What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 130 Stack Exchange communities.

I see that accept rate has now been removed from display on questions. I think this is not a great solution for reasons that waffles explained well. Take it away waffles!


On one hand accept rate bothers me a lot, it is a one-dimensional metric that does not really give me enough information about how upstanding a user is in the community. It is used frequently to bully users and leaves a horrible taste.

On the other hand, introducing accept rates heavily increased the amount of accepted answers in the system. Which in turn helps close loops and motivate the community.

That said, I do not think it would be beneficial to simply retire accept rate if there is no replacement offered.

When we introduced the accept rate stuff we were careful not to add too much of a value statement. On hover we do not say "horrible user, never accepts answers".

Going forward I would much prefer that we retired accept rate in favor of a more general metric that covers a variety of "citizenship" metrics that do not result in rep.

  • Does the user vote?
  • Does the user accept answers?
  • Does the user answer questions?
  • Does the user edit or suggest edits on questions?
  • Does the user flag stuff?

I don't know, perhaps we should go the other way and show a tagline of honor for the top N percent of users.


Rather than removing accept rate, I propose replacing it with a citizenship percentage or level, based on the above criteria. This would be a powerful motivator for better sites, and gives much more useful context for the question.

It is not that displaying accept rate was bad, so much as incomplete. I always had plans to circle back and improve it to be a citizenship level.

Displaying this metric on questions might help participants think of the game at a bit higher level. What is more useful to the longer term health of a community: a single ok question, or an engaged community member who assists and participates -- as a citizen, not just another drive by hit and run?

share|improve this question
19  
I asked it before when you mentioned this, but I'll ask it again: "What is the desired effect of it. Say we have a broader metric visible to others. What should I do, or how should my behaviour change if I stumble upon someone who is not acting responsibly? Other than the edits I already make or the guidance I already attempt to provide based on the actual content (the question) I see." –  Bart Jan 26 '13 at 18:02
5  
@bart well you could visit their user page manually and try to suss it out manually, but that puts the entire burden of effort on the answerers. Why not make it easier and quicker to see what kind of citizen is asking? Rep score does not measure some rather critical good behaviors. –  Jeff Atwood Jan 26 '13 at 18:05
85  
I still find that deciding whether or not to answer the question based on whether the question itself is high quality gives me a better experience than using accept rate, to be honest. I certainly appreciate users who will make edits, vote etc - but a well-written question is likely to be of more use to others in the future. –  Jon Skeet Jan 26 '13 at 18:06
22  
What do I care what kind of citizen is asking anything? If the question is good, it's good. If the question is utter crap, it's crap. That's all the info I need. If I can answer a good question and it's helpful to the wider community (as judged by the upvotes I received) then my job is done. –  Bart Jan 26 '13 at 18:06
20  
@bart and your argument also applies for suppressing reputation scores. We display rep because it motivates positive behavior, do we not? –  Jeff Atwood Jan 26 '13 at 18:07
10  
@bart how does it hurt to have additional useful context for the question? And doesn't answering questions from those users who never give back, encourage them to continue to selfishly absorb answers without voting, accepting, flagging, contributing answers of their own, etc? –  Jeff Atwood Jan 26 '13 at 18:11
9  
Useful context how? I fail to see how it's useful. In general users are playing the game just fine. And isn't one of the requirements for questions that they have to be of benefit to he wider community?Then who cares about the user asking the question? Their acceptance is a mere 15 rep and the community will far outweigh that through their upvoting. I'm not arguing answers shouldn't be accepted, but I don't expect a big downfall of the system happening either. –  Bart Jan 26 '13 at 18:14
12  
@JeffAtwood: So as Bart says, encourage behind the scenes - I don't think people aspire to a high accept rate (or that they'd do anything similar with a "citizenship level"). Instead, it's just been used as an excuse for some users to beat up on others - which does little to help engage a user. –  Jon Skeet Jan 26 '13 at 18:25
3  
I agree with Waffles; acceptance (as a concept and an action) is very important to SE. However, doesn't this suggestion just propagate the problems that existed with the accept rate without even addressing the root cause for its existence; getting people to accept answers? I don't really see how providing a different metric that can be used to hector people or have people ignore questions encourages engagement or new users to stay on the sites. I would be very interested in a method of increasing the acceptance of answers without pressurising the OP to do so. –  ben is uǝq backwards Jan 26 '13 at 18:27
39  
"Why not make it easier and quicker to see what kind of citizen is asking?" - Why do we care? The primary focus is on the quality of the post, not on the user who posted it. –  casperOne Jan 26 '13 at 18:39
7  
@JeffAtwood For Stack Overflow, I'm showing around 37% of unclosed, unaccepted answer questions. Is it that big of a problem? Have we shown that people who accept answers are that much more engaged in a community? I share the goal of wanting longer-term health for SO, but you're not showing me how this particular issue (low accept rates) is a detriment at worst or that these users are neutral in other activities at best. In other words, show me that this is a problem for the site long term (or at least, lay out the thinking if you don't have data to support it). –  casperOne Jan 26 '13 at 18:49
7  
I don't see how a higher-dimensional metric would change that. Sure, the exact flaw is more difficult to pinpoint, but then again the "increase your accept rate" comments didn't involve a whole lot of investigative skill either. "Increase your citizenship level" it will be. –  Bart Jan 26 '13 at 19:39
9  
Related: meta.stackexchange.com/questions/103247/… - a citizenship level would also help beta sites see how active their community really is, because a 200+ rep counter doesn't cut it. –  Troyen Jan 26 '13 at 21:20
19  
The people who care only about the question are wonderful, selfless people. I am less noble; my main motivator is helping nice people. Not being thanked (and a "thank you" comment means more than accepting my answer, but either works) or knowing in advance that someone is not inclined to niceness is demotivating. It's selfish, but that is how I am configured. If you want to encourage such selfishness, a citizenship metric could help; otherwise, it's a bad idea. –  Jordan Gray Jan 27 '13 at 14:28
4  
@Jordan: well, I suspect a lot of participation here is driven by a sort of enlightened self-interest. Even the folks who don't care about reputation often have some reason other than pure altruism for answering, editing, etc. The trick is to encourage selfishness that just so happens to manifest itself in actions that benefit others as well as the person making those actions - a "citizen score" could work toward this end. –  Shog9 Jan 27 '13 at 20:21

26 Answers 26

I like the idea of citizenship, but I don't think it should be in lieu of the accept rate. Let's close the accept rate chapter for now; interested users can always use either this script or use the accept_rate entry via the API.

I would much rather see the rep based privileges be replaced with an effort based privilege, thus leading to different "citizenship levels". There are several high rep users who absolutely do not deserve most of the privileges they have right now (a recent example), and there are several low rep users who do an exemplary job of assisting in janitorial work, but don't gain access to the tools, because they're not motivated primarily by reputation.

This was also the central theme behind my previous request to roll out the privileges based on effort. One positive change that came out of that was in making the 5k privilege more substantial. I feel that there is potential to go further than that.

share|improve this answer
8  
Displaying a nuanced, complex citizenship level would be the first serious step in this direction. And you are right, equating "general competence as a programmer" (rep) with "willingness to be an active citizen of the community" is... Not great. It generally works and is mostly good enough most of the time (or most of SE would be broken) but it is not really correct. This is a design mistake -- albeit thankfully not a critical one -- I hope not to repeat in the next system I design. –  Jeff Atwood Jan 26 '13 at 18:41
2  
That said your answer is not really an answer to my question but a feature req. –  Jeff Atwood Jan 26 '13 at 18:43
11  
Well, if we're being pedantic, your question wasn't really a question, but rather a long statement :) You're proposing a feature request based on a loose concept of "citizenship" and I'm just steering it towards a slightly more focussed direction. As I said, I do like the idea and I don't see the need to open a new request only to slightly deviate from what's already here. It's better to keep the discussion here and let the community voice their opinions. –  Lorem Ipsum Jan 26 '13 at 18:51
    
@JeffAtwood, that is called "approximate with the scanty, noisy data at hand". A design mistake is something quite different. ;-) –  vonbrand Apr 11 '13 at 18:40

While I liked the idea of a citizenship rating on first glance when I read about it, after thinking a bit more about it I have two problems with it:

I don't think such a single value can effectively express the value this user adds to the community. There are many different ways to contribute positively to an SE site: asking good questions, providing good answers, voting (but not indiscriminately), making good edits, flagging appropriately, participating on meta and more. The actual value of the contributions of the user depend much more on the quality of each action than the quantity, and that is something the system doesn't really know.

With the exception of some extreme cases, it doesn't really matter to me who asks the question, it only matters how this specific question is written. If the question doesn't show any effort, I might decide not to answer. I would not decide not to answer if a user doesn't edit much, and even if he doesn't vote at all it would not be that important, as the asker is not the only one who votes on answers. In the cases where user behaviour would stop me from answering it is usually evident from the bad question itself, and such users are also typically caught by the question ban.

And as a last point, I suspect that this would only lead to replacing the "please improve your accept rate" comment by "please improve your citizenship", "please vote for other user's posts", "please help by editing posts" and similar noisy comments.

share|improve this answer
    
I agree completely with your answer but disagree slightly with your statement that the system can't know. The system can know by comparing your actions with those of other "good" citizens; all you need is a reasonable starting point. I also think it would lead to a horrible positive feedback loop, which isn't good as it encourages no deviation from the behaviours deemed to be acceptable by those initially given the stamp of a "good" citizen. –  ben is uǝq backwards Jan 26 '13 at 18:59
3  
True, but a single reputation score can't really measure your competence as a programmer, either. Doesn't stop us from trying, and it is still generally a useful measurement despite its imperfectness. Accept rate is, unlike rep, a really narrow measurement of a very specific action. Too narrow and specific is the problem... –  Jeff Atwood Jan 26 '13 at 19:00

Let's start with your question:

What is more useful to the longer term health of a community: a single ok question, or an engaged community member who assists and participates -- as a citizen, not just another drive by hit and run?

