What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 131 Stack Exchange communities.

If you take a look at this person's profile you'll see that 3/4 of his reputation is from the +2 bonus for accepting an answer. Yes he gains rep very slowly, but I still feel that it is a problem.

I therefore propose some kind of threshold for when the +2 bonus should not be awarded. Possible candidate thresholds are:

  • When rep gained by acceptance bonuses is greater than 1/2 of their total rep
  • When their question to answer ratio is more than 10:1
  • When their total rep is greater than 1000

What are you ideas? Personally I like the question-to-answer ratio idea as a way of nudging people to try to answer more questions.

share|improve this question
    
When their question:answer ration is more than 1:1. –  Lawrence Dol Jun 25 '10 at 0:07
    
@Softw to be honest my question to answer (depending on site) is usually like 0.9:1 so that's a bit too harsh really. I think it starts getting ridiculous though after 5:1 –  Earlz Jun 25 '10 at 0:10
3  
I gave up answering his questions, he likes to dup his questions a lot, and he doesn't seem to listen to guidance. –  Lance Roberts Jun 25 '10 at 0:18
1  
Are you perhaps drawing inspiration from my and Bill the Lizard's comments here? meta.stackexchange.com/questions/54925/… –  Ether Jun 25 '10 at 0:36
    
(-1) for the reasons in my answer. Acceptance bonuses are a good thing. –  devinb Jun 25 '10 at 8:24
1  
Where does the 75% figure come from? Accept Rate is 98% of 549 questions asked. = 538 accepted questions. 538*2 = 1076 (59.4%). –  Martin Smith Jun 25 '10 at 10:56
    
@Martin: I think that's a mistake on my part. When I (or anyone with > 10K on SO) visit that user's page I see the user has a total of 741 questions asked. That includes deleted questions, which aren't visible unless you flip through several pages of results depending on how you have them sorted. Since those deleted questions don't figure into the accept rate, your figure is the correct one. –  Bill the Lizard Jun 25 '10 at 12:39
    
Ok, fine, but don't complain about people's accept ratio if its implemented :) –  Tim Post Jun 25 '10 at 15:47
    
-1: People always forget to take into account bounties –  Casebash Jun 29 '10 at 23:51

5 Answers 5

up vote 8 down vote accepted

My general perception is that these kind of problem users aren't really so common as to warrant mechanical measures. We do get, in addition to those kinds of users, people who ask a lot of really good questions but give few answers. The person I just linked in fact asked this Meta question, where it seems we resoundingly agreed that asking lots of good questions is perfectly fine even if you aren't contributing answers.

It's only 2 reputation, so it isn't a big loss to them. But it feels kinda defeating, that suddenly you aren't getting anything. Some people just fit better by asking questions, and it feels rude that they can give others 15 reputation but cease to get the normal returns for it.

For your candidate selections...

  • The half reputation bit is almost the most reasonable, until you consider the fact that many early users can end up completely walled by this. Early users will need to ask really good questions that get a lot of upvotes, or get some good answers in, which isn't very friendly to the many users who join in order to ask questions that they need help on. But past those early users, it's the only one that will generally strike problem users more, as the good question askers tend to get upvotes... unless the user posts extravagant bounties. Even I've been tempted to spend so much on bounties that I'd be broke, if not close to it. The user that inspired this thread has posted at least 900 reputation worth of bounties, which is what makes it look like acceptance is such a significant fraction of the total reputation gain. So in the end, this also has some strong potential faults.
  • Question ratio is difficult because it's difficult to have a good ratio that will catch people with really problematic ratios but not catch the people who simply haven't been around to catch good places to give answers. The fellow I linked is 72:6. Anyone with only 1-10 answers can easily ask five times as many questions, or more, if their line of work or curiosity demands it. Meanwhile, with 50 answers (good or bad, as it were) you can cruise clear for hundreds of questions. So it's not really a safe metric - it's quite simple to workaround with satisfactory answers, yet too strict of a ratio will harm legitimate users.
  • Reputation BLOCK is about the worst possible idea I can imagine. The point of earning reputation is to gain new privileges. It is quite backwards to subtract abilities for that.

