-10

Accept rate dunning has become a meme around here. Yes, we have the accelerated comment deletion feature, but I don't think it's enough. So long as the green check is such a large rep bonus, it will continue to inspire pointless harassment.

Meanwhile, there are other limitations of the big green check. Notably, the OP is not always the best judge of the answer that will be the most useful to future readers of the question. There are sporadic posts here complaining of accepted answers which are counterfactal (i.e., dead wrong).

What if we just didn't have the concept? An OP would have a single upvote, just like everyone else, and future readers would see the community's judgement as the only rating of answers.

An additional note:

I find myself frequently observing the following pattern:

A user with only a few questions (not nearly enough for 'accept rate' to be statistically meaningful) posts a poor-to-awful question. Two kinds of comments arrive: accept rate complaints, more or less snarky, and really snarky criticisms of the question. The later phenomenon deserves attention on its own. The former suggests some of the variations in the answers involving hiding accept rate for people who don't have enough questions to make it meaningful, or from people who don't have some amount of rep.

9
  • 29
    Can you work on your accept-rate it's too low :-). Jun 9, 2012 at 16:16
  • 2
    Maybe we could add a filter. (btw if you want an answer improve your acc3pt r4t3 plz)
    – user7116
    Jun 9, 2012 at 16:17
  • 1
    Any idea about the accepted answer bonus being exempt of the daily reputation limit?
    – Arjan
    Jun 9, 2012 at 16:34
  • 1
    IMO the bonus is already relatively small. It's mainly relevant for users who hit the daily cap. Jun 9, 2012 at 16:46
  • 1
    @TimPost so providing correct answers to solve a problem is significantly less important than a single upvote? And as important as editing "thanks" out of a post (suggested edit bonus)? I don't really like that idea at all.
    – Ben Brocka
    Jun 9, 2012 at 16:50
  • 3
    You do know that many of the people asking questions don't have the upvote privledge Jun 9, 2012 at 16:52
  • @BenBrocka Gah, I deleted my previous comment before I noticed you replied. For consistency sake, I had tossed out the idea of reducing the answer bonus to the same that the question author received for accepting (+2)
    – Tim Post
    Jun 9, 2012 at 16:56
  • I can see the rep cap making this a bit of a problem. (Well, not a problem... just annoying.)
    – Ry-
    Jun 9, 2012 at 17:23
  • 4
    This question can't possibly be downvoted enough. The reputation gain from an accepted answers is one of the fundamental pieces of SE's gamification.
    – user164207
    Jun 9, 2012 at 18:27

9 Answers 9

6

Removing the huge rep bonus from the accept mark (However while keeping the accept mark itself!) sounds like a sensible idea to me. Alternatively reduce the bonus to 2 points to both OP and answerer, as suggested by Tim Post.

Possible positive effects:

  • It might help calm down accept rate harassment, which surely is often about the points and the expectation of "giving back" to the answerer.

  • It would more accurately reflect that the OP isn't always the most qualified person to judge on the perfect answer - the community with its upvotes is.

  • It would discourage answering very localized questions, and debug sessions, in the hope that the OP will reward one's work with the accept mark. I think this is a good thing - helping a user debug an extremely localized problem is a nice thing to do, but it's not beneficial to the community at large. Stuff like that should stay in comments or chat.

  • In the same vein, it would encourage making an answer understandable to everyone, not just the OP, as it's the community that will reward you with votes.

To provide a bit of consolation, a high number of accepted answers could still be worked into some other kind of bonus, like a badge. (If one doesn't exist already.) But I totally agree the bonus could go.

