Reward finding duplicate questions - +10, +2, -5

Question

Give an incentive for finding duplicate questions deals with various ways of rewarding the finding of duplicates, a tough and increasingly necessary janitorial task.

Here's a concrete suggestion:

1. Reward the finder of a duplicate with +10 points if the question gets closed as a dupe. If multiple duplicates are found, reward all finders whose suggestion was voted for at least once by another user. (The limitation is to avoid gaming by making frivolous suggestions when it's obvious that a question is going to get closed as a duplicate.)

2. Reward each following closevoter who picks one of the previously suggested dupes with +2 points if the question gets closed.

3. Punish everyone involved (dupe-finders and dupe-voters) with -5 points, and revoke all rep gained from the dupe-voting, if a question closed as duplicate gets reopened.

A punishment for mis-voting is going to be controversial but in my eyes, it is absolutely essential. Finding a good dupe is a work of love: You need to make sure you understand the OP's question, and you need to scan every potential duplicate for whether it really contains an answer that will help the OP. Everyone dupe-votes carelessly from time to time, so the threat of punishment needs to be present to keep everyone on their toes, just as the threat of downvoting does when you post an answer.

If you think +10 is too high a reward: Consider that finding a duplicate can often be more work than writing an answer, and a correct answer usually nets you at least 10 points, if not much more.

If you think -5 is too much of a punishment: -2 might work as well, although I really think there should be a harsh punishment on an unfair dupe-closing.

Answer 1

Finding a duplicate (especially if it's an old one) can be hard. But if one is found, the other votes are a lot easier. So it's probably better to only reward the first voter and only if the question is closed. For example:

1. First closing a question that is less than 48 hours old, should receive a small award (+1).
2. First closing a question that is older (requires more detective work), should receive a bigger award (+5).

This way you reward the actual work.

Answer 2

This has been shot down before IIRC. But it's discussion-worthy in light of all the feature and moderation enhancements over the years. (And the more recent dupehammer of course).

Problems I see with it:

• Unfairly benefits users which already have >3000 rep over newcomers.
• The moderation system and reputation feature should stay separate (that was the conclusion from the previous discussions).
• The interpretation of exact duplicateness is susceptible to moon phases. (And for vague questions only the OP can ever conclude if a duplicate really fits. Or throw a fit if it doesn't.)
• Being able to propose one's own answers as duplicates is both a recipe for abuse to rep addicts, or worse yet: a deterrent for modest posters.

I believe the potential reward and the penalties are too low in any case. It would certainly spur more closevoting, but won't fix the problems of our nightshift. (Also: not all duplicates are bad duplicates.)

1. If awarded automatically

   However such a feature change would never go back in the bottle. If implemented, then it should really be slowly. The rewards should be minuscule for a start. Yes! Why not use floats for our virtual internet money? Let a closevote trade for e.g. +0.25 at the NY StackExchange.

2. Flag-Weight (not a thing anymore, or is it?)

   Obsolete counter-proposal: Keep the reputation system and moderation separated. Instead of rewarding with score, benefit closevoters with flag weight. People are still obsessed with that, right? So give them +1 flag weight for correct closevotes. That's a proper moderation feature after all. And someone who correctly closevotes most of the time will also have a higher probability of correctly flagging. The benefit here is coming closer to obtain the badge thingy that's available (?) for high flag weight.

3. Upvoting DCs

   Now some time has passed. Changed my mind. Automatisms aren't always workable.

   If we want to reward helpful duplicate suggestions, then it'll likely only work out if it was implemented as UI gimmick.

   • Introduce a subtle ▲ button in the "marked as duplicate by …" box.
   • Next to each user that proposed a link.
   • Hand out a few points (+2 or +0.5 whenever clicked.)
   • Exempt highly/well-known duplicates from being eligible though. To encourage suggesting case-specific or harder-to-find and most-exact duplicates.
   • Let every voting-eligible user have like, dunno, 10 such votes per day. (Not deducted from regular Q&A votes.)

   Putting just a few such rules into place kinda makes penalties redundant, I'd say. And if it's not automatic, the democratic community will is gonna sort it out.

Answer 3

I do not vote to close a question for any reward. I do it because duplicate questions are annoying. Reducing the clutter is enough reward to me, but perhaps it's not for others. I would say it's worth trying it and see what the results are.

The main issue I believe are the scattered answers on all those dupes.

To prevent duplicate answers on duplicate questions, what about punishing those who answer the duplicates even when the "possible duplicate of..." comment is there?