That is of course trivially simple to answer. I want a citizen. I want a user who asks good questions, gives good answers, upvotes and downvotes correctly, edits things into shape and....accepts the correct or helpful answers he receives. And to make this crystal clear, my objection to accept rate was not in contradiction with any of these things.

But the question becomes, does displaying a citizenship metric help in creating the positive effect of ultimately getting more involved users? And does it not suffer from similar downsides that became the downfall of the accept rate?

Let's for a moment assume we implement the metric. And I stumble upon a user who asks a question. What to do? Well, that's simple really, isn't it? It's a great question, so I'll answer it because I can. Community wins and the site has some additional great content for new visitors. Or the question is bad, in which case I can address that through voting, closure, editing or whatnot.

Strangely enough, at no point did I need the metric there. Before your edit you argued that it would be a

proper guide for "should I answer this question, was it asked by a fellow, engaged citizen?"

Ehm, who cares? Was it a good question? If so, then not answering it just because the user is not an engaged citizen would only end up doing harm, wouldn't it? The site would lose out on adding great content. Sure, I might not be guaranteed to get my 15 rep, but is that so much of a problem?

And if it's a poor question, how engaged the user is doesn't really matter either. It's a bad question. Present me the most engaged user of the site. If he asks a bad question, it's still a bad question. And I would respond to that exactly the same.

But what about the positive effect of a citizenship metric? Wouldn't users want to be displayed as a "good user"? Sure, those who already care would certainly care about the metric as well. But they are fine anyway. And those who only care about simply having their questions answered won't care a lot about a metric either.

And displaying any metric has another side-effect, similar to some of the problems with the accept rate. You are implicitly putting the policing of users into the hands of the community. And while I don't want to discredit all the nice users who were merely informative, there were also those who went overboard. I'm talking about the negative comments. The badgering.

Sure, a broader metric might make it harder to pinpoint the actual "problem" with the user, but if there is something "wrong" users will be quick to pick up on it and comment on it. And do we really want that? There were already measures in place to nuke accept rate comments with a single flag. So apparently we don't.

Engaged users will always be engaged. Users who simply don't care will never care. And those who don't know can be informed by the system. You don't have to leave that up to the community.

My proposal:

Implement your metric. Implement several metrics if you want. And if you realize something is amiss, have the system poke the user behind the scenes.

  • Tell them they might want to evaluate the answers they received.
  • Inform them that there are review tasks awaiting them if they've never even visited a queue.
  • Perhaps even lower their privileges. "You only ask in high volume, never accept anything, rarely get upvotes and don't contribute otherwise? Okay, less questions for you per week/month"

That I can see have a positive effect. A visible metric...not so much.

share|improve this answer
4  
Ok, so we should remove visible display of reputation score? Because at what point do you need to see a reputation score to decide if a question is worth answering? Everything above applies just as equally to display of reputation score, as it does to display of accept rate. Unless, as I've been saying, the problem is that accept rate is, unlike rep, an overly specific and narrow measurement.... –  Jeff Atwood Jan 26 '13 at 19:08
5  
They are entirely different things, as I have argued before. Never have I seen users negatively comment on low rep to the best of my recollection. Rep is an ultimately positive measure, especially with it being always 1 or higher. As someone once said, graph it and it will ultimately always go the right and up. A metric, if not 100% positive, is a negative statement to some degree. Do I need to see a reputation score? Mwah, not really. –  Bart Jan 26 '13 at 19:11
1  
and can you explain to me how a citizenship level could go down? Rep can technically go down as well, but does it? –  Jeff Atwood Jan 26 '13 at 19:16
3  
Rep goes from low to high. Not from leech to perfect. They are inherently different measurements and send completely distinct messages. I ultimately want to see the same goal achieved, but think the publicly visible metric is a rather negative way of getting there, if it has the desired effect at all. –  Bart Jan 26 '13 at 19:20
    
And does not citizenship go from "felon" to voter to community council to mayor to governor to president? How is that negative? (Well, unless you dislike politics.. But expecting a programmer rep score to capture willingness to be a community citizen is the deeper underlying issue isn't it?) –  Jeff Atwood Jan 26 '13 at 19:26
1  
As I said in my answer, users who care would already care without the metric. And they would subsequently have a great metric value. Users who don't, well...they don't. And users who don't know can be informed behind the scenes. Do you really think that you'll pull a significant group to the good side that otherwise would not have been there? Do you have anything to back that up with? I'm not saying you don't, but I would honestly be very interested in seeing it. It would make your point a whole lot stronger. –  Bart Jan 26 '13 at 19:29
    
I didn't have any proof that reputation score would work prior to implementing it, either. You tell me -- did it work? And a citizenship metric is far, far closer to rep score than a mindlessly simple, never intended to be the final step, display of a single accept rate percentage. –  Jeff Atwood Jan 26 '13 at 19:32
    
@JeffAtwood It's not so much of an "I want to see proof", rather than a "What do you base this on?". I'm honestly simply interested. Because I don't see a reason why it would have a significant effect. Still, it's a higher-dimensional metric than the flawed accept rate, I'll give you that. But I ultimately feel it's unnecessary and might suffer from the same negative community effects that accept rate did. –  Bart Jan 26 '13 at 19:37
    
@JeffAtwood A citizenship level can definitely go down: deface posts, be offensive, suggest (or make) bad edits, approve everything in /review, ... Not that consider that an argument against it. –  Gilles Jan 26 '13 at 21:12
1  
@giles does rep go below zero? And you start out with 1 rep, not zero. –  Jeff Atwood Jan 27 '13 at 20:01

If the end goal is to increase the amount of accepted answers, the straightforward way of accomplishing that is to award points for accepting answers.

share|improve this answer
18  
...like the 2 rep you get? –  Bart Jan 26 '13 at 19:05
    
Make it more like 20. –  Apocalisp Jan 26 '13 at 20:51
1  
@Apocalisp - It has to be between 2 and 4 . –  Adel Jan 26 '13 at 21:07
2  
@Adel Interesting, why the upper cap of 4? –  Manishearth Jan 27 '13 at 9:12
1  
@Manishearth I guess to have less of an influence than a question upvote. –  Bart Jan 27 '13 at 9:16
    
@Bart - yes, that is it. –  Adel Jan 27 '13 at 16:14
1  
@Apocalisp 20 rep is more than the answerer receives for solving the problem in the question. –  Asad Feb 6 '13 at 4:21

When we introduced the accept rate stuff we were careful not to add too much of a value statement.

100% == complete. < 100% == incomplete. Without any on-page guidance as to what was acceptable, this is how the number was actually interpreted.

While I like the idea of a general-purpose "citizenship" metric in theory, watering it down with stuff like flagging just makes it harder to interpret. The best idea we could come up with ended up being sort of a parallel reputation system that increased whenever you did something positive on the site - while interesting, I'm not convinced this would do much to encourage specific forms interaction the way accept rate did.

I don't know, perhaps we should go the other way and show a tagline of honor for the top N percent of users.

I like this idea more; fits with the old "praise in public, criticize in private" advice.

What if, when posting a question, folks who accept answers to most of their questions (say, over 70%) got a little "responsive asker" indicator below their name?

We could even expand this to other metrics without watering anything down:

  • Neophyte editors whose recent edits have been 100% approved get an indicator of this in the suggested edits queue.

  • Flaggers whose recent flags have been helpful get an indicator in the flag queues.

  • Answerers whose past answers are seen as helpful (up-voted / accepted) get an indicator on their new answers. (Kinda uneasy about this though; could unfairly skew voting)

Displaying only positive indicators in situations where they actually matter gives folks something to work toward, without encouraging public witch-hunts.

Chiding in private

Another suggestion for modifying the accept rate display was to display it only to the author - this avoids some of the public shaming, without actually making this obvious to the asker. If this number also linked directly to a list of questions that lacked an accepted answer, it'd also give them specific action items to improve that number, without requiring them to dig up documentation on what it actually pertained to.

Frankly, doing this alone might be enough to provide most of the benefits of accept rate, without the noisy comment harassment.

share|improve this answer
1  
Answerers whose past answers are seen as helpful (up-voted / accepted) sounds like it would just end up being another @JonSkeet badge, or reward answerers of sock-sorting questions. –  CodeGnome Jan 26 '13 at 19:08
    
Jon's hardly the only person to consistently post helpful answers, @Code. –  Shog9 Jan 26 '13 at 19:13
4  
I think you are underestimating the ability of rep to capture a lot of positive behaviors in one bucket: posting great questions? Rep. Posting great answers? Rep. Accepting answers? Rep. Suggesting edits? Rep. Granted you don't want dozens of things there but citizenship could fold in the top 3 most important without breaking a sweat. Because rep certainly does... –  Jeff Atwood Jan 26 '13 at 19:19
5  
If you're not interested in encouraging all behaviors, then rep is fine, @Jeff. Some folks only ask, others only answer, and that's ok. If you want the folks who ask to also accept, then displaying a "top citizen" score for someone who asks, never accepts, but flags a lot isn't gonna help. I think the "citizen score" idea is interesting, simply because it recognizes that there's a dimension of positive participation that isn't (and shouldn't) be captured by the current reputation score - but if the goal is encouraging one specific behavior looking for a silver bullet is just a distraction. –  Shog9 Jan 26 '13 at 19:24
4  
Where did I say I wanted to encourage one specific behavior? I want to replace (rather than mindlessly delete) one useful-but-overly-narrow visible metric with a broader and much more useful one. –  Jeff Atwood Jan 26 '13 at 19:29
8  
"Without any on-page guidance as to what was acceptable" - Technically this was there in the form of range colouring, but the perfectly acceptable "middle ground grey" was probably not the most intuitive colour choice for the common range. It also required you to know that the colour was something that changed in the first place, although I suppose the tooltip could have hinted at that if it wanted. The larger issue was that the system couldn't know when you had a good reason for not accepting certain answers, so people opted to assume the worst regardless. –  Tim Stone Jan 26 '13 at 19:56
2  
+1 for 'responsive asker'. It could also be expanded if we could find a metric evaluating responsiveness to requests for clarification in comments, so that users who ask a question and ignore follow-ups are penalized yet those who participate in the discussion about their question and just never find an answer that helps them are able to get some credit toward responsiveness. –  Carl Veazey Jan 26 '13 at 20:27
    
