What is the ideal number of questions SE wants for any answer for the best search effectiveness? [duplicate]

Question

After reading former SE co-founder Jeff Atwood's blog post Dr. Strangedupe I came to this problem. If I post a wonderful description about the features of a 1967 Chevrolet Camaro, do we want users to find this post easily with different search terms? For example, is it desirable to find this answer with

• What is the first Chevrolet muscle car?

• What was the pace car in the 1967 Daytona 500?

• Did the Camaro ever have window vents?

• What was the black car in Better Off Dead?

• What was the Chevy II Nova frame based on?

These five questions may have the same answer, but certainly the site works better if there are 5 different ways to find a description of a 1967 camaro. Are more questions better? Shouldn't we reward users who improve the validity and usefulness of an answer rather than down-vote them?

Instead it appears SE policy stifles search engine effectiveness by reducing reputation when a presumed answer is already posted under a different question. Currently we treat search-engine augmenting questions as such:

1. OP presents a question.

2. An arbitrary user with 3K+ reputation guesses what the answer is then searches SE for it.

3. Using their judgement, they vote to close the OP question as duplicate and link the answered question (related or not) to it.

4. The OP begins receiving downvotes for the duplicate question and is denied reputation which answers and views would have brought.

The end result is that users who broaden the search terms which can arrive at a good answer are stigmatized and in fact punished with loss of reputation because a duplicate answer was confused with a duplicate question.

There is a logic to connecting certain questions which ask the same thing, but there is no logic to associating several uniquely asked questions together only because they happen to arrive at the same answer. This also hurts the site.

Someone tell me I am missing something, this is my experience.

HOW MANY DIFFERENT questions are ideal for any single answer?

Answer 1

I think you have created a fallacy by taking a point to an absurd extreme. Of course a single word answer is not sufficient and would not be accepted.

The answer is, rather sensibly, that with a well formed question and answer the decision to close as duplicate can be straightforward. If the answer gives sufficient information to answer more than one question it may be sufficient to allow closure as duplicate.

But it may not - depends on circumstances

Answer 2

Fundamentally the questions need to be about the same subject for instance "Why is the sky blue?" and "What is Rayleigh scattering?" may not appear to be the same question but I imagine if you asked these questions on Physics they'd basically be duplicates because you're asking about the same phenomenon in different ways.

If however you asked two completely unconnected questions that happened to have the same result they aren't duplicates.

Having said that a single word answer is not really a good answer and we really want answers with rather more information than that. In your case a good answer to the first question really ought to say more than birds. It might say who John Audobon is (if the question doesn't already cover that) and might cover things information about what birds he painted, where he painted them and what mediums he used for instance.

The animals question might say that not all birds can fly amongst other things.

Which means that now we have two different answers of course.

On top of that we do split sites into different subject areas so I can't conceive of any site where both your proposed questions would be on-topic.

Answer 3

The basic purpose of the duplicate question feature is to ensure that users don't have to constantly a question that is just a rephrasing of another question. This does not mean that an applicable answer makes the question a duplicate, but it is a strong point in its favor.

For example, a person might ask a more specific version of a general question. If the meat of their specific question is essentially the general question, then it adds very little to the site to spend time providing an answer which will be 98% the same as the general question, with some trivial tailoring to the user's circumstance.

Remember: the purpose of this site is to build a repository of problems and solutions. If the core of their problem has already been solved, then it doesn't make our repository any better to solve a trivially different version of it again.

Let us consider your example:

E.g., the answer is “Birds”

I ask a question, “What was John Audubon’s main art subject?”

The answer in question is rather incomplete, is it not? Wouldn't a more complete answer explain that Audubon's ornithology study was a great influence on his art, which is why most of his artwork centered on birds?

See, the problem you cite only really manifests for bad, incomplete, trivial answers. Those aren't good answers, and we would much rather than more fully formed answers. And once you provide that, the problem you suggest disappears.

But if you want to be pedantic, then two can play this game:

However later someone wants to know “what animals have feathers and can fly?”

Will that question be closed as duplicate?

No, because not all birds fly.

Also, it's highly unlikely that both questions would be valid for the same site.