritorious questions.

\$\endgroup\$
5
\$\begingroup\$

To add to the other answers:

Are my observations correct?

No SEDE, but I guess so as I haven't noticed any bounty yet.

Is it really a problem that questions with a bounty do not get more attention?

I think so. On the one hand, bounties are damn cheap here as Pimgd stated. OTOH they have hardly any effect. IMHO this should be changed by making them more visible, e.g., randomly occur on the top positions.

Showing them in the place for hot questions makes IMHO sense, too. As Pimgd stated, hot questions are easy, which is a good thing, but it gets too boring. Mixing in a bounty question from time to time would make it more interesting.

There's one overlooked aspect of bounties: Despite they low value here, they show that the OP is really interested in an answer.

Should the system be changed StackExchange-wide?

Probably. For sure, CR is special in this respect. I'd suggest to make bounties slightly more prominent on all sites, but more so on CR. This could be controlled via a simple scalar parameter.


Nick Udell wrote:

One gripe I do have, is that the bounty period is too short for Code Review.

I'm not sure about that. But I'm sure, I'm not gonna spend more than one week with answering. :D And the OP might not need it anymore.

I guess, that drawing more attention to bounties would do. If you spotted the question early, then one week should do.

\$\endgroup\$
  • \$\begingroup\$ You know that all active bounties show up in the featured questions tab, right? ;) \$\endgroup\$ – Mathieu Guindon Oct 19 '14 at 18:50
  • \$\begingroup\$ @Mat'sMug You know it's been months/years I recently visited the featured tab? I always open my 5 bookmarked SO/CR "newest" pages and none of them brings me to "featured". It's not interesting enough to be bookmarked and so it could be directly non-existent. \$\endgroup\$ – maaartinus Oct 19 '14 at 19:03
  • \$\begingroup\$ @Mat'sMug What's worse, the featured "tab" is sort of persistent. After having switch it once, opening my bookmark brings me to "featured" (I guess, there's a workaround, but I'm too lazy). \$\endgroup\$ – maaartinus Oct 19 '14 at 19:19

You must log in to answer this question.

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