+1 for "whatever display" links the asker directly to a page that allows them to improve whatever is bringing the number (however it's calculated) up. –  Chase Florell Jan 26 '13 at 20:41
3  
I'm starting to really like the idea of rewarding good behavior without drawing public attention to bad behavior. That way, when I get to a question, even if the user has a 0% accept rate, it's really none of my business at this very moment and I am only here to answer. But if the user is over 70% , it's possibly just a little positive boost. –  Adel Jan 26 '13 at 21:53
    
Does "I got flagged" count negatively against me for my citizen-score? AKA if so then it would encourage the use of flagging to indicate poor community participation when it's seen. –  jcolebrand Jan 26 '13 at 22:04
3  
@tim the design was not quite right in several dimensions. Citizenship is hugely important. How many answers you accept is not that important. –  Jeff Atwood Jan 27 '13 at 19:57
1  
What if, when posting a question, folks who accept answers to most of their questions (say, over 70%) got a little "responsive asker" indicator below their name? This is going to have the exact same effect as you say displaying the accept rate did: 100% == complete. < 100% == incomplete. Except that 100% will be replaced with 70%. Instead of "improve your accept rate" comments, users will simply make "go earn responsive asker" comments. –  Esoteric Screen Name Jan 28 '13 at 15:04
4  
Unless there's a corresponding "unresponsive asker" display, I'm skeptical of that, @Esoteric. –  Shog9 Jan 28 '13 at 20:44
4  
I would not want to see "responsive asker." People will figure it out and start harrassing people again. Most of everything else, I can agree with. I've long advocated the system hammering users with notices about being a better citizen. We don't need users hammering other users. –  Anthony Pegram Jan 29 '13 at 16:41
    
You could always turn accept rate into color shades with a very fine scale. This way, the difference between 95% and 100% would be imperceptible, but the difference between 10% and 100% would really stand out. I've used this technique with some of my customers' dashboards as a way of steering them away from placing too much weight on arbitrary thresholds and more on the overall "sense". "Greenish" = ok, "redish" = not ok. –  JDB Feb 19 '13 at 23:18

Two observations come to mind:

  1. It appears as though active users that have higher accept rates, have them because the person asking the question will take the first response that unblocks them from the problem. So frequently the accepted answer tends to be very short and not frequently 100% correct.
  2. The fully correct answer eventually boils to the top, but it is never marked as accepted. This could be months to years later when an authority on the subject - not someone terribly active - responds to a question. And then add the time it takes for the answer to be up voted.

So there are three problems as I see it that need to be addressed:

  1. Not all answers were created equal. Partial answers are common. Authoritative answers are more rare. The person asking the question rarely goes back to mark the authoritative answer as the accepted answer.
  2. Not all accepted answers are complete - they're only good enough to get a user unblocked with respect to their current problem. At the time the person asking the question gets their first answer, they don't have any perspective for what the realm of possible answers could be.
  3. Not all people asking questions have the wherewithal to understand a correct answer vs. just the answer that will get them unblocked. Similarly, the user asking the question may not care about the health of the community.

The accepted rate has a few problems because it only rewards two of the three roles involved in answering and asking questions:

  1. Asks questions - currently rewarded when marking a question as accepted
  2. Answer a question quickly - currently rewarded when marking a question as accepted
  3. Answer a question authoritatively after an answer has been accepted - And there's the rub

So here are my thoughts re: how to solve this particular problem, with the above as my frame of mind:

  • Allow multiple answers to be marked as accepted, either by the community ("candidate answer") or the asker. "This is an answer, but not the answer." Up votes are free, but marking an answer as a "candidate answer" costs rep, like asking a question with a bounty.
  • Setup a reminder to nag an asker to review partially accepted answers. If the reviewer doesn't mark the candidate question as a "viable answer," the users get their bounty back (an "abstain"). If the user has their question accepted, they get 2x the rep back, plus a "viable answer" tally. If an asker rejects a question, the user looses their rep.
  • A different member of the community (not necessarily the respondent) can submit a response using their reputation.
  • Interacting with "candidate questions" gives the asker some kind of rep bump for doing the right thing.
  • Of "viable answers", there can be an "authoritative answer" flag that can be shifted around between "candidate answers."
  • Askers should have the ability to delegate responsibility for what's a viable vs authoritative to the community.

re: that last point: maybe add a points per tag system. "I have N points for the PostgreSQL tag, therefore I can interact with questions tagged PostgreSQL, but I can't necessarily answer questions regarding Flash." It seems that this would solve a few problems.

As for the UI, it'd be nice if there was a mouse-over for a user's avatar that would show their authoritative/viable accept rates and the relevant tags for those answers (or a metagroup - database vs CSS vs ...)

Bottom line: the citizenship/activity level is currently already scored using rep. Accept rate, however, needs an extra dimension - hopefully the above is good food for thought.

Cheers.

share|improve this answer
1  
To the end of your post you talk about something I wish we did. We should, in my opinion, show cumulative tag-rep in addition to total rep on specific posts. –  jcolebrand Jan 26 '13 at 21:04
4  
Rep doesn't always measure activity/citizenship level. Not really an issue on SO, but on beta sites there are people with high rep that haven't been around in months and people with low rep who don't answer questions, but constantly help guide new users into using the site in the hope they stick around. Accept rate didn't capture that either, but "citizenship level" is a broader metric than just checking a green checkmark. –  Troyen Jan 26 '13 at 21:18
1  
Troyen you raise an excellent point. Leaving guidepost comments is much more helpful to the community whit large sometimes. –  jcolebrand Jan 26 '13 at 21:27
    
A community accepted answer is interesting - perhaps, the ability to flag an answer as more complete, which would allow some collection of moderators to override the currently accepted answer, in the case of questioner inactivity? –  Hannele Feb 7 '13 at 15:38

We already do kinda sorta have a mechanism in place to measure your "level of citizenship": badges. All of the sample actions you've mentioned are covered by badges.

  • Does the user vote?
    • Supporter
    • Critic
    • Suffrage
    • Vox Populi
    • Civic duty
    • Sportmanship
    • Electorate
  • Does the user accept answers?
    • Scholar
  • Does the user answer questions?
    • Teacher
    • Self-learner
    • Nice answer
    • Necromancer
    • Revival
    • Enlightened
    • Good answer
    • Guru
    • Great answer
    • Populist
    • Reversal
    • Generalist
  • Does the user edit or suggest edits on questions?
    • Editor
    • Excavator
    • Archaeologist
    • Strunk & White
    • Copy editor
  • Does the user flag stuff?
    • Citizen Patrol
    • Deputy
    • Marshal

You may argue that badges don't do a good job of measuring citizenship, but:

  • They're already coded
  • They're already shown everywhere
  • They can be tweaked
  • They can be revoked

We don't need a new statistic, a new number or set of numbers to minmax or berate others over. We just need badges. Perhaps we need "better" badges, or "more balanced" badges, but I posit that the badge counter is already an okay way to estimate your involvement with the community in a deeper way than reputation does.


If you think just the badge count doesn't cut it (sigh), you can be cruel to Jin and give... badges for badges. For example, take flagging. You do Citizen Patrol and get a first-level flagging badge badge: dashed flag. For Deputy, you upgrade to blank flag. For Marshal you get colored flag. You make the flag to end all flags and you get glowing flag, then my shitty drawing makes your eyes burn and you ragequit Stack Exchange. This looks unnecessarily... PHPForums-y to me, however. I think that badge counts alone can do it.

share|improve this answer
12  
Badges are extremely broad, too broad to be useful for much other than encouraging a huge smorgasbord of behaviors ranging from "keep trying!" (Tumbleweed) to "you read the FAQ" (analyst) –  Jeff Atwood Jan 26 '13 at 19:46
6  
@Jeff «Perhaps we need "better" badges, or "more balanced" badges» –  badp Jan 26 '13 at 19:53
2  
The Badge Approach is a good one. Maybe simply replacing the "Accept Rate" bit with a citizenship info BASED on badge acquazition. And that info might differ based on the tags of a question. IE: I'd have a pretty good citizenship badge for .NET questions and not a very good one for PHP ones ... unlike @CodingHorror's PHP involvement –  Chase Florell Jan 26 '13 at 19:57
1  
Also, it wouldn't encompass ALL badges per se, but I think that the ones listed above along with individual tag badges with some magic in the back could really make for a good general citizenship representation. –  Chase Florell Jan 26 '13 at 20:03
5  
I agree with citizenship being determined by some kind of aggregate of the existing badges. That's not only the parsimonious (i.e., elegant for Apple fanboys) thing to do, but it also adds legitimacy to badges which are currently kind of a bastard child SE metric. –  tim peterson Jan 26 '13 at 20:10
    
Or you could make awesome badges more prominent! –  JDB Feb 19 '13 at 23:22
    
I would say, use a point system, assign a number of points for each of the above behaviors mentioned by badp, decide on appropriate point thresholds, and you have a citizen badge (or two or three). –  Tom Au Sep 29 '13 at 1:46

You raise excellent points. The only thing that troubles me is that we might, by adding a new metric system, find ourselves at the same predicament - people might start discriminating based on the metric.