Instead of afflicting everyone, I think it's simpler to highlight the problem users individually, such as by flagging. In fact, I bet it isn't too difficult to devise some Data Explorer query to find the users with moderately high reputation and >50% reputation gains from acceptance.

share|improve this answer

It seems a bit contrived to explicitly check for the source of reputation this way.

This is my personal opinion but I'd be just as happy to see the +2 acceptance bonus done away with completely. When I describe reputation as something that others award to you, I always have to preface that with the exception that accepting an answer awards yourself reputation. It's a philosophical distinction.

There are other ways we encourage accepting answers without allowing users to award themselves reputation.

share|improve this answer
    
I kinda like getting the +2... it's like getting cookie –  jmfsg Jun 25 '10 at 1:01
2  
Meh. It's not getting a cookie, though. It's taking a cookie. Self serve. –  Robert Cartaino Jun 25 '10 at 1:06
    
To me, +2 is so small (even back when I only had 26 reputation) that it doesn't seem all that bad to lose it. But on the other hand, it's so small that I don't think it's all that bad to leave it alone, as well. It's barely a motivation for trying to build anything substantial off of it, and most of the time it's insignificant compared to getting a good answer. Except, naturally, cases like the one that brought up this discussion, but those are the kind of users that stick out. –  Grace Note Jun 25 '10 at 1:15
2  
For new users who don't understand everything about the community, the +2 is a good way to get them on a habit. –  TheLQ Jun 25 '10 at 1:36
    
@Grace Note: Agreed, not a big deal at all. Like I said, it's a philosophical argument. –  Robert Cartaino Jun 25 '10 at 1:38
1  
Mmm, from a philosophical standpoint, yeah, I can see that. –  Grace Note Jun 25 '10 at 1:40
    
I was curious about how much rep the average user gains for accepting answers: meta.stackexchange.com/questions/54949/… –  Jon Seigel Jun 25 '10 at 2:02
1  
(-1) I think the +2 is valid and should be kept. Please see my response to the question. –  devinb Jun 25 '10 at 7:30

(This should be a comment on Robert's answer, but I need more space.)

I was curious about how much rep an average user has gotten from accepting answers, so I did a bit of Data Explorering since it's pretty smart.

Amount of rep gained from accepting answers (query link):

SELECT COUNT(*) * 2 FROM Votes v
    INNER JOIN Posts a ON a.Id = v.PostId
    INNER JOIN Posts q ON q.Id = a.ParentId
    WHERE
        v.VoteTypeId = 1 AND
        q.OwnerUserId != a.OwnerUserId AND
        (q.CommunityOwnedDate IS NULL OR q.CommunityOwnedDate > v.CreationDate)

returns 782,732.

Number of users who have asked at least 1 question (query link):

SELECT COUNT(*) FROM Users
    WHERE
        (
            SELECT COUNT(*) FROM Posts
                WHERE Posts.OwnerUserId = Users.Id AND Posts.PostTypeId = 1
        ) > 0

returns 126,165.

So taking 782,732 / 126,165 = 6.2 rep/user, which is pretty miniscule in the grand scheme of things.


EDIT: I wrote another query to give a summary of all the users with a lot of rep-gaining acceptances. You can find the query here (query statement is too long to paste here). There are some pretty interesting results in there. (The specific user in this question has changed his/her user name since the data dump -- inspect the user id, it's near the top of the list.)

share|improve this answer
    
If I've made any mistakes in either of the queries, feel free to point them out. –  Jon Seigel Jun 25 '10 at 2:04
    
Why not link straight to the queries on DE? –  Michael Mrozek Jun 25 '10 at 2:34
    
I'm getting back 472,057 total accepted answers from the API right now. Do you not gain rep from accepting answers on CW questions? (I feel like I should know this.) Also, Happy Birthday! –  Bill the Lizard Jun 25 '10 at 2:41
    
