6
$\begingroup$

I just want to react to a recent question that was asked about the R rank function: Rank in R - descending order

To my opinion, this is clearly unrelated to statistical analysis and is purely a technical trick, specific of one particular language. From what I seem to remember, such questions were likely to be closed or downvoted one or two months ago. I know this has been discussed in the mean time, but there it seems a little googling would have been fine.

So, my question is: Are we going to open the door to such questions, at the risk of "obliterating" real and interesting questions about statistical modeling?

P.S. I didn't donwvote it but I don't understand why it has been upvoted.

| |
$\endgroup$
8
$\begingroup$

I agree that the question isn't particularly statistical. However I would just answer these questions and if no answer is forthcoming, then point them to SO. Not voting the question up for these types of questions will help too.

I think it's easier to allow a wider range of questions rather than endless debating whether questions are on or off topic.

Also if we close questions too quickly, it can put off people returning to the site - first impressions last and all that.

| |
$\endgroup$
  • 2
    $\begingroup$ That second paragraph really hits the nail on the head for me. $\endgroup$ – walkytalky Oct 5 '10 at 11:59
  • $\begingroup$ Ok, so it seems I'm a little spaced out and misunderstood the purpose of this SE :) $\endgroup$ – chl Oct 5 '10 at 14:14
  • $\begingroup$ @chl: In theory it would be a good idea to bounce computing questions to SO, but in practice it just doesn't really work. $\endgroup$ – csgillespie Oct 5 '10 at 14:17
5
$\begingroup$

We agreed to welcome such questions, mainly because it is just hard to judge what is good and what is bad in this direction and we don't want to assume everybody here have also SO account. Altogether we have "statistical computing" in scope, see FAQ.

Related question: How much programming here?

| |
$\endgroup$
  • $\begingroup$ Indeed, the link with "statistical computing" isn't obvious to me. $\endgroup$ – chl Oct 5 '10 at 11:30
  • $\begingroup$ @chl Because it is not obvious; the real idea is what I or csgillespie wrote, this is rather sort of justification. $\endgroup$ – user88 Oct 6 '10 at 8:24
4
$\begingroup$

As one of those who up-voted - I apologize - I was unaware of the disinterest in R questions. I'm a C++ guy and working on a Stat PhD. I up-voted because I saw it as a different perspective on a programming language which I'm still learning. I would've expected the answer to use an option to the function, ordering=less, or something.

Instead, I think the answer was more statistical than the question because -- I think -- knowing the stats, @ucfagls, suggested simply flipping the sign. Something intrinsically obvious to a statistician thinking about rank statistics, but not so much so to low-level programmer thinking about implementation specifics.

Further, I would suggest you encourage R questions here for precisely this reason. Being a language written with statistics in mind, I think the statistician can provide unique insight that a generalist or low-level programmer would miss. However, I don't think all R questions should be moved here because certain interface aspects or databasing backend questions, etc., would probably be better answered by someone who has the generalist / low-level programmer insight who would probably frequent StackOverflow more.

Noise: I forgot to mention, that with voting, once someone has access to it you are probably going to get a bit of noise in the importance of a question/answer: "+1 - Hey I didn't think of it that way", "+1 - That was cooool.", "+1 for use of the word Nefarious.". It's certainly limited because you have a max of 30 -- which I think is the correct limit -- but you're still going to get a few like me who up-voted and didn't know better or up-voted for the wrong reasons... smirk

If the SO/SE gods were willing, we could consider a multivariate rating scale (perhaps one that's only visible above a certain reputation) so as not to confuse the newbies. Then, perhaps like slashdot, you could set a threshold for fodder below which you have no interest in viewing it. Slashdot has +1 funny, +1 insightful, probably don't want to encourage "funny" here, but what about "+1 algorithm", "+1 perspective", "+1 wave to C.R. Rao for me..."

| |
$\endgroup$
  • 1
    $\begingroup$ No. The general rule is: If you like, upvote. $\endgroup$ – user88 Oct 5 '10 at 16:24
  • $\begingroup$ Ok. Cool. Thanks! $\endgroup$ – M. Tibbits Oct 6 '10 at 14:47
2
$\begingroup$

This is my second day using R. I've had some issues using it, and I'm not sure how appropriate it would be to ask those here, since they are not really statistic questions, but R usage ones. In FAQ there was nothing about allowed topics, I've thought of checking meta and noticed this question :)

For example, after successfully doing some work yesterday, I started R today and I typed in load("test.r"). R complained with Error: bad restore file magic number (file may be corrupted) -- no data loaded. After some googling, I figured it out that I should use source instead. But what if I hadn't find anything on google? How would such a question fit here?

Another issue I have converting some S-Plus code to R: I get this warning 'seriesData' is deprecated. Use 'series' instead. I'd like to know if it's safe to just replace seriesData with series, or if this can lead to some silent problems.

So what I'm asking, should this site be a one-stop shop for R? I think it should be, it's quite unpleasant to subscribe to 'r-help' just to ask a basic question. This is why stack exchange sites were created for. But I can understand why statisticians would be bothered by such questions.

I think the other extreme is mathoverflow.net, where I've seen people with 3000 reputation first request permission on meta to ask a borderline question. That gave me the impression that they view the site as some sort of a temple which they must not dirty.

| |
$\endgroup$
  • $\begingroup$ It is ok to post R questions here. It is also ok to post R questions on StackOverflow, that's why the confusion. $\endgroup$ – user88 Oct 6 '10 at 8:20
  • 2
    $\begingroup$ I'm in the minority and the debate is already closed, but I'm also easy to ignore, so... why not ask such questions on StackOverflow? All of the very good R users here are also there, and there's already a wealth of technical questions about R answered there. If this site should be a one-stop shop for R, why not SAS, SPSS, MLwiN, HLM, JMP, Stata, MATLAB, Octave, Clojure, Incanter, Scala, Hadoop, Pig, etc.? I think that allowing purely technical questions opens the gate to tons of questions with only tangential relations to this site's core purpose. $\endgroup$ – Matt Parker Oct 6 '10 at 16:15
  • $\begingroup$ Whether or not a flood will come through that gate is another issue, and one I can't predict. But, as you say, this is what the StackExchange sites were created for... to have a place to ask every question. Notably, it isn't ONE place to ask every question. So, although my opinion is in disagreement with the policy here, I still encourage you to ask your technical questions on StackOverflow. It's the right forum. $\endgroup$ – Matt Parker Oct 6 '10 at 16:17

You must log in to answer this question.

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