An example is this question: How to delete entry and video file in a listview file browser?. I marked as a dupe of this one: How to delete entries and video files in a listview file browser?, which is an exact copy from the same unregistered user. If you see my comment, I even asked the user to not to post duplicate questions. It can't be more obvious to anyone reading the question. Regardless, this user posted an answer there. I do believe this is the kind of behaviour that should be discouraged. Now, since it's already marked as answered, nobody will take the time to check and vote it (I even flagged it for mod attention).

A possibility is yet another annoying orange popup when you post an answer if the question has been voted or flagged as a dupe.

Answer 4

I wouldn't suggest putting reputation on finding duplicates.

I could just track the new questions stream, quickly check for duplicates and thus award myself reputation...

An alternative though is to introduce a new sort of moderation reputation, loose from Q&A reputation.

It shows how much people are interested in moderating the site, we could house the following things in it.

Meta:

• Asked/answered a meta question (+1)
• Accepted question on the meta (+2)

Editing:

• Editing questions (+1)
• Edit gets rolled back (-1)

Closing:

• Closing questions (+2)
• Closed questions gets reopened (-1)

Flagging:

• A flag was good (+1)
• A flag was bad (-1)

Tag wikis:

• Edit on a tag wiki (+1)
• Rollback on a tag wiki (-1)

This is of course merely an idea, this might need to be worked out so the scoring is right.

a tough and increasingly necessary janitorial task

You might be interested in my script Duplicate Question Suggestion Boxes which makes this more easy.

An example of its effectiveness, where otherwise the link would be hidden somewhere in the side bar:

Of course, we can't enable this for all users.

Answer 5

Data collection

A list of questions I've seen in the past few weeks in which users with close vote privileges answered (what seem to be) fairly obvious dupes rather than voting to close. CW so that others can add to the list (these are mostly in my "home" tags).

• Objective-C: `@synthesize fooBar;` vs. `@synthesize fooBar=_fooBar;`
• What is the difference between int and NSInteger?
• Rounding up double numbers in Objective-C
• Objective-C properties
• What's the Difference Between nil and Nil
• Posting message along with an image on Facebook
• Why alias a new name and explicitly release the IBOutlet property?
• @property, @synthesize and releasing objects in Objective-C
• How to define a global variable that can be accessed anywhere in my application?
• Properties and accessors in Objective-C
• So what's the difference between sigaction() and signal()?
• How to call a method inside another method in Objective-C
• Is setting a property to nil same as releasing the property?
• any good tutorials re finding memory leaks?
• How to change the dimensions of picker view?
• iOS: Compare two dates
• Should all properties be declared "nonatomic"?

_{I'm not trying to specifically call out any of the 3k+ users who answered. I think that there may be other valid reasons besides "lust for rep" that made any one of them choose to answer, such as "lust to help out". This is just to help with the analysis of this question.}

Answer 6

3) Punish everyone involved (dupe-finders and dupe-voters) with -5 points, and revoke all rep gained from the dupe-voting, if a question closed as duplicate gets reopened.

This makes it unworkable, IMHO. As you note,

Everyone dupe-votes carelessly from time to time,

Yeah, everyone makes mistakes. Heck, you can turn out to be wrong even when you put careful thought into closing, simply because the asker didn't quite express the problem correctly the first time around and returned to clarify after seeing the duplicate you proposed in good faith...

...Fortunately, questions can be re-opened when closed incorrectly. The last thing we should be doing is discouraging folks from fixing mistakes. And such a penalty would do exactly that.

so the threat of punishment needs to be present to keep everyone on their toes, just as the threat of downvoting does when you post an answer.

Well, you're probably right - this would be a quick (and terribly destructive) way to get some easy rep if there was no cost to it. But I don't think linking the reputation system to the process of question curation will do much to improve the results, with or without penalties. Keep in mind, you can already get reputation by voting to close duplicates... If you answered the original.

I think the solution here lies in tying dup-finding and answering closer together somehow. See: Revamping the Duplicate question System

Answer 7

Imho, the main part of the problem is finding a dup question on SO. There's absolutely no point in rewarding/punishing for finding/posting dups if it's hard to find dups in the first place.

I mean seriously, the Search functionality is inadequate to say the least... You're invariably better off searching from Google directly. A newcomer might not necessarily think of doing so -- and even more experienced users might not do so either.

It's like, either you've hung out around enough to have a vague idea of what the dup might be, or you don't bother looking because doing so and finding something relevant breaks down to sheer luck.

Maybe add the option to search SO using Google (site:stackoverflow.com [search terms])? It will yield more relevant results than the current search tools in almost every case.

