13
$\begingroup$

It seems that questions on analysis of particular algorithms are deemed superfluous in the view of the community. Although many of them are probably routine, some are not, and the reference function is not helpful for solving them.

I suggest setting a new standard for closing such questions: only vote to close a question if you are confident that you know how to solve it.

A case in point is a question which was recently closed, although I would bet that none of those who voted to close it know how to solve it off hand.

$\endgroup$
  • 1
    $\begingroup$ thx for bringing this to attn & caring enough to do so. one-size-fits-all policy is inflexible. am in favor of flexibility & exceptions. zen questions exist even in STEM fields. ps whats the asymptotic complexity of SAT? :) $\endgroup$ – vzn Oct 4 '15 at 19:12
  • $\begingroup$ That'd be a great rule-of-thumb for any closing vote. $\endgroup$ – Ran G. Oct 4 '15 at 22:22
  • 5
    $\begingroup$ No. The asker has to show why their question is not routine (i.e. a duplicate). Why should I waste time to try the routine approaches if they alread did? Close as duplicate, reopen upon clarification. $\endgroup$ – Raphael Oct 5 '15 at 10:26
11
$\begingroup$

The proposed standard requires five people to verify that the standard techniques all fail or, at least, are difficult to apply. That's a fairly large fraction of the amount of work needed for those five people to answer the question itself.

The burden is on the asker to demonstrate that the question is suitable and, really, it doesn't matter if they do that in the initial version of their question or after it gets put on hold. There's nothing wrong with putting a question on hold and then reopening it once it's improved.

However, it might be useful to have a template comment along the lines of "Most questions of this kind can be answered using the techniques from our reference question. Please try those and, if they don't work, edit your question to explain what you tried." That would be friendlier than just closing as a duplicate, which quite a lot of people seem to object to on grounds of, "Hey, my question wasn't explicitly answered there!"

$\endgroup$
  • $\begingroup$ A second, specific +1 for the template comment idea. $\endgroup$ – Luke Mathieson Oct 26 '15 at 4:55
8
$\begingroup$

Thank you for this helpful post! I agree it shouldn't be closed as a duplicate of the reference question if the techniques in the reference question aren't sufficient to answer the question.

However, I have a slightly different perspective about what lesson to learn from this example. I do consider asymptotic analysis of an algorithm to be on-topic, if the question is not a duplicate of our existing reference question. However, it seems to me the burden is on the poster to show their work and demonstrate that the techniques found in the reference questions don't work for their particular problem.

Examples:

Bad question: "Here's my algorithm, what is its running time? I haven't tried anything / I don't know how to start." ==> will often get marked as a dup of our reference question

Good question: "Here's my algorithm, what is its running time? I've tried each of the following standard techniques and here is why each one doesn't work." ==> not a dup

It seems to me that the problem with the question you gave is that there was no evidence in the question that the poster had tried any of the standard techniques and no indication of any barrier to them working, and in the absence of that, it seems like trying the standard techniques is the obvious answer.

So, while I do see your point (and I'm glad you caught the issue with that particular question!), I'm not sure a new standard is necessary. We already ask that people should do research before asking and show us in the question what approaches they've tried and rejected and why. If people do that, I don't think this issue will arise. And in my mind this kind of question is a rare case, so I'm not convinced it is frequent enough to warrant a major change in policy. Just my sense.

$\endgroup$
4
$\begingroup$

I am not for or against. But I want to state a point:

I think we should keep in mind that we can refer people to go read the textbook and understand it and solve their problem in place of asking on CS.SE if we push the idea of closing because there is a general reference.

For a reference question to really work when closing a question it should be really easy for the OP of the question to read the reference question and its answers and understand how to use them to solve their own question. If it is not so then it really does not answer the OP's question, so it probably shouldn't be closed as duplicate. If someone has read the reference and still has not got the answer to their question then I am not sure if it is right to close the question as duplicate. Ideally the OP should explain (either in their initial posting or by an edit after closing) why the reference does not answer their question.

$\endgroup$
  • $\begingroup$ A minor comment: "If someone has read the reference and still has not got the answer to their question then I am not sure if it is right to close the question as duplicate. Ideally the OP should explain (either in their initial posting or by an edit after closing) why the reference does not answer their question." - I've always assumed that was basically our current policy. However, my sense is that it's very rare to see a question like that. If anyone has specific examples of this kind of question where it remained closed but shouldn't have, that would be interesting to see. $\endgroup$ – D.W. Oct 19 '15 at 6:47
  • 1
    $\begingroup$ Regarding close as standard reference: when it was proposed on Meta.Stackexchange, it was rejected with "status-declined". I suspect an individual site could still establish this as a site policy if there was consensus to do so. Just sharing the link so you can read the arguments that were presented on both sides, when it was proposed there. $\endgroup$ – D.W. Oct 19 '15 at 7:26
  • $\begingroup$ @DW, that discussion is more about external sources of enlightenment. Here we are talking about pointing at local reference questions. Different kettle of fish. I still upvote your comment, looking at general discussions is certainly needed here. $\endgroup$ – vonbrand Oct 25 '15 at 2:24

You must log in to answer this question.

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