@Bill: Thanks! Also, nice catch, I have to exclude CW accepted answers. I will attempt to fix my query. Note also that the query here already excludes self-accepted answers which generate no rep. –  Jon Seigel Jun 25 '10 at 2:50
    
@Michael: Done. –  Jon Seigel Jun 25 '10 at 2:57
    
@Bill: Fixed. It seems the API has more up-to-date data (it's live data, isn't it?). If I remove all the constraints in the WHERE clause except the vote type, I get 434,725 acceptances. –  Jon Seigel Jun 25 '10 at 3:00
    
It's not minuscule if some users have garnered 98% of their rep from accepting. –  Lawrence Dol Jun 26 '10 at 2:51

Reputation and badges are both meant to describe one action contributing to the community.

Every time you do something which contributes to the community, the system is designed to reward you. (No, it's not perfect.)

First, on the topic of users who ask too many questions.

As has been mentioned over and over. This is a question and answer site. Some users (most users) provide many times more answers than questions. They are generating content by providing answers.

Some users ask many more questions (hundreds in your case.) These users are generating content by providing questions. Even if they never accept a response, the users who answer those questions are still able to gain reputation by providing thoughtful and insightful answers.

So, basically, we never want to punish users for asking questions, because that is what those users have to offer the community. That isn't quite what we are talking about here, but it is a foundation.

The more important part that we are concerned with is the issue of gaining reputation for marking something as an answer. As I prefaced at the very beginning, reputation is designed to positively encourage good behaviour. The badges are all designed that way too.

Community concerns caused the SO team to create the "Acceptance Rate" which is actually a negative tool that is designed to shame people into good behaviour, but that is not the norm. Normally, StackOverflow is designed to help you integrate by rewarding positive behaviours.

Marking an answer as accepted is positive behaviour. You are contributing to the community by selecting the specific answer which solved the issue. Arguably it could be claimed that this contribution is worth more than a mere 2 points. It is the quintessential purpose of the site to have questions and answers. Without marking something as "the" answer, we have questions and a stack of suggestions.

Bottom Line, Marking something as accepted is important, and it is worth incentivizing in general and I have no problem if someone is gaining reputation this way. It is also incredibly difficult to gain that much reputation the way this user has.

This answer of mine has a few links which are also related somewhat.

share|improve this answer

Your math is wrong.

If you count up the votes on his questions, up to the 206th question have at least 1 vote. 206*5 rep = 1030 rep

75 questions have at least two votes. 75*5 rep = 370 rep

30 question have at least three votes. 30*5 rep = 150 rep

15 questions have at least 4 votes. 15*5 rep = 75

10 questions have at least 5 votes. 10*5 rep = 50 rep

9 have at least 6 votes. 9*5 rep = 45 rep

And the remaining 4 questions have an additional 14 votes between them:

14*5 rep = 70 rep

So the total amount of this user's reputation gained from votes on questions is: 1790

Or 98.8% of his current total reputation.

I'm guessing he didn't actually accept answers to a whole ton of questions, or else didn't even get answers to them. That and a great deal of the accepted answer reputation was probably negated by down voted questions.

EDIT: Wait a second, something weird is going on with this guy's reputation. Because 549*2 is 1098. I checked nearly every page and there was almost never more than one question with out an accepted answer. I looked at his down votes and there couldn't have been more than a couple hundred rep lost to down votes. There were only 30 or so questions with down votes at all and only the last handful had more than one.

So he should have gained 1790 from questions, 1098 from accepted answers, only lost a couple hundred to downvoting, but only had 1811. Unless he's posted a couple of questions that were epically down voted and then deleted - where'd the reputation go?

share|improve this answer
2  
He has assigned several bounties, at least 900 reputation worth by what I've divined on his reputation graph (50, 150, 200, and 500). There is also a -100 on a deleted question that may be a flag-based deletion. So it looks like he's big on bounty assignment, which explains why his reputation is much lower than calculated. –  Grace Note Jun 25 '10 at 16:59

You must log in to answer this question.

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