Maybe leverage Yahoo!'s (or Google's?) content analysis web service to find related content on SO as one types his question or before it gets published?

Answer 8

There was recently some discussion about reputation awarded from questions and some ideas about limiting maximum reputation received by asking questions. So why to introduce another source of reputation?

If you want to make some award for good Close as duplicate calls, create some badges. Many badges are already dependent on the reputation (you cannot upvote if you don't have 15, you cannot downvote if you don't have 125, etc.). To avoid close as duplicate rushing I would introduce some penalty for wrong close calls - let say if you have several bad calls within short period of time you will temporary loose this ability.

Btw. another point for discussion is how to motivate users to upvote linked question / answers. Many users tend to upvote only new answers or answers on their own questions.

Edit: Actually when I think about this little bit more, removing reputation for wrong close as duplicate calls is not a bad idea. Reputation gave as some moderators privileges so if we used them incorrectly there can be some penalty.

Answer 9

There's a corollary to this.

The suggestion is to reward people for finding (correct) duplicates. However, as Josh Caswell noted there are plenty of users, with enough reputation to close a question, who'll answer questions and not bother to find the duplicate. This suggestion does absolutely nothing to address that.

Anyone can easily get 20-30 rep (add another 15 if the user is nice enough to accept) by answering a question that they know should be a duplicate of another. If the goal of participation is reputation a measly 10 rep for doing the decent thing and trying to keep the site clean doesn't stack up¹ against the potential gain for an answer.

Effectively, you're trying to reward doing good when most of those involved know that they can obtain a greater reward by being "bad".

Before I continue I think it's very important to note that any suggestion should not detract from each sites main purpose. That is, to make it easy for people on the internets to gain a good, quick, answer to their question. I'm not entirely certain that the following proposal sticks to this purpose absolutely.

Counter Proposal:

I recognise that this might seem a little harsh but something has to be done to stop the larger sites (Stack Overflow in particular) from drowning in a sea of questions.

Rather than trying to reward good why not try to game the system so that people are less likely to do bad? I'm defining bad as asking or answering a duplicate question rather than tracking down the original and voting to close.

Any user who asks or answers a question that subsequently gets closed as a duplicate gets 0 reputation for their post. In order to stop this being too harsh I would add a couple of caveats:

1. You still get reputation for your answer being accepted. This reduces the risk that people will stop answering obvious duplicates completely.
2. There should be a time-limit, something reasonable, for instance 30 days. This means that a someone who posts a perfectly decent question or answer won't get punished a year later when the question subsequently gets closed as a duplicate..
3. It doesn't matter if the reputation gain was limited to some other arbitrary (small) number. The point is to partially disincentivise answering dupes.

I doubt the askers care anywhere near as much but it's unfair to punish the answerers alone when the askers are identically at fault.

The obvious problem with this is that all closing as a duplicate might stop instantly. I doubt it as there will always be some decent people around. However, when combined with an incentive, for instance 5 or 10 reputation for being the first to find a duplicate (as per the feature request) you raise the spectre of the dupe finders being rewarded more than the answerers. That should make finding duplicates more attractive.

There's also the darker side of human nature that can be brought into play. You create the perverse incentive to find the duplicate to stop the person who got there just before you from getting 50 rep for answering an easy question. It's not pretty but I can see it happening. Thinking about it, and because of this, maybe it should be 10 rep and not 0. You don't necessarily want the duplicate finder to be rewarded more than the answerer and it'll stop some of the extra nasty stuff.

_{1. Forgive the pun.}

Answer 10

I think your rewards are too high, also the punishment is quite high. But both ideas are quite reasonable. A lot of questions by newbies are closed very quickly, even if they aren't really exact duplicates as it is a requirement by the stack exchange sites.

So this might give an incentive to think again if a newbie might just rephrase his question to keep it from being closed...

Answer 11

I think the idea of mixing rep with moderation is a bit iffy.

However, if this is implemented:

• What to do about the granted rewards if the question is subsequenty reopened? Do you revoke those points?
• Do the reopen voters get rewarded if the question is reopened?
• Are more rewards granted/revoked if the question is closed again?

It gets messy, so I'd suggest keeping it as simple as possible.

• Only reward the initial finder of the dupe if and when the question is closed.
• Revoke the reward if the question is reopened.
• Rinse lather and repeat for all subsequent close/reopens.

Answer 12

What about:

• The reward for finding a duplicate is 10 plus all the rep gained on answers to the closed question.

• All rep gained on the closed questions and it’s answers are removed if the question is closed as a duplicate apart from a max of 2 up votes worth on the question.