Skip to content
Prev 200 / 15274 Next
Original Raw

R vs. S-PLUS

Ajay Shah 
I am personally an R user, but I believe that one of the reasons which
made me more comfortable with R was that under certain states of
nature, I could drop into a commercial version.

So far, I have generated no revenues for the firm :-) but I do believe
that's possible (under certain states of nature).

I think there's a powerful duality between free software and
commercial software in this kind of situation. I'm reminded of the
time - long ago - when a bunch of guys close to the Free Software
Foundation created a company named Cygnus which gave commercial
support for GNU related software. The company got bought out by Red
Hat later <http://news.com.com/2100-1001-232971.html?legacy=cnet> but
it does serve to illustrate the principle.

Similarly I see a nice duality between linux and solaris - firms can
delicately calibrate how much they want to use free software, and how
much they want commercial software. I think both have a role. I think
both sides benefit from working together.

  Example: in finance, date and time is of great importance, and right
    now, it looks like the two are diverging - R is going the its/zoo
    way, and S/finmetrics have done their own thing. It will be of
    benefit to both sides if a common free software library does
    datetime.

  Example: The Eric Zivot book is great - but it should have pitched
    higher, at becoming something like MASS. That, of course, requires
    a much richer set of public domain libraries, not proprietary
    libraries.

I have thus far had 0 interactions with S. I am actively using R and
love it. I am completely unpersuaded when arguments are made that S is
technically superior. But I do recognise a potential role for having
commercial support for R. Peering into the crystal ball, I expect that
the creative and intellectual part is best done in the free software
community, and the company will work best saying they are "a
commercial supported R", instead of saying they have proprietary
code. They can perhaps be like a stable release, and R can be
risk-taking.