Bart said:

But as long as you're displaying the accept rate, people will comment on it, criticize users for it and possibly reconsider answering because of it. And while I'm of the opinion that a low accept rate should not stop you from answering (we're here to make the internet better after all) and that the possible reputation gain (or lack thereof) should not factor in either, I can't blame users for considering it. It's in their face after all. And we're only people

Let's face it, some of us have the luxury to be good citizens(voting, accepting answers, editing questions, etc.), while others just need an answer .. if we're aiming to help the greatest number, shouldn't we remove any potential barriers to someone getting helped?

I think we should reward good users, yes... but just in a way where it wouldn't interfere with the simple case of a question needing an answer.

share|improve this answer
7  
+1; I know many people who just get on SO to get help, but never bother to give help. Sure, it would be preferred if they would also answer questions, but is this really a problem? I don’t think so. After all, the sites are there to help. And we don’t know if they won’t return at some point to give help as well; but it is important to make their first impressions good and a metric that just leads to other people pissing them off is never a good idea. –  poke Jan 26 '13 at 20:16
2  
If this was true, I would expect to see a lot of "your reputation is too low, please raise your reputation level!" comments on questions. But we don't... so my working hypothesis is that the measurement we are showing with accept rate is wrong. It was never intended to be the final iteration, citizenship metric was always in the cards. –  Jeff Atwood Jan 26 '13 at 20:34
    
@JeffAtwood - Well, I see what you're saying. Citizenship, if implemented wisely , may bring more "community" into it. My only issue is that I think it should be a very coarse grained metric on the question screen - I'm thinking 3-states(so people don't go too far in commenting): low-Citizenship, good-Citizenship, and Excellent-Citizenship .. however - Low-Citizenship should be hidden-by-default(perhaps a mouse-over view). Kinda what Shog said - praise in public, criticize in private. –  Adel Jan 31 '13 at 16:17

I was unhappy with the removal of the accept rate.

IMO responsible SO citizens used it responsibly; I know I tried very hard to, particularly after participating for awhile. I left my share of "accept rate" comments; eventually I changed them to be (a) more explanatory, and (b) deliberately shifting the blame on to answerers, not askers.

I don't know how "citizenry" would be calculated. I think I'm a mostly-good citizen, although I tend to be a douchebag sometimes, especially in a few known situations. That should count against me, but evaluating douche-baggery is pretty fuzzy.

I'd like to see something that takes in to account how a user interacts with SO overall.

I don't know how it could be implemented. IMO the accept rate is actually reasonable, but the use of the accept rate is where things tended to fall over. Flagging of "just" accept-rate comments took care of a lot of that (IMO).

share|improve this answer

At a first glance, we have a single-valued citizenship metric. It's called 'reputation'. We have a multidimensional citizenship assessment. It's called 'badges'. Given that the OP invented them, it is somewhat humorous that he would pose this question.

It seems to me that the problem comes in assessing people who don't have much rep or badge real estate. Is this because they are deadbeats? Or is it because they haven't been around very long?

However, I suspect that any metric will have this problem, unless it has a time axis, and, even then, it's dubious. If we say, 'rate of rep/badge gain', we penalize well-behaved people who stop in when they have a question or answer, and otherwise have lives.

One possible direction to go here is to observe that we measure signals of goodness but not badness (give or take downvotes) and so we don't get much discrimination between 'good but casual' and 'icky'. This might lead to consideration of making downvotes much more punishing. Thus, people who participate lightly will tend to have more rep than poor citizens.

The other way to look at it is to focus on the banning system. If the system is doing a good job of banning the truly awful, is it really that important to draw fine distinctions between the others?

I generally don't need rep, badges, or accept rate to get a good idea of what sort of OP I am dealing with. I simply apply ask the following questions -- and weigh the answers to all of them:

  1. Have they registered and selected a user name? A few userXXXXs are worthwhile, but most are not.
  2. If they have a user name, is it boastful, flakey, or otherwise cringe-inducing?
  3. Does the question have an informative title? If not, chances are that the experience of reading the body will be a further letdown.
  4. Is the question 'plz send me the codez', or a rant, or founded on counterfactual premises?

Somewhat like a questionnaire in a fluffy magazine, one can add up the results of these, and pretty easily make a decision whether to spend time. Thus, if I see a question that has bad premises but looks otherwise good, I might invest the time in comments or an answer to try to comb the thing out. If, on the other hand, I see some of the other hallmarks of badness, I won't do more than downvote or vote to close or both.

Finally, I am not really excited about the concept of acceptance as it plays out. Far too many questions get bad answers accepted, either due to the OP wandering away after the first answer, or due to the OP not knowing enough to judge. You can write 'the check only means that it helped the OP' as many times as you like, but readers will still view it as some sort of official endorsement.

share|improve this answer
    
How is having more information a bad thing? Whereas it is clear that accepted answer alone was too narrow and as you noted -- not that important, since an accept is just a thank you. But the community does not see it that way. Anyway, focusing on a narrow metric, and one that isn't that important (and arguably widely misunderstood) is clearly incorrect -- but it should be improved not deleted. –  Jeff Atwood Jan 27 '13 at 20:05
1  
My argument, such as it is, is that people reading questions already have plenty of information. However, I've written my piece here and I'm not interested in defending it; if you, the team, or the community don't agree, you don't agree. –  Rosinante Jan 28 '13 at 17:54
    
Even Jon Skeet answers questions that he shouldn't. Our reputation system currently heavily encourages answering even if the question is bad or inappropriate, even if a particular question is ultimately deeply unhealthy for the site. In all cases, whatever the outcome, answering is very likely to generate reputation and up your rep score. I propose showing an improved metric on questions -- beyond accept, complementary to rep -- that helps balance and counteract this. –  Jeff Atwood Jan 28 '13 at 21:24

I don't mind having this, but it should not be displayed on questions. The quality of a question can be determined by reading it. Knowing more about the asker should not change your vote -- Stack Exchange is about posts, not users.

A user's rep/stats may help when determining the correctness of an answer. But for a question... there really isn't any positive change brought forth by it. It all boils down to this:

What do you expect a user to use this newfound knowledge about the asker for?

I myself don't see this newfound knowledge being used for good.


Instead of this, why not display some metric like this publicly on the user's profile? Or, if you feel that it may lead to a lot of complaining (similar to the whining brought forth by the public display of flag weight), then display it privately.

share|improve this answer
4  
Then we should also suppress reputation score display on questions. Because why would you need to see someone's reputation score to evaluate the question? –  Jeff Atwood Jan 27 '13 at 19:50
1  
@JeffAtwood: I never said I agree with rep being displayed on qs :) –  Manishearth Jan 28 '13 at 5:37
1  
@man then these might not be the droids you are looking for... –  Jeff Atwood Jan 28 '13 at 6:05
1  
The point he's making here is good. Deliberately shifting the focus of an expert from the question to the person that asked it, while on the other hand waving our hands around saying that it's not about the person it's about the content is questionable. Could it really be done in a way where comments complaining about something lacking are speaking directly to information missing from the question and not something else about the user? I tend to think not. I like every aspect of the idea besides showing it on questions unless it's obvious that we learned something by the way it's implemented. –  Tim Post Jan 28 '13 at 12:10
    
@tim every single word of that applies to displaying rep on questions. Every. Single. Word. I got a crazy idea, how about we use some science and try something different? Rather than just deleting stuff out of fear. (Granted, there was lots of data to support the idea that showing accept was problematic. But that's just it: science!) –  Jeff Atwood Jan 28 '13 at 21:20

This sounds good at first: a citizenship "score" could become an integral part of your profile, maybe on the same level with reputation. It could be a way to award those who put a lot of time and effort in cleanup / quality control in their tags instead of just accumulating rep.

However, looking at how even the prospect of getting a fricking badge leads to widespread abuse of the system (voting, editing), it's a very real danger that this would become just another number you can increase through harmful actions like random voting, or approving BS edits.

I'm sure Rosinante's approach is shared by many community members: we often already check out users before we answer a question. How invested is this user in the community? Did they bother to register a user name? What kind of contributions do they make? Do they answer questions as well? I tend to be much more willing to help fix an imperfect question when the OP has a track record of answering questions, or shows a real interest in the community. How about instead of creating another metric, making the existing ones easier for us to look up in the user's profile?

Maybe even in a separate "citizenship" tab, designed in infographic style.

We could display....

Definitely:

  • Successful close votes (ie. votes cast that lead to a closing)
  • Helpful flags
  • Question / Answer ratio, or maybe even better: answer upvotes
  • Number of comments
  • Some of the "hard" badges, maybe those that are displayed in the pre-election stats
  • Meta participation

Maybe (I'm unsure about these because they are so easy to game):

  • Voting
  • Edits

That would show us all we need, but without the problems of a new metric.

share|improve this answer
    
Interesting view on the potential abuse. Didn't think of that. And perhaps "number of upvoted comments" instead? (And is Meta participation really that positive of a thing? :p ) –  Bart Jan 27 '13 at 12:07
8  
+1, but just to provide another point of view, I check out nothing before providing an answer: if I got one, I give it. Period (and I edit, and upvote the question). I was fond of the accept rate only because it could be an incentive, later down the road, for the user to finally accept the answer. –  VonC Jan 27 '13 at 12:38
    
Similar logic applies to rep. Why display rep? It might be abused. Hide rep and keep it on the users profile in handy info graphic form... –  Jeff Atwood Jan 27 '13 at 19:51
1  
@Jeff well, fair point. Generally, a citizenship metric really would be cool, especially one that is on par with rep in terms of visibility ("citizenship" could e.g. replace badges in the flair)... it would bring recognition to those who do a lot of janitorial work. But it needs to be super certain that stuff like this never happens again –  Pëkka Jan 27 '13 at 20:00
    
