# New proposal at Area51: R statistical computing

I think we could use a site dedicated to programming in R (similar to the Mathematica SE). I went ahead and made a proposal in Area 51 (link).

• This has been discussed some time ago, see R-specific StackExchange site or greater integration of R community within CV? (with follow-up comments). – chl Apr 28 '15 at 16:05
• There are, of course, a great deal of R programming questions already on Stack Overflow. It seems to me the issue is of greater concern for them than for us. Are you thinking the existing SO Qs should be migrated to the new R.SE? How do you see the new site being related to SO & CV generally? – gung - Reinstate Monica Apr 28 '15 at 19:27
• @gung I think the situation would be analogous to mathematica.stackexchange.com. They also had a lot of questions on SO, which eventually got migrated to mathematica.SE. – becko Apr 28 '15 at 19:35
• I'm open to the idea of an R.SE site, but I'm ambivalent about it. I think it would create more confusion than it's worth to have R Qs on CV, R Qs on SO, & an R.SE site. If the R Qs on SO were all migrated to the new R.SE, & we could maintain an otherwise identical arrangement--that would mitigate against the 3-site confusion. OTOH if the new site is really the same as R on SO, there is some question of why making the new site is worth the effort & (SO) disruption. – gung - Reinstate Monica Apr 28 '15 at 19:43

The issue of an R-specific SE now appears to be moot, as judging from the link the proposal has been closed with the message:

Announcements

The R community embraced Stack Overflow as The place for R support and went on to gain a strong following on our Cross Validated site. – Robert Cartaino♦ 2 hours ago

(Robert Cartaino is "the Director of Community Development for the Stack Exchange Network".)

• I have become so used to the nifty "edit" button I even had a glance to see where it had gone so I could change "embrased", but of course there was no such option :-) – Silverfish Apr 28 '15 at 19:57
• I am intrigued here. How can one user close a proposal like this? (I have serious doubts about the merits and viability of this idea, but that's not my point at all.) – Nick Cox Apr 28 '15 at 19:58
• Me too; embrace your favourite dictionaries! – Nick Cox Apr 28 '15 at 20:00
• It's like saying that SAS users embraced the use of CARDS statement to input their data. Yes, it's there; but it does not mean that it's the best solution to the problem. – StasK Apr 28 '15 at 21:05
• I'm thoroughly confused--exactly what "went on to gain a strong following on (CV)"? Is Cartaino saying CV is the place for R questions? That it's not? That CV supports keeping R questions on SO? – whuber Apr 28 '15 at 22:08
• @whuber Other than a vague sense that he is seeking to avoid splitting up R on the SE system any further, I'm not sure what he means. (But he really did write "embrased"!) – Silverfish Apr 28 '15 at 22:19
• Maybe he was trying to type "embarassed" and it just looks like he meant "embraced". – Glen_b Apr 29 '15 at 5:14
• I don't think being "Director of Community Development" marches well with just squashing the proposal immediately and unilaterally. How much community involvement is that in discussing the merits of a proposal? (Or, if this is not what it seems, it should have been better explained.) – Nick Cox Apr 29 '15 at 9:23
• @Glen_b: Or "embarrassed"? – Nick Cox Apr 29 '15 at 9:57
• @NickCox I typed it that way but the built in spellchecker that's running on this machine underlined it in red. I changed it until it stopped complaining. I mistakenly assumed it knew better than me. I presume one-r is US spelling. – Glen_b Apr 29 '15 at 10:13
• @Glen_b How embarrassing. (I won't milk this any more....) – Nick Cox Apr 29 '15 at 10:15

StackOverflow defines the following on-topics:

• a specific programming problem, or
• a software algorithm, or
• software tools commonly used by programmers; and is
• a practical, answerable problem that is unique to software development

SO does not define programming per se, so we'd have to go elsewhere. According to Wikipedia,

Computer programming (often shortened to programming) is a process that leads from an original formulation of a computing problem to executable computer programs.

OK. So how many R questions in SO are asked in the context of developing executable code, or at least in the context of developing a standalone, self-contained R package? How many are asked about the software algorithm (which, in case of R, would probably be a statistical method, and hence more on topic at CV)? How many are asked about IDEs for R? There are some, I imagine, but mostly the questions are along the lines of "How do I tweak ggplot to do this" or "How do I manipulate data.table to get that". None of that leads to executable code; at best, you can say that the code being created is batch code, i.e., simply sequences of R commands and functions, although you could probably refer to some of that code as scripts if they involve control flow and/or input parameters.