15
  • Perhaps lessened? Give both parties the same +2 and be done with it.
    – Tim Post
    Jun 9, 2012 at 16:32
  • @Tim sounds good as well! Those +2 could also be exempt from the daily cap as a tiny tiny bonus
    – Pekka
    Jun 9, 2012 at 16:33
  • @TimPost: Seems fair to me.
    – Makoto
    Jun 9, 2012 at 16:33
  • 1
    I think it's a good compromise prior to considering just not displaying the accept rate to begin with. Putting it in the user card just gives it too much emphasis. I'm not saying it's not important, but not sure if it's still important enough to go there.
    – Tim Post
    Jun 9, 2012 at 16:34
  • 1
    The third bullet might indeed be a nice side effect. Nice.
    – Arjan
    Jun 9, 2012 at 16:36
  • @Tim If you do the +2/+2 how do you deal with the first point in my answer? Jun 9, 2012 at 16:37
  • @Ben I'd distract them with colorful ribbons and mention something about the lost rep going to feed starving unicorns in Atwoodistan. In all seriousness, that would be a heavy hit for some, perhaps removing a privilege, which brings me back to just taking away the accept rate display completely as a start to see if that curtailed the badgering and pressure to accept answers that really did not help them.
    – Tim Post
    Jun 9, 2012 at 16:39
  • 5
    50% more than a single upvote is a "huge rep bonus"?
    – Ben Brocka
    Jun 9, 2012 at 16:43
  • @Ben it is IMO, seeing as it comes on top of the OP's upvote
    – Pekka
    Jun 9, 2012 at 16:44
  • @Pekka, maybe that's the answer then. Remove the ability of a user to up-vote an answer to their own question? Jun 9, 2012 at 16:46
  • @Ben, 15 rep is also almost 10% of the daily reputation limit, for those who reach that? (Which might be a good reason not to get rid of it.) But disallowing a user from upvoting answers their own question would also disallow them to appreciate multiple answers?
    – Arjan
    Jun 9, 2012 at 16:55
  • @Arjan hmm, yes good point. I'll retract that statement. Jun 9, 2012 at 16:57
  • 3
    I'm settled on not having the accept rate displayed any longer. I'll make a feature request early next week if someone does not beat me to it. The badgering has been brought up here a few times, a few times an OP has asked if they should accept unhelpful answers just to make the percentage go up. Such a proposal needs to point to that, and this proposal as well.
    – Tim Post
    Jun 9, 2012 at 16:58
  • When I first read the title I was like "wth?" but reading this answers, this makes complete sense. +1
    – animuson StaffMod
    Jun 9, 2012 at 17:26
  • @Ben: Remove the ability of the OP to upvote and accept the same answer is the only reasonable option that I see.
    – jscs
    Jun 10, 2012 at 8:13
11

I don't think you can remove the concept entirely. You'll cause a revolt. I was going to say that this is the sort of question I think @Pekka would agree with but he's just answered so I don't need to guess any more.

I'd agree that reducing the bonus could have positive behavioural effects. However, there's a problem.

Would it be retroactively applied? If so meta will be deluged with howling people who've lost thousands in reputation and had privileges removed (revolt).

If it's not retroactively applied then you've immediately given a massive advantage to all users who've been here longer. It becomes that much harder for a new user, or users with less than 10k now to reach that higher level. You would effectively entrench privilege, which I disagree with.

I would be more in favour of modifying the accept-rate system than removing the bonus altogether. Maybe by using bands rather than percentages.