Baseball cards put all the juicy stats on the back. And don't try to lump the ones that make for a good pitcher in with the ones that make for a good batter... I kinda like the idea of a "player stats" screen somewhere off profile-way. –  Shog9 Jan 28 '13 at 4:49
2  
@shog9 I'm pretty sure when a batter steps up to bat, the TV screen displays some relevant stats about them to determine whether or not they're likely to hit the ball.. broadcastengineering.com/products/… –  Jeff Atwood Jan 28 '13 at 8:15
4  
If a citizenship metric is established, it should exclude voting at the least and maybe even editing. Having a number in my profile that I can make go up by applying random clicks on vote buttons and making microscopic edits is a recipe for disaster. Peer-reviewed contributions like flags and successful closes (well, peer-agreed in that case) will work so much better. @Jeff –  Pëkka Jan 28 '13 at 9:52
    
@Jeff, why display rep? Well, to be honest, on a question I'm not sure there's much point. Admittedly I'm more of an asker than an answerer, but I tend to look at rep on answers more than on questions. Apart from here on meta, where rep is a handy approximative marker for 'newbie or not'. –  Benjol Feb 6 '13 at 11:39
    
I am definitely more likely to help a user with a decent accept rate, or put more in to the answer. At the very least, seeing that a user accepts answers shows that they'll even SEE my answer. A 0% accept rate leads me to conclude that they cast a wide net in the heat of problem-solving, then resolved their issue through some other mechanism. By the time they see my answer (if ever), it is useless to them -- they're back casting a new net on a new problem. Depending on how broad their question was, I may have literally wasted my time. I hate feeling like I am wasting time. –  Chris Feb 6 '13 at 14:53

I would be fine with some nebulous klout-like calculation of global SE "kudos" that would apply across all your activities on the network, and would hopefully be constantly made more sophisticated in terms of measuring in a rich, deep and vivid way, of your role within the community -- it might be related to the quality of particular activities and contributions (but would be weighted towards those actions and contributions that are for whatever reason "most wanted" by the community.)

To some degree reputation accomplishes half of this -- it can function more like a global reputation exchange, which gives a certain kind of raw quantitative scale. But the problem is that it becomes purely vertical. I'd definitely welcome a recognition of more "horizontal" civic SE effort -- maybe something like a 0-1k "karma credit" score.

share|improve this answer
1  
Make it complex, and nobody will understand what it is all about. Make it simple, and it won't really capture the meaning intended. And any "community kudos" will by force be a multidimensional measure (what about somebody who rarely answers, just comments? Never evaulates posts put is the author of the highest-ranking questions/answers? Just too complex, IMHO. Reputation is a very crude measure of community involvement, but easy to understand. Raw number of badges also says a bit... –  vonbrand Apr 12 '13 at 14:42

Not sure if revolutionary...or new way to flame users.

Here's why I feel this way.

We've demonized talking about the accept rate, and comments making reference to such were always considered inappropriate comments.

I'm playing devil's advocate here, so I want to challenge the merits of bringing yet another measure of a Stack Exchange participant, and if it will benefit or adversely affect Stack Exchange at large.

1. It's yet another way to measure participation.

We have badges, reputation, number of answers, number of questions, number of reviews...all of this information, while not readily available on a single screen, is a great measure of active participation on any SE site. Maybe this suggestion combines this information into one screen (heh), but that brings me to...

2. It's yet another way to flame users.

Remember that commenting about a user's accept rate was considered a no-no. What would having a "poor" citizenship rate earn a user?

Say, for example, we have some user117243*. They're new to Stack Exchange and participate on Stack Overflow and TeX.SE. They're fairly active in both - over 500 reputation on both sites, asked plenty of questions, answered their fair share, flagged some answers/comments, and participate on Meta. However, they don't always accept their answers, don't have many badges, and don't review or make peer edits. Are they a bad citizen? If so, why? They're one of the many people that actively participate on Stack Exchange at a level I personally feel comfortable with - general active participation.

Let's presume that they're stuck with a low "citizenship rate". They're going to get comments about it, they're going to get flamed for it, and that will actively discourage their participation on Stack Exchange, something that is a huge detriment to the system. That's already occurred with the previous measure, but I shouldn't be beating a dead horse.

*: Sorry to user117243; I pulled that number out of thin air. You're awesome, d00d!

3. It will feel incomplete on its first implementation.

There's lots of ways that someone measures valuable participation. Personally, I feel it comes not just from a collusion of badges and general participation, but also meaningful participation - I mean, I've been vocal about the way the reviews are done, and I feel that anyone that's a "good citizen" and participates in the review queues should be doing them well. I would personally feel that a measure of quality of reviews not being present in the system would render it "incomplete".

There's a lot more out there that could be thrown into the mix, too.

  • How often do they participate in peer edits?
  • Do they post on Meta?
  • Is the average score of their questions above 5 (minimum 10 questions)?
  • ...and so forth...

I'm not entirely sure or convinced that we can cover all the bases of "activity on this.

share|improve this answer
2  
The perfect is the enemy of the good. –  Jeff Atwood Jan 28 '13 at 6:04
4  
I don't think SO has a problem with participation. So what if we piss off a few hundred users who aren't contributing properly? There are plenty more where they came from. –  Lightness Races in Orbit Jan 28 '13 at 7:56
4  
@lightness I would argue it is our job to piss off some of the users. This is a system of fairly strict rules for a reason, to promote quality and produce strong signal. –  Jeff Atwood Jan 28 '13 at 8:17
7  
@JeffAtwood The perfect is the enemy of good proverb doesn't apply as a criticism of this answer. That proverb means if something is worth doing, it's worth doing well enough (and sometimes badly is well enough). If something is fundamentally impossible to do well enough, it's not worth doing. Also, even if this could potentially be done well after many tries, there's a basic ethics problem if the first try (and the 2nd? and the 3rd? ...) consists of thousands of knowingly wrong evaluations of how much/well people have contributed. –  Eliah Kagan Jan 28 '13 at 8:52
    
@JeffAtwood: I couldn't agree more. :) –  Lightness Races in Orbit Jan 28 '13 at 9:21
1  
In defense of the Citizenship metric, people used to pester askers over their accept rate because it is so specific. It's a specific action (or rather, inaction) that you can pester people about. I don't think this would happen with an accumulated metric like Citizenship - as Jeff pointed out elsewhere, you don't see "your rep is a bit low, I don't think I will answer this" type comments either –  Pëkka Jan 28 '13 at 9:48
    
@Pekka웃 The accept rate comments were often not based on a whole lot of investigative background either. "Value low. Must leave comment about value". I'm not sure this would change with another metric. –  Bart Jan 28 '13 at 9:53
    
@Bart but that metric would be much less specific. Commenting is socially acceptable on specific (perceived) infractions, which low accept rate was a convenient indicator of. I think a general metric showing some nebulous amount of community participation or non-participation would be perceived much more like rep. –  Pëkka Jan 28 '13 at 10:01
    
@LightnessRacesinOrbit: I think that's where the divergence for me hits hardest. Defining "contributing properly" is so wildly hit and miss for all sites across the network, it'd be tough to establish good, solid criteria for that. Not everyone elects to get tens of thousands or even hundreds of thousands of rep on any SE site, but that doesn't necessarily mean they're improperly contributing. I've yet to see a reliable, assured metric that strikes a balance between the casual user of Stack Exchange versus the devoted user. I'm not really sure there is such a balance. –  Makoto Jan 28 '13 at 18:48
    
@Makoto: And I've yet to see any evidence whatsoever that the established regulars and experts on SO routinely get this wrong. On the contrary, I think the fact that the site has continued to succeed proves that the regulars and experts who keep the damn thing afloat routinely get it right. –  Lightness Races in Orbit Jan 29 '13 at 3:09
1  
@LightnessRacesinOrbit: I think it's wrong to use Stack Overflow as the golden standard here. This change didn't just affect SO; it affected the entire network. So anything that has to replace this feature has to apply to the entire network. I maintain that there are regulars and experts on smaller sites who don't have a stellar reputation, but have a handful of very insightful answers, or participate in minor but significant ways, such as editing a handful of posts or flagging questions. These people must be taken into account if any sort of "citizenship" calculation applies. –  Makoto Jan 29 '13 at 5:13
1  
@Makoto: Mmm that's a good point. That does change things. To me it almost seems like SO needs its own rules as it is indeed quite a unique place. Almost. –  Lightness Races in Orbit Jan 29 '13 at 16:04
    
Reputation is clearly aimed to encourage frequent, active users of SO sites. Maybe, a citizenship metric can demonstrate recognition to the second-order users that only take part occasionally, but can appreciate that their behaviour is noted, and not penalised for the occasional participation. –  Pekka Jan 29 '13 at 19:53
    
@LightnessRacesinOrbit: Almost, but not entirely. While SO is one of the original trilogy sites, it shouldn't be the exception to global changes to how we measure something new like this. Changes should be applied globally and equally. –  Makoto Jan 31 '13 at 18:39
    
@Pekka: It really depends on how the metric happens. Again, I feel that having it visible would make users likely to attack those with lower "citizen" values. If it weren't visible, I don't think I'd have any problems with it. But if it's visible, then that's the rub. –  Makoto Jan 31 '13 at 18:44

I think this a very good idea for the reason given; it promotes good behavior while at the same time aligning with (in my belief) basic psychology.

I like that whole altruistic notion that it's all about the question and if the question is good than it's reward enough to answer it and to contribute to the quality of the site. The problem is that I don't think it holds water. If I honestly look at myself, I answer questions for three reasons:

  • The satisfaction of helping another person (samaritan)
  • Increase my standing in the community (vanity)
  • I enjoy the challenge (recreation)

Improving the quality of the SO and the value for future users does not motivate me one bit. Firstly, StackOverflow is an abstract concept, a thing, so I get no samaritan satisfaction in helping it. Secondly, I have no stake in SO. Increasing the overall value of the site and the repository of questions and knowledge gives nothing back to me and does not activate any of my basic motivators.

