On the question here, it was eventually discovered that this question was a duplicate; however, if we go through the same process that the OP has done and craft a question as they have (with the same title that is), the possible duplicate (which is practically a near-perfect duplicate other than the earlier question not giving a hint as to the limit) does not show up in the recommendations. Even the seemingly minimal search phrase

"\lim_{n\rightarrow\infty}a_{n+1}-a_{n}"

does not recommend a previous question slightly related. In all reasonableness, if I were asking this question I would not have believed I were asking a duplicate question. Furthermore, if I take parts of the old question and search for questions with the same string, the new question doesn't pop up.

Even though I have the accepted answer on the duplicate, I wouldn't mind if the question was closed as a duplicate and a moderator would merge the answers (I hate duplicates). However, the old question with the same title would remain as blind to the new question with the same title, and I imagine the question would creep up again. They're both reasonable titles and formulations.

I further imagine that this certain predicament happens occasionally.

What does the community think should be the course of action here? When a question is, strictly speaking, an exact- or near-duplicate of a previous question, but both questions are effectively blind to each other through the search algorithm.

Jeff Atwood has a recommendation, and this the first time I've thought his argument has some merit.

  • Once a question is marked as a dupe, it will be forever marked as a dupe, unless otherwise altered, so upon marking one as a dupe, anyone finding either one will find the other. Besides, the "Related" column of suggested similar questions has long been known to be faulty. Go with quid's answer below. There's a better, smarter, and more likely successful search method than trusting the "Related" lists on the right-hand side-bar to identify hits. – amWhy Sep 5 at 22:43
up vote 5 down vote accepted

Blind or not, we close as a duplicate. Duplicates are not deleted or suppressed from searches. I cannot see what advantage there is in not marking as a duplicate now. Let me add, to mark one of them as a duplicate now. One might discuss which one.

The only thing that is prevented is that more answers are given to one of the two versions. Yet, if somebody sees it now and has a different approach they could post it on the other version.

The way I read Jeff's blog post is that he argues for not agonizing about a new duplicated being on the site, for not berating the poster of the question for not having found it or everybody that might have answered the duplicated, etc. (Because on the one hand it is hard to avoid that it happens and on the other hand duplicates also have good aspects to them, such as enticing anew that a problem is considered and maybe generating new and good answers.)

So, we won't do any of that, yet of course we still mark one of the questions as duplicate when somebody finds a duplicate.

Tangentially, I found this question near-instantly with https://approach0.xyz/

$\lim_{n \rightarrow\infty} ~ x_{n+1} - x_n= c , c > 0$ . Then, is $\{x_n/n\}$ convergent?

Or somewhat further away but maybe still sufficient to extrapolate

If $\lim_{x\to\infty} [f(x+1)-f(x)] =l$ then $\lim_{ x\to\infty}f(x)/x =l$ ($f$ is continuous)

and likely I'd could dig up still more.

You must log in to answer this question.

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