13
  • 8
    I'm almost to the point of reconsidering the value of showing the accept rate to begin with. Not quite there, but almost. A significant number of comments that I delete every day are accept rate badgering.
    – Tim Post
    Jun 9, 2012 at 16:29
  • Well, the "revolt" argument is certainly true. No matter which way you do it - retroactively or not. Still, it might be worth weathering. Also I don't think Rosinante is suggesting to get rid of the entire check-mark system - just the bonus associated with it. That's a huge difference
    – Pekka
    Jun 9, 2012 at 16:30
  • 3
    Ah, now I see one advantage of keeping: it's exempt of the daily limit, right? Hence, it might make people put a little extra effort in answers, to get the best and hence get it accepted?
    – Arjan
    Jun 9, 2012 at 16:31
  • @Arjan I thought of that too, but how many people are affected by that really? How many exceed the daily cap on a regular basis? Very few methinks.
    – Pekka
    Jun 9, 2012 at 16:32
  • 1
    But the top posters might, @Pekka. I can hardly imagine people like Jon Skeet cry over new rules that affect all, but it might make top answerers log out of the site just a bit earlier? I don't know.
    – Arjan
    Jun 9, 2012 at 16:33
  • @Arjan maybe the OP's normal vote (+10) could be made exempt from the rep cap for the answerer whose answer gets accepted? Although that sounds complicated... How about Tim Post's +2/+2 suggestion underneath my answer?
    – Pekka
    Jun 9, 2012 at 16:35
  • @Pekka fair point. Done; it was less relevant than the first by a long way. Jun 9, 2012 at 16:40
  • The problem with bands - most programmers that write answers worthy of acceptance can also do simple math. I'm closer to the point of suggesting that we just remove the accept rate display altogether. If we're going to be at all consistent, lowering that bonus value would be a retroactive change on the heels of rep updating in real time. I see villagers with pitchforks if that happens.
    – Tim Post
    Jun 9, 2012 at 16:49
  • 3
    @TimPost A data point for you: accept rate has been hidden on Programmers since the early days and I haven't noticed any particular issues with that. Mind you, help vampires are probably a bigger issue on SO, so you might not want to discourage a visible reminder to be nice and actually give back to the answerers/community a bit.
    – Adam Lear StaffMod
    Jun 9, 2012 at 16:51
  • @Tim, definitely the pitch-forks. I'm only at 6.2k on SO but removing the 15 point bonus would take off 1,455 rep. I'd be back below 5k immediately. Jun 9, 2012 at 16:52
  • 4
    You know what, I think I will propose to have the accept rate no longer displayed next week. If you care that much about rep - you'll have two choices: 1 - Answer the question before you lose your chance at being the first with the correct answer or 2 - waste five minutes digging through the OP's history to see if you want to answer a question that could earn you lots of rep. That would be a good change, I think :)
    – Tim Post
    Jun 9, 2012 at 16:53
  • 1
    @AnnaLear Yes, they could be reminded after a certain time has passed since they asked a question that received up-voted answers, and that would definitely be more effective if done by the system rather than snarky comments that most mods delete on sight to begin with.
    – Tim Post
    Jun 9, 2012 at 17:03
  • @TimPost I would actually like to see the system also reduce the number of questions one can ask daily on a sliding scale based on the number of questions they have with upvoted answers but no accepts. System reminders will only go so far.
    – joran
    Jun 9, 2012 at 23:41
10

What if we just didn't have the concept?

Well, I think the site would've done just fine without it. Could've just as easily given an asker the ability to up-vote an answer on his own question, and accomplished roughly same thing. Probably reduced confusion a bit too.

But that's speculation. Accept has been part of SO from the start; folks love it so much that when it wasn't possible - self-answered questions - they petitioned heavily to have it added.

Removing it now would cause a major disruption. And for what?

Accept rate dunning has become a meme around here.

Yeah... Y'know, we don't actually need to display that. Let's face it: it was pretty much created to encourage harassment motivate folks asking a lot of questions to accept a reasonable number of answers. The exact number is pretty meaningless; we could just as well display "high / medium / low" and accomplish the same thing with less fixating on it. Or remove it entirely...

But let's be careful here to distinguish between harassment and teaching. Not everyone who shows up here figures out what that checkmark is for, and it's quite easy to forget about it once you've gotten an answer and moved on. A friendly introduction can be perfectly appropriate. But there is no excuse to ever be rude to someone because they didn't accept an answer - the "accept" feature is completely optional, and if someone knows about it and doesn't want to use it, that is their choice to make.

If the latter becomes too much of a problem, I completely support modifying or removing the accept rate display.