However, by giving me an arena (or a playing-field) where my motivators can be engaged, great value can be generated for the site. But that is an effect of my behavior, not a primary motivation for it.

Because of this, the quality of the question is equally important to the perceived quality of the questioner. If someone posed an interesting question but I perceived that person as a "bad member" or less deserving of my help, I might still answer the question just to get some community recognition (i.e. up-votes) but I would not put my best effort into it. I would get no samaritan satisfaction in helping someone "undeserving". A good question from a "good citizen" however hits all my motivators and triggers me to make a better contribution myself.

And I have a strong belief that my attitude is shared by a great portion of SO's users.

EDIT: People are going to pass judgement on questioners regardless. The question is, does StackOverflow want to influence what they base their judgement on, or not? Take away all quantifiable information (e.g. answer history, past behavior etc), and judgement will be based on non-quantifiable (and/or possibly irrational) factors, like spelling, username and so on. And I think that would be worse, plus a missed opportunity to promote general behavior in a desired direction.

share|improve this answer
    
That's a good way to capture it. The game is played on multiple levels, it is not incorrect to "only" look at the question, but we can also provide context for the higher level game -- if you wish to play that part of it. –  Jeff Atwood Jan 28 '13 at 8:30
3  
I don't see how your argument is a positive. Assuming you have two equally valid questions. @Jeff's new score has told you that one was posted by a "bad" user. Now, you answer these two questions differently. The question of the "bad" user you don't answer as well. This is a net negative; if you didn't know that one user wasn't as good as the other then you would have answered them equally well, which is surely what the sites are about? –  ben is uǝq backwards Jan 28 '13 at 8:37
2  
"I might still answer the question just to get some community recognition (i.e. up-votes) but I would not put my best effort into it"...that's a pretty convincing argument against the proposed feature –  Bart Jan 28 '13 at 8:44
    
@benisuǝqbackwards or I might not answer the questions at all (or not with any great effort) since I don't know anything about the questioner. You get all questions answered equally bad. –  pap Jan 28 '13 at 8:53
    
@ben even Jon Skeet answers questions that he shouldn't. You are right that the correct behavior is not to answer but our reputation system currently encourages answering even if the question is bad or inappropriate, even if a particular question is ultimately deeply unhealthy for the site -- answering is very likely to generate reputation and up your rep score. So yeah, something that would help balance that out would be healthier for SE communities. –  Jeff Atwood Jan 28 '13 at 9:12
    
That's quite not what I said @Jeff. If there are two equally good questions I think they should be answered equally well; no matter the community engagement score of the user asking. I agree that answering bad questions is unhealthy for the site. Answering good questions from a "bad" user badly cannot be construed as healthy whereas answering good questions well cannot be deemed unhealthy, I believe, no matter the engagement score of the user. –  ben is uǝq backwards Jan 28 '13 at 9:53
    
@benisuǝqbackwards No argument about your desired end-result. But it's the way there. If we stipulate that people are more inclined to give good answers to "good people", we should encourage people to be good. Having something like a "citizen status" is one way. If being a "bad citizen" causes you to get answers with less quality, people will be encouraged to be "good". If you truly believe that the vast majority of users here don't care about who the questioner is, then disregard this. But you have to align the rules and the field with what motivates the users. –  pap Jan 28 '13 at 10:06
    
@Bart "that's a pretty convincing argument against the proposed feature" - maybe in the short term. But we are talking about promoting certain behaviors over the long term. Both in the people asking and answering questions. –  pap Jan 28 '13 at 10:10
1  
The behaviour that should be promoted is giving good answers to good questions. And downvote/close/delete the bad ones (and firmly so. No objection to that whatsoever). I don't want a user giving mediocre answers because the asker has a mediocre standing in the community. Now that would be harmful on both the short and long term. –  Bart Jan 28 '13 at 10:12
3  
I don't think we should be taking the relative merits of the asker into this @pap. I think we should be giving good answers to well asked questions. That's the point of the sites and is what has made them successful (and why I'm here). I don't believe we should be expanding a lot of effort in working out whether the user deserves an answer to their question or not but whether the question deserves an answer. I would much rather time was spent on better duplicate detection methods etc than on creating some score to encourage prejudice, and in your own words bad answers. –  ben is uǝq backwards Jan 28 '13 at 10:21
    
@Bart You may not "want" people to give mediocre answers because the questioner has a mediocre standing, but that doesn't mean that they won't. I'm saying that the standing of the questioner is a factor, regardless if you want it to be or not. Attempting to hide the standing may cause people to assume it's a mediocre user. Plus, we want to encourage people to be "good users", don't we? If being a "good user" means getting better answers, we should use that motivator, not hide it. –  pap Jan 28 '13 at 10:22
    
@benisuǝqbackwards See also my comment to Bart - we have to work with the reality of what motivates people. It would be nice if we all could just "[give] good answers to well asked questions" but I don't think that's the reality of how people work. If someone asks me a question, irregardless of context, my answer will always be prejudiced on how deserving I find the person asking is, not just how deserving the question is. If you disagree with my assessment of the psychology of the users, say so. But you have to model the system on how people do act, not how you wish they acted. –  pap Jan 28 '13 at 10:28
2  
If you're not told with some metric (which you have no control over how it's constructed. How do you know it's "right") how do know whether the user is "deserving" of having their question answered? You have to expend a lot of effort investigating previous questions/answers etc, which, psychologically speaking, will put the majority of people off. Those that it doesn't put off will do it anyway, score or no. The answer to your last argument is to not have the score, so that people don't take it into account. I'm leaving this here as I've said my piece and I'm taking up a lot of the thread... –  ben is uǝq backwards Jan 28 '13 at 10:31
    
@benisuǝqbackwards That's fine, we'll agree to disagree. –  pap Jan 28 '13 at 10:56

It would be interesting to give it a try by showing "citizenship level" value at MSE for 2-3 months.

As for how to calculate the citizenship level, I am pretty positive that whatever would be measured, it would better be expressed in logarithmic scale: it would be much less interesting to know if one user made 1.25 more actions than another as opposed to if there was an order or two difference. Logarithmic scale looks like the best fit for that.

Regarding what to consider for citizenship, from what you wrote it looks like general idea could be briefly described as

all the actions user does in the system

Note above naturally leaves reputation out, since reputation is given to user by others. The (maybe incomplete) list of actions includes everything that may be potentially visible to others:

  • registration, association bonus
  • posts
  • answer accepts
  • bounties offered
  • edits (including those done to own posts)
  • votes cast (including those to close, delete and votes on comments)
  • comments
  • flags (probably including declined ones, more on that below)
  • chat messages
  • favorite questions that one feels slippery; included only because it formally fits above definition
  • reviews
  • days visited

Since the actions listed above are very very different, making it somehow "sorted" or "weighted" could make the system too complicated to understand. Because of that, it would make better sense to me to just use sum of plain numbers, without making a difference between eg helpful and declined flags etc.

Above approach (plain sum of plain numbers) intentionally ignores details and nuances of specific actions because I believe that these should be addressed by badges.


It's that easy, double amount of actions => citizenship level increments by 1.

   actions | level
   --------|------
         1 | 1
         2 | 2
         4 | 3
         8 | 4
       ... | ...
     16384 | 15
       ... | ...

As an example, current citizenship level at MSE for gnat is about 14, for Jeff Atwood: about 16.

  • Note how this would also serve a purpose previously addressed by accept-rate indicator: seeing that user performed hundred(s) / thousand(s) actions gives one a sufficient confidence that user won't skip accepting an answer "by mistake", without substantial reason. And the opposite, seeing the low citizenship level means that user has not yet much experience using the system, sending a "signal" that they might benefit from a friendly advice / guidance.

One thing to decide on is whether to count deleted posts / comments or not. I for one have no particular preference here. As far as I can tell, current user stats do not give a consistent preference for one way or another: eg profile summary page shows count of votes including those cast on deleted posts while activity sub-tab shows count of votes excluding these (per discussion at Prog meta). Whatever is chosen though, it would better be easy to explain to those reading the "citizenship level" values.

share|improve this answer
    
    
interesting that this resembles the way how is meta participation calculated: "combined number of posts, votes, comments and edits... Take the sum of those four values and sort the users based on those sums..." –  gnat Jan 20 at 21:29

The gray rectangle in the bottom left corner is quite meaningless without the accept rate. It helps us to learn about the citizenship level of the OP. Because I like to be appreciated for sparing my time to give my suggestion to someone

Also it allows me to encourage other users to give their valuable answers to my questions by showing them Yes I have 100% accept rate. You will surely be valued if you could provide me with good answers

I am not saying this because I don't want to help the people with 0% accept rate. but what I think that is there should be loyality for the users who put their research to solve someone's problem.

What I have found is that most of the users only accepts answers, when they are asked to improve their accept rate. So at time, it's not shown on the posts, the users will not bother as much to accept the answers, somehow it could halt the good answers,

  • Good answers will not get appreciation.
  • It would be difficult to know which answer exactly solves the problem.

So I don't think it's a good idea to just remove the Accept rate from the posts without providing a good alternative for it.

share|improve this answer
    
Having a 100% accept rate will not (or at least should not) impress other users into answering your questions. It is perfectly legitimate to have an accept rate much lower than 100%, particularly if you pose challenging questions. –  Asad Feb 6 '13 at 4:19

This citizenship level seems to be intended to solve a problem that I don't think was the primary problem with displaying accept rate. The specificity of accept rate may have been a minor issue, but I think the real issue was it provided a metric to the community but lacked a way for the community to provide negative feedback other than comments which were then deemed to be rude. If this citizenship level doesn't include a nice way for users to say become a better citizen they will say it in comments.

These comments would not be at all similar to increase your rep comments which don't really occur. We currently don't see "increase your rep" comments for two main reasons. The first is rep is not something the user can directly affect by their own actions, other users must deem them worthy of a rep increase. The second is closely related, users are able to affect each others rep through votes to say whether a particular user's posts are worthy of gaining rep.