Hence, I come to the conclusion that R questions have been off-topic on SO to begin with, for all these years, unless they deal specifically with development of R packages (checking the inputs, establishing dependencies, interacting with the S4 object system, etc.). In other words, R community should be grateful that their stream of off-topic questions is being tolerated on SO. Having a dedicated site would have been a better idea. I feel really weird every time I flag R questions and send them to a website that primarily deals with OOP in Java and C#.

I would feel the same way about other statistical packages like SAS and Stata. Just like in the case of R above, you don't create reusable executable code in SAS or Stata to solve problems with arbitrary inputs (although you could, of course, write SAS MACROs and Stata programs), but rather to deal with your single existing data set trying to answer your specific research question. Judging from CV activity in these tags, these software titles are all but dead -- not an impression you would get from their annual user conferences, though, or from the expensive looking booths at professional conventions. The users of these packages, however, are but a very humble minority on CV -- may be because the strong emphasis on R intimidates them here; but mostly because the documentation in these packages is so much more extensive than the typical R listing of available options, and because the tech support type questions are answered in other online communities that SE system failed to incorporate -- as well as by the real tech support that these commercial software vendors do provide. (IMHO, Statalist should have migrated to SE platform when they closed the mailing list format a year or so ago, instead of forming a traditional forum on which I miss everything that makes the SE platform so productive like edits, extensive tagging and multiple tiers of response. While the R questions get bounced to the Java/C# SE site, Stata questions are getting bounced away from the SE system to statalist.org.)

• I am not an active SO user, but your interpreation of what is (or rather should be) on-topic there seems to be pretty restrictive. Think of any specific question in any specific programming language and google it: e.g. python remove element from dictionary -- the first hit is often to a SO thread that answers the question. There are myriads of similar questions on SO. How are they more "in the context of developing executable code" than a question about tweaking ggplot to achieve something? – amoeba Apr 28 '15 at 23:27
• On Statalist: As @StasK knows, I was party to the decision to relaunch as a separate web forum. There were, and are, some minuses as well as pluses to the decision, but the most important point is that people wishing to use SE for Stata questions were and are free to do so. – Nick Cox Apr 29 '15 at 9:28
• I would echo the reservations here about the kinds of questions asked about statistical software on SO. Many of the questions asked about Stata on SO are not programming questions at all: some don't even include any code but are just "How do I do X in Stata?". It's only because SO operates as a set of separate forums which largely police themselves that many of these questions survive at all. – Nick Cox Apr 29 '15 at 9:55
• @amoeba, Python leads to an executable code, eventually. People learn Python to write programs that deal with pretty much any area of application. However, I am yet to hear about anybody having written, say, a web server that runs in R. (This is of course about me as an occasional R user, at best; may be such webservers do exist, and may be Univ of Auckland Dept of Statistics website is being run in R...) – StasK May 4 '15 at 15:27
• @NickCox, people who are not aware that CV/SO exist cannot be in the category of those wishing to use them to ask Stata questions. Arguably, this could be classified as a failure to self-educate of what the existing learning resources are. However, one indicator of the lack of visibility of statalist.org is that, unlike say "How do I do X in LaTeX?" questions that invariably land on latex.se, "How do I do X in Stata?" are unlikely to land on statalist.org, although SO/CV won't be among the top links, either. – StasK May 4 '15 at 15:36
• @StasK Not clear of your point. But Statalist posts are of the order of 100 posts per day while Stata posts on SO are more like 2 per day. Which is evidently more visible to Stata users? – Nick Cox May 4 '15 at 15:41
• statalist is visible to the "proper" Stata community -- those who come to user group meetings/conferences, subscribed to the Stata Journal, and take short classes from Stata. Beyond that, there's an ocean of students on whom Stata was dumped by their economics professors without explanations, and who have to find some answer somewhere. 100 posts/day is a very solid activity, likely comparable to R tag activity here, and probably would have satisfied the healthy beta requirements quite easily. – StasK May 4 '15 at 15:55
• Statalist is intended as a one-stop, one-shop forum covering Stata and statistics. I don't think that would work here as part of SE, but anyone is welcome to try to propose that. Judging from the thread here a proposal would not get far. I wouldn't want to work on the criteria for Stata on SO vs Stata on CV vs Stata somewhere else. – Nick Cox May 4 '15 at 16:01