2
  • 1
    Increasing the window for when accept rate starts to appear might be an idea...I think it shows up after 4/5 questions? I noticed I was at 0% accept rate with 5 questions on Mechanics, which suddenly (and quickly) looks really bad. With only 5 questions maybe you don't yet have a perfect answer to any....but with 10? 20? 100? Then 0% accept rate starts to look more like you don't care.
    – Ben Brocka
    Jun 9, 2012 at 19:52
  • I guess removing the accept bonus would cause civil war, yeah... ah well. Seeing as this feature request is unlikely to be implemented, I totally support what @Ben says. Bitching about users' accept rates is laughable when they've only asked 10 questions and 5 of them are still unanswered. Introducing a threshold that shows the accept rate only when there's a lot of questions sounds reasonable. Or even combine that with showing only really, really bad accept rates (like "accepted < 30% of at least 30 questions with answers with at least one upvote")?
    – Pekka
    Jun 9, 2012 at 20:00
8

For the record, this OData query tells you how much rep you would lose from the policy change:

Percentage of your total rep that came from accepted answers

The numbers for the top 5 users of the site:

Jon Skeet        39.9%       -174030 points  
Darin Dimitrov   39.8%       -124320 points
Marc Gravell     22.6%        -69360 points
BalusC           36.4%        -98685 points
Hans Passant     31.0%        -77565 points

Different from the 2010 reduction of rep from questions, this measure would disproportionately hit the most productive users, while favouring those with few accepted answers.

While I still think it would be a good change in principle, I guess this is enough to make sure it's never going to materialize.

1
  • 2
    Forgot: me - 15 points
    – bobobobo
    Jun 9, 2012 at 21:42
5

Well uh, I agree with the harassment issues and the fact that the OP may not know what's best, but I also don't like the fact that sometimes the drive-by-n-vote community doesn't seem to read the question either.

So how about an "OP's choice award",

trophy

That's a bit cheesy looking (and I meant it to be here).

Anyway, point is the checkmark now symbolizes "OP's choice". This doesn't mean the website claims it is right anymore, like a checkmark implies, it just means the OP appreciated the answer and would like to award additional rep, and mark it as the answer that "helped him most".

3
  • 1
    Eh, for the most part, accepted answers DO work. I don't think we should cater the system toward those who don't use it properly.
    – Ben Brocka
    Jun 9, 2012 at 19:48
  • hmm interestingly way back in August 2008 Jeff had this to say when working on the "Hot Questions" algorithm Note that accepted answers weight not at all in hotness. This is intentional, as I feel accepted answers are a fine social contract, but not a good data point for question or answer quality. Jun 12, 2012 at 22:31
  • BECAUSE I AM JEFF ATWOOD.
    – bobobobo
    Jun 13, 2012 at 17:03
4

IMHO the original rationale behind the acceptance bonus still stands. Earning the green check signifies that a post was most helpful to the question's OP, and solved his problem best/most clearly/most insightfully/[other]. There are also definitely cases when the most popular answer is not necessarily the most helpful answer to OP; both are valuable, and the most popular answer gains rep by virtue of its popularity. In summary, both the acceptance mechanism and its bonus provide real value to the site. And if anything on the site deserves a bit of rep, I'd say "OP acknowledging your answer was the best" is it.

That being the case, it's sad but predictable that gamers and griefers will lay undue stress on those extra 15 rep points (which, I'll add, isn't always insignificant - e.g. for low-rep users still gaining basic privileges, in questions that often don't generate a ton of outside interest). But removing the rep bonus feels to me like throwing out the baby with the bathwater - it's discarding acknowledgement of superior knowledge and effort (and the value thereof) simply because some users care so much about it (and its on-site value).

Griefers can be flagged and warned. As for gamers, I do like the idea of re-concealing the acceptance ratio - I really don't see what purpose it serves to display that publicly, unless you want people to be choosing their responses according to OP's ratio. (Perhaps displaying the ratio only users above X rep is a good solution here, because above X rep nobody cares about an extra 15 points here and there, for sufficiently large values of X). In any case, I am sure there are better ways to deal with abuse and manipulation of acceptance bonuses than to simply eliminate the bonuses entirely.

0

What if we just didn't have the concept? An OP would have a single upvote, just like everyone else, and future readers would see the community's judgement as the only rating of answers.