A citizenship score would be similar to accept rate in that its value is controlled almost exclusively by the individual, not the community. If you want to create more systems to encourage participation, you need to build ways for the community to feel they have some level of control over them or you will see the same comments, because that is the only option available to the community to exert control.

share|improve this answer
    
As a colleage here is fond of saying, "the community doesn't exist, there is only each one of us". –  vonbrand Apr 12 '13 at 14:38

The aim of the reputation was to say how much someone is rewarding the people who spend their time to answer his questions. There are many users who ask a lot and later even won't care for accepting or even upvoting.

But the problem is that not all questions that have answers have an answer that can be accepted. I've asked even once a meta questions about that questions: What's the best way to deal with orphaned questions with partial or wrong answers?. I've asked a few such questions, usually when working with niche technologies or rare technology mixes.

The problem is that such questions have answers which either aren't helpful (the answer simply was not working), and there are no more answers even after half a year, or I've got an answer when I don't work with that framework for a year (project change, company change etc.). In my case, most of my questions were less niche, but someone who has mostly such questions could end up being ignored because "he's the egoist who doesn't care about accepting", which was very unjust.

So I'm grateful for removing accept rate from user profile. Citizenship idea is quite nice, but the criteria are quite fuzzy.

Instead, we could create user involvement rate. Involvement rate means if someone cares about other people's work. It means reaction to the answer. How can I react:

  • accepting answer
  • upvoting answer
  • downvoting answer
  • commenting answer

The last would be the most appropriate in my case. If something is not working for me, but I know it could be solution by other people, because it sound reasonable, I don't want to upvote or downvote, but I leave comment that unfortunately it is not working for me.

If I can't test it because I don't have access to the old project or environment, I can answer that unfortunately I can't test it.

But the most important is, if I care about other people's answer by reacting to them. You know, as answerer, that someone has read and analysed what you have written. It makes you feel you're doing something important for other people.

share|improve this answer

I agree that there should be an alternate metric, but most of your general citizenship idea is not useful since this site is mainly a Q/A website. So, the primary purpose of a new metric should be to improve answer quality, and to promote wise allocation of time in answering questions. Since everyone's time is limited, spending it on less useful questions implies that you have given up the opportunity of answering questions that you could answer better. Therefore, when you design a new metric, it should try to match the right people to the right questions!

From this point of view, I think the following metric would be useful:

  • Reputation: Quantifies how much work a person has put in this website
  • Activeness: Quantifies how active they are in the last 30 days
  • Responsiveness: Quantifies whether the OP takes care of their own question, using any form of response, including upvote, downvote (!), accept, or comment on the answers.

The last two metrics are very important motivation for me to give an answer here. Those two metric are not shown now, but I know that the OP (@Jeff) must give me responses and in a short time and therefore I am writing my answer here. There are already twenty answers here so most people will not read this answer. If the OP does not even read and give me any responses, there is no point for me write this answer, right? Even if I have good answer.


Let me elaborate why I think these metrics are helpful. The value of a question certainly is determined solely by the question itself. However, I will only spend a certain amount of time to read it, unless it is written by a reputable person (more likely to be valid), or a responsive person so that I can ask them to clarify.

As an answerer, if the question is very esoteric (and maybe useful) that only small group of people understand and have interest in it, I would consider to answer it only if the asker would give me a reply, since the OP might be the only audience that would read my answer! Another situation is old questions with answers. Even though the question itself are good and I know a better answer, I might answer it only if OP is active and will give me a reply. Otherwise, the question will pop up in the front page, the top rated answers get upvoted and the new answers at the bottom stay there without ever being seen, so pathetically.

As a learner, I would only read those answers that get a response, written by high rep user or with the clear answer in the first few lines.

Showing this data would allow us to decide how to spend our time answering questions. The removal of the accepted rate already prevents people from wasting time. Now you should give them information on how to allocate their time more wisely.


Some replies to other comments here:

What is the desired effect of it. Say we have a broader metric visible to others. What should I do, or how should my behaviour change if I stumble upon someone who is not acting responsibly? ..." – @Bart

You should move on to the next question. Just don't waste your time commenting if the poster is not active anymore. If the OP is not responsible and there is a long list of answers, please don't add new answer because no one will see it.

I don't mind having this, but it should not be displayed on questions. The quality of a question can be determined by reading it. Knowing more about the asker should not change your vote - @Manishearth

I agree that the quality of question is determined by itself, and extra information would not change my vote. But it would change the time I am willing to read it and answer it.

What is more useful to the longer term health of a community: a single ok question, or an engaged community member who assists and participates - @Jeff

Both, but not all people are willing to participate in both of them. Also, most information does not help any questions and answers, which you consider the first class citizen, at all. I would also suggest you to remove those blade counts because it is already correlated with reputation, the details blade are important though. I am in favor of the information being put on their user page.

Remark: The Activeness and Responsiveness involves information that is not in public so it is much more helpful the user page. You can put a high rating on accepted answer to motivate them to accept more answers.

share|improve this answer
1  
Responsiveness is an interesting idea - it is frustrating to have what you feel is a better answer, and to have it buried. But I don't think encouraging an OP to act on each one is the best choice, as that'll likely be noisier - extra comments, or upvotes, or downvotes, especially when the OP has already found the answer he/she needs. –  Hannele Feb 7 '13 at 14:08
    
@Hannele your first edit improved the post, thanks. However, your new edit is giving new meaning to sentences, it's not grammar and not spelling. Please don't do that. –  Shadow Wizard Feb 7 '13 at 15:38
    
@ShaWizDowArd It appears to have been accepted now, and I do believe it preserved the original intent. Perhaps the comment should have been clearer (word choice, or something like that). –  Hannele Feb 7 '13 at 17:21
    
@Hannele The response is not necessary from the OP, it can come from other people. As far as questions are good with most answers getting upvote/downvote/accept/comment, then the OP has a good "grade". It is not a punishment, but bonus as a status of questioners. It has no effects on those high rep user only ansewring question. The main point is that this information is not accessible at the moment, not by rep nor blage, etc. –  hwlau Feb 8 '13 at 1:01
    
I got intrigued by some questions, and answered them (or even a generalization/extension of them) for my own amusement. If OP even sees my answer is mostly moot to me there. YMMV. –  vonbrand Apr 12 '13 at 14:47

In the interest of maintaining functionality until a citizenship level is added, I have created the following script:

function getPercentage(user, totalanswered, totalaccepted, min, max) {
    min =  min || 1;
    totalanswered = totalanswered || 0;
    totalaccepted = totalaccepted || 0;

    var percentage = 0;

    $.ajax({
        url: "/users/" + user + "?tab=questions&sort=votes&page=" + min,
        type: "get",
        dataType: "html",
        async: false,
        success: function(data) {
            var jdata = $(data);

            max = max || jdata.find(".page-numbers").length - 1;

            if (isNaN(max)) {
                max = min;
            }

            var answered = jdata.find(".answered").length;
            var accepted = jdata.find(".answered-accepted").length;
            totalanswered += answered;
            totalaccepted += accepted;

            min++;

            if (min > max) {
                percentage = totalaccepted / (totalaccepted + totalanswered);
            } else {
                percentage = getPercentage(user, totalanswered, totalaccepted, min, max);
            }
        }
    });

    return percentage
}

$(".owner > .user-info > .user-details").append("<div style='color: #700'>Accept Rate: " + (Math.round(getPercentage($(".owner > .user-info > .user-details > a").attr("href").substr(7)) * 100)) + "%</div>");

Or you can use it as a bookmarklet, by copying it and pasting it into a js minifier, then a bookmarklet generator like the one at http://chris.zarate.org/bookmarkleter

I would have posted the bookmarklet itself here, but SO doesn't allow javascript links.

For ease of adding this yourself, I have put a link in a jsFiddle that you can just drag into your bookmark bar: http://jsfiddle.net/zxkPk/1/

share|improve this answer
2  
It says Jeff Atwood's rate is 0 for me. –  Mechanical snail Feb 6 '13 at 4:22
    
Sorry about that. I only did it for SO, not any of the StackExchange variants. It's been updated now. –  d_r_w Feb 6 '13 at 4:33

Citizenship, to me, implies well-rounded-ness in the use of Stack Overflow. Not only do you ask questions, you accept answers, you write new answers, you edit, you review, etc.

So, what else is well-rounded? Delicious pie, of course! Or cake, if you prefer.

I propose showing a wheel on a user's profile, with wedges that are representative of the user's activity in a particular area of the site. Once you've performed some quantum of activity in that particular category, you are granted the wedge. A similar feature was introduced on Kickstarter last year, where you collect a wedge for supporting a project in each category. Classically, of course, it takes inspiration from collecting wedges in Trivial Pursuit.

You could use more saturated colours for higher activity levels in a category, or make it a simple binary. You could have a count on the number of activities in a wedge, for example, you wouldn't get the question wedge until you've asked a certain number of questions, or accepted answers for them, or upvoted/edited them, or earned a certain number of question-related badges, or some combination of all of the above.

I don't believe it would dilute the existing systems, as it would be more about displaying the existing badges/reputation gains in a more consolidated way (a meta-badge, if you will). For examples of categories, see the above list, or just about every other answer here.

I don't believe it would unfairly penalize active contributors, as I've rarely seen a contributor that hasn't asked a single question. Even so, there needn't be specific question and answer categories - a content category could cover both ways of contributing content to the site. Other answers with useful categories: badp, lechlukasz, and even the original post.

