How can Stack Overflow discourage duplicate questions?

Question

Ok, so questions and suggestions about duplicate questions are pretty common, often being duplicates themselves. My question is really more about how Stack Overflow could change in order to discourage a problem that many people see as one of the leading problems of Stack Overflow. And, as a sub-topic, how can we encourage people to answer older questions, or update answers to them?

Jeff Atwood has posted in his blog that he sees a moderate level of duplicate questions as healthy, and I agree with that. Apart from the fact that differently worded questions may lead to easier searching, I think Stack Overflow has a process problem that encourages people to ask duplicate questions, and more importantly to discourage people from answering older questions with new or better answers. Duplicate questions help to encourage better answers, IMO.

However, I don't think this is the way it should be. I would love to get behind the "one question to rule them all" camp, but because of Stack Overflow's design this makes it impractical. Here are the reasons why Stack Overflow encourages duplicates, and discourages improving old answers:

1. As the number of answers to a question grow, the likelihood of gaining any rep from an answer goes down. If a question has 20+ answers, the likelihood of my getting any rep from an answer is almost nothing, regardless of how good the answer may be.

2. When a question is marked "answered", it discourages people from writing a better answer than the selected answer. This is particularly true of older questions in which the technology may have changed and a better answer now exists.

3. People seldom read past the first page, and even more seldomly read past the second. Merging questions and adding further answers makes it harder for people to find answers. If you merge two questions, and there are now 50 answers, and the best answer is down at the bottom (maybe because it was recently answered) then people just aren't going to find it.

4. People disregard older questions as no longer relevant. If I'm searching for an answer, I'll prefer more recent questions to ones asked 2 or 3 years ago. This is the same way that I tend to discount old how-to's, because chances are the answers no longer apply to the current technology.

5. Even if an older question is unanswered, the user that asked the question may no longer be on Stack Overflow, or may not have visted the site in 6 or more months making the likelihood of getting that answer check mark very slim. The fact is, old questions just don't get the attention that newer ones do.

Simply put, Stack Overflow gets duplicate questions because its workflow encourages it. No matter how much moderation you put in place, and how much closing and merging you do, you're just plugging your finger in the hole of the dike. The only way to fix the problem is to stop Stack Overflow from encouraging it.

So, what are some ways to discourage duplicate answers that do not involve moderation, question merging, or yelling at users for asking duplicate questions?

I think we need to either change the Stack Overflow workflow to discourage duplicates, or we need to accept that duplicates are a way of life and create process that makes them useful. What is not useful is having a workflow that encourages something and then running around after people and trying to clean up their messes.

Answer 1

Ok, let go through this one by one:

1. Tough luck. If it's any consolation, when number of answers increases high enough, the question and all the answers will become a Community Wiki, so no one would get any rep anyway.

2. Does it? If it is still relevant question to others, you may still get rep for upvotes. There's even a fancy badge for the extreme version of this scenario.

3. If there are 50 answers, then it's a CW already. So, what's keeping you from including the parts of other answers to those on the first page. It's not like you'll steal rep from the other answer authors (it's a CW and they are past the second page anyway), and if you really want you can reference it.

4. This only reflects the technology you use. If I'm using a 2-3 years old API, I'd look for answers in older questions.

5. The checkmark isn't the only reason to answer a question. Ok, there's that thing about helping each other out, making the web a better place, etc. Even if you consider all that a bunch of hippie talk and prefer to just get good old rep, you'd gain much more through upvotes than accepted answers anyway.

So, now I got to the main part of your question:

So, can anyone suggest ways to discourage duplicate answers that do not involve moderation, question merging, or yelling at users for asking duplicate questions?

Frankly, I'd rather keep it as is, but if I did make any changes that would be instituting a obligatory test to all new users which would introduce them to the way this site works, and then test them on already prepared fake questions/answers. Like a tutorial level in a game.

But, this will probably never be done, because it would lower the arrival rate of new users. It would probably annoy me too if I had to pass it. I always skip tutorial levels, too (Yeah, yeah... I get it. Mouse drag selects the units, right click tells them where to go. Let's start already!)