I don't like the idea of removing the concept entirely. The OP should be able to express which of all the helpful (upvoted) answers was the most helpful (accepted).

Alternatively reduce the bonus to 2 points to both OP and answerer, as suggested by Tim Post.

I see two problems with this:

  • New users don't yet have the privilege vote up, as it requires 15 rep.

  • Reputation form accepted answers is exempt from the rep cap.

    For instance, this would imply a mayor reputation hit for our top users.

The real issue, in my opinion, is that the OP can upvote and accept one of the answers, multiplying the reputation earned by 2.5.

My suggestion

Retroactively disable upvotes from the OP when an answer gets accepted, without changes the reputation bonus either one generates.

This leaves us with:

  • Most helpful answer: 15 rep (exempt from rep cap)

  • Helpful answer: 10 rep

2
  • That's an interesting idea, but it would lead to similar results in rep loss as the original suggestion. Which is probably too much: meta.stackexchange.com/a/135291/138112
    – Pekka
    Jun 9, 2012 at 20:28
  • I'm not sure. For example, I doubt Jon Skeet would lose any rep with my suggestion. Those extra 10 rep get capped anyway.
    – Dennis
    Jun 9, 2012 at 23:31
-1

As primer, let me state up front that I'd be fine with getting rid of the bonus for accepted answers, but I'm not sure that getting rid of them would resolve the accept rate harassment issue. Perhaps removing that from individual users as a whole would be a better course of option?


I'm comfortable with letting the community decide what they think is the best answer through their upvotes, since it's been the case (and will continue to be) that users will accept answers that aren't based entirely in fact, are subjective, or are just dead wrong.

The main purpose that accepted answers serve, to me, is that an asker has received an answer that they deem is applicable and correct to their situation. This may seem counter-intuitive, at least in the sense that sometimes the truly best answer is accepted, or that some other answer is accepted, but that's how it seems.

How about this for a twist/food-for-thought: Let the Community decide which question merits acceptance, for +0 reputation, based on the total number of upvotes, revisions, and comments to the post. Quality and accuracy could be weighed in a similar manner that "poor" questions are found.

1
  • 1
    Did you say Primer? Washing machine, washing machine, tumble tumble, back in time.
    – bobobobo
    Jun 9, 2012 at 16:40
-2

There are several points being missed here.

  1. Questions with no accepted answers should either be rewritten or deleted. If no one answered a question "properly," it's usually the sign of a bad question.
  2. The check mark should be an indication that a question was addressed, and not just left hanging. Questions with no accepted answers are generally bad questions.
  3. The whole Q&A format idea is a good one, but it typically degenerates into "How can i fix my localized problem?" This isn't a bad thing, necessarily; if you take away check marks, then localized answers (which may sometimes be generally applicable) will get no love.
  4. Knowing that someone has a 10% acceptance ratio is a good metric for deciding whether it's worth spending significant time answering their question. Why answer the question at all if it won't get accepted, and it's not applicable to 3,000,000 other people who might vote it up?

Rep is really just a way of keeping score. If you make it too hard for people to earn rep, they may decide it's not worth it. YMMV.

3
  • 2
    #1 is dangerously close to Experts Exchange logic...regardless of the question, if it's old and no answer, delete it! really the wrong way to handle "broken windows", hard questions aren't broken windows, bad questions are. Just because it hasn't been answered and accepted yet doesn't mean the question (or answers) are bad.
    – Ben Brocka
    Jun 9, 2012 at 20:47
  • @BenBrocka And just because an answer hasn't been accepted doesn't mean it's hard. Sometimes it's just a bad question.
    – CodeGnome
    Jun 9, 2012 at 20:57
  • We've no shortage of means to handle bad questions. Hard or obscure questions, on the other hand, add value (because they give an opportunity to light up a unique knowledge niche) but have a lower accept rate - and are somewhat harder to identify clearly. (Also, a good answer in a non-obscure question is likely to net good rep even without the +15 bonus.)
    – Ziv
    Jun 10, 2012 at 7:55

You must log in to answer this question.

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