But most importantly, it would help give new users a sense of the broader goals of the site. The current collection of badges is fairly useful in that sense, but there is an overwhelming number of them, in alphabetical order. Which is fine if you're looking for criteria on a specific badge, but not so great if you're trying to find out which badges are for good answers (for example).

share|improve this answer
3  
So once you've performed a group of actions you get your slice of the wedge; just like with /review badges? Encouraging people to "catch 'em all" has a multitude of problems and encourages needless and counter-productive action. It also means that decent contributors may be unfairly castigated because they don't have their "question" wedge, or whatever they represent. If this has to be implemented I'd much rather it was obfuscated through an algorithm so that there's isn't a specific requirement that certain actions be performed. –  ben is uǝq backwards Feb 6 '13 at 22:38
    
I doubt that decent contributors would be castigated for not, say, having a Question wedge (u great answers, y u no ask question?!), although I've rarely seen a good contributor that has asked absolutely no questions. Perhaps, it could problematic for someone answering their first question, or someone who has only the question wedge. –  Hannele Feb 7 '13 at 14:09
3  
There are users that only answer and never ask (or the opposite) on some sites, including me. :P –  Alenanno Feb 7 '13 at 14:36
    
@Alenanno But would you consider yourself active and a good, well-rounded citizen on those sites? I'm also not proposing a question/answer dichotomy - there could also be wedges for maintaining the site (edit/review/tag wiki), for voting, for meta activity. –  Hannele Feb 7 '13 at 14:40
    
@Hannele I haven't posted lately, but I'm a 10K user so I'd say yes. –  Alenanno Feb 7 '13 at 14:42
    
@Alenanno Fair, I see your English Language profile. But I'm also not proposing a question/answer dichotomy - there could also be wedges for maintaining the site (edit/review/tag wiki), for voting, for meta activity. –  Hannele Feb 7 '13 at 14:46
    
@Hannele Oh I was just letting you know since you stated something about it in your answer. :) –  Alenanno Feb 7 '13 at 15:04
1  
@Alenanno It is interesting to see how that can happen! And I've clarified my answer because of it, so thank you. –  Hannele Feb 7 '13 at 15:07
    
@Hannele You're welcome! :) –  Alenanno Feb 7 '13 at 15:08

I think showing only reputation on questions is fine. When I'm deciding whether to spend time on a question, I consider things like:

  • Is it understandable?
  • Do I already know the answer?
  • How extensive an answer (plus possible research effort) does it need?
  • Am I interested in the answer for myself?

If the answer will require significant effort and doesn't interest me, only then might I care whether this is a user deserving of that effort by virtue of their own contributions.

The accepted answer rate doesn't help with any of these. Reputation gives me some warm fuzzy that the user won't abandon their question, in addition to a measure of their contributions to the site, without differentiating contributions of good questions, good answers, or site maintenance.

As such, reputation is also a reasonably good metric for awarding powers (comment, edit, etc).

Before introducing a citizenship score separate for reputation, someone needs to answer "If something matters so much, shouldn't it count toward earning powers?" If the answer is always yes, the reputation is the only citizenship score we need.

Where I'd like to see a metric other than reputation is on answers. Reputation doesn't tell me anything whatsoever about how trustworthy an answer is, simply because reputation draws from so many sources. On an answer, it would be useful (and especially so to newer users) to see:

  • Does the user have badges on THIS TAG?
  • Is the user getting upvotes (and not downvotes) for other answers on THIS TAG?
  • If a combination of tags is popular enough, look at score on that combination particularly.

Questions aren't authoritative. Answers need to be. Put the metric of expertise on the answers. (And no, I'm not suggesting that acceptance rate is a measure of expertise, nor does the "good citizen" score appear to be one)

share|improve this answer
    
Yes, but... if the user is clearly a newbie (low reputation) and shows some effort, I'm more willing to help out, other factors (roughly the same as yours) being the same. –  vonbrand Apr 12 '13 at 14:37

The problem that I see with the new proposed metric, it wants to replace a metric that was exclusively used for judging questions with a general-purpose metric. We've already got a general-purpose metric, two metrics are just confusing, not useful. So I'm concentrating on a useful metric to replace the Accept Rate.

Reputation measures excellence, any chance the new metric will measure mediocrity?

Reputation is absolutely awesome for a couple of reasons:

  • Allows one to "grow" it's reputation doing the thing they're good at. Let good askers ask and let good answerers answer. Let the native English speakers with a knack for proper english do the editing.
  • The really high reputation goes to the ones answering questions. This might sound elitist, but it's fine that way; It's a Q&A site, answering questions drives the site, not asking or editing.

Since the new "citizenship" meter can't possibly weight in the exact same way as reputation, yet it wants to measure everything, there's a good chance it's going to be a measure of MEDIOCRITY. One'll get a good score if one does a little of everything. If it's any other way it will either raise with up-votes (and then it'll look exactly like reputation), or it will not and then it will be far worst then mediocre, it'll end up promoting everything but votes.

Any new measurment we do get should help Askers understand if the questions they ask are good or bad

This is a Q&A site and it's measure of excellence lays exclusively with the question and answers. Everything else is just an aid: edits and flags are nice, but they'll be useless without good questions and smart answers.

I do approve of a new metric, used the same way as the old "accept rate", but one that measures the pertinent factors in what makes a good QUESTION and nothing else:

  1. Are the question properly researched, written in a proper way and in accordance to the site FAQ?
  2. Is the asker actively answering to questions from the answerers?
  3. Is the asker argumentative in comments?
  4. Does the asker accept answers?

All those aspects of question asking can be determined using automated systems. (1) can be determined by examining question scores and question close-rate. (2) can be determined by measuring the time it takes for the answer to answer anything in comments. (3) Is a bit more difficult but very important, because it erodes the morale of answerers. We're all p****d with smug users that are so much smarter then us yet somehow need to ask a question. (4) Is related to (2) and it's a way to measure user involvement.

When asking a question, those elements are not interesting:

  • Does the user vote?

    Useless and very dangerous. Let the Askers know that votes count, and they'll vote, and there vote will only represent there desire to get a better score so they attract better answers.

  • Does the user answer questions?

    I'd say this poses a small danger as well. Example: If I'd ever show up on the english stack exchange I'd surely exclusively ask questions, because I know I'm not the right person to answer anything there. Do you think that'll work for SuperUser or StackOverflow? Any chance we'd get dummy answers because one needs to have answers for the score? Less of a danger since bad answers are promptly downvoted.

  • Does the user edit or suggest edits on questions?

    According to the last StackOverflow survery most users are not from native english speaking sites. Should users start editing questions simply because they need to improve the presumably visible citizienship score?

  • Does the user flag stuff?

    Flagging is to many a "gray" area; On the one hand there's the theory of doing what you can yourself, without involving the moderators. Then there's the (un)desire of some to become a "snitch". Think of all the ex-comunist country members where "snitching" has a particularly bad history attached. I just don't see how this should be relevant to asking questions.

share|improve this answer

On the rare occasion that I looked at accept rate (because I seem to have a blind spot; maybe that makes me a Good Citizen) it wasn't because the question was good. It helped me decide on whether the asker of a godawful question was worth the effort of trying to help them improve it, and I think that was a good thing to have.

If you want to pursue this further (which seems in doubt, given the feedback here - most of which I agree with) then perhaps this aspect would be a more useful metric for a "secondary rep" than the be-all "citizenship" concept.

In other words, devise a metric which focuses on respect for answerers and their time and effort.

It could include the following, although some of them are dubious.

  • Voluntarily removed questions with negative vote
  • Closed or migrated questions (negative score!?)
  • Accept rate
  • Flags (negative score!?)
  • Long comment threads which refuse migration to chat (negative score!?)
  • On the other hand, "responds to comments", while not very focused, might be a useful counter-measure

Maybe only display the result if it's positive? I.e. let the absence of a "nice buddy star" mark a user as a potential waste of your time and spirits, as it were.

Because most of these are hard to remedy, I would like to see a time cap of some sort. Because some users visit only rarely, a fixed time window isn't very good; so maybe calculate it on their last 30 "active" visits or something like that. (I dont't know what exactly "active" means but just logging in and passively reading the site shouldn't count.)

Still not sure how new users should be tackled. They're a problem in any system which bases judgments on previous actions, anyhow. I suppose "innocent until proven guilty" would work here, especially in combination with the low rep of new users.

share|improve this answer
1  
"Voluntarily removed questions with negative vote"...given that that is advised against with regards to post-bans, that might not be a great thing to include. –  Bart Feb 9 '13 at 10:54
    
I'd not worry so much about the one asker as the multitude of latercommers stumbling on the question. If the question is worth asking and answering, it certainly deserves to be kept shipshape. In this vein, I have taken extra care to improve high-upvote questions and answers, as they presumably are the most important ones here. –  vonbrand Apr 12 '13 at 14:50

At the risk of flogging a dead unicorn, I think that there's another angle here.

All this discussion turns, I think, on the concept of 'citizenship.'

I'd suggest that we could usefully consider users as sorting themselves into three categories:

  1. Members of the community
  2. Well-behaved casual visitors
  3. Trouble

The first category are the only people who are going to build up a significant inventory of reputation. If they are experts, they are answering questions. If they are not answering questions, they are engaged, repeat, visitors who ask questions. They ask good questions, they vote, and, yes, they even accept when an answer actually helps them.

The second category are people who turn up, at most, from time to time. They don't infest the first page looking for things to answer. They don't create a lot of questions. When they have a question, they ask it. And, again, they ask on-topic, coherent, questions. We'd like to hope that they would click on the vote arrows or the check mark from time to time, but I'd hate to think that we'd hate on them if they did not. Simply showing up and asking a good question should be enough. That 'makes the internets better.' That provides fuel for our more obsessive participants to answer and to vote for.

The last category, are, of course, the reason for all this discussion. Whatever their frequency of appearance, they are people who do not add good content to the site.

If you want to go creating metrics, by all means. Just, please, design them to draw a line between (2) and (3), not between (1) and (2).

share|improve this answer

You must log in to answer this question.

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