LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (March 2000, week 2)Back to main SAS-L pageJoin or leave SAS-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:         Tue, 14 Mar 2000 10:25:35 -0800
Reply-To:     kmself@IX.NETCOM.COM
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         kmself@IX.NETCOM.COM
Subject:      Re: SI announce SAS for Linux 4Q2000:wq
Comments: To: Fabrizio Carinci <carinci@cmns.mnegri.it>
Comments: cc: Tim Churches <TCHUR@doh.health.nsw.gov.au>
In-Reply-To:  <no.id>; from carinci@CMNS.MNEGRI.IT on Tue, Mar 14,
              2000 at 11:29:23AM +0100
Content-Type: text/plain; charset=us-ascii

On Tue, Mar 14, 2000 at 11:29:23AM +0100, Fabrizio Carinci wrote: > Dear SAS-Lers, > I would try to break the Karsten-Churches oligarchy by adding my small > impression to the SAS-linux thread from a biostatistical point of view.

Cool, I've always wanted to be part of an oligarchy.... <g>

Actually, commentary is what we'll be looking for in spades, though we're trying at this point to create a departure point for the discussion.

> In our work, the position of SAS is almost monopolistic, i.e. it is > straightforward to have the computational aspects of a paper accepted by > using (and citing) SAS and its well known procedures. > > However, my impression is that this WILL NOT be acceptable anymore in the > near future, and that the change will be quite rapid. > The fact is that the 'open source' philosophy is not corresponding to > 'freeware' philosophy:

Crucial point. Misunderstood less these days than it has been in the past, though I still shake my head at the confusion which remains, far too often.

> - the point for scientific acceptability is that any scientist should > have the possibility to control the source, sometimes to evaluate > strengths and weaknesses, some other to better understand what's going on > or fit the procedures to his needs (even in terms of execution times). > - the reason for the rapid change is that open source has been found not > equivalent to condemn themselves to poverty ! Netscape, RedHat and SUN are > growing fast while they are giving away their secrets.

Open source is at least a viable leg of a business venture, though the model is still shaking itself out.

> Thus, SI will find their strategy outdated very soon if they will > continue to ignore the true 'open source' meaning.

Among my fears for SI.

> By the way, there are great options available. and they are free. > These include sql for database processing, java and gcc for data > analysis, php, perl and so on.

...and much, much more.... I'd throw in R and MySQL for starters.

> Such software is conceived for distributed processing, where SI is just > emulating or adding stuff to their products, that may perhaps only > significantly slow down processes, relative to the existing alternatives > (not to comment the price). > > And, SETINIT stuff will appear simply ridiculous to the Linux community !

"Hello, my name is Leenus Toorvalds, and your Leenux license has just expired...." Not!

> Question to SAS-Lers interested in this thread on new technologies: > do you want to use your SAS background as a knowledge-base > to develop a statistical use of new tools (which is a good point to avoid > jumping on other lists), or do you consider this thread just to send > messages to SI, such as "unveil the code" ?

I'm hoping the former. As I said in an earlier reply, IMO SAS's legacy is hurting it critically, for a number of architectural reasons. I'd prefer a clean break. Among the background activities we've been engaged in are putting together the resources necessary to organize a distributed development project, thankfully much easier these days than even a few months ago. This includes web space, online sources, discussion lists, etc.

> Question to Tim and Karsten: are you aimed to building "new" > SAS-resembling software, or addressing developing collaborations to look > for modern solutions?

Much more the latter. We don't ourselves have the resources to build a ground-up SAS-alike system, don't see it as a positive goal, and find that there are existing tools which would be much better integrated, than trying to reinvent the wheel. However there will be issues of developing code and defining the system itself.

> Statisticians of the next century should be able to build their own > software by interfacing different software languages - again, looking at > network functionalities: no future without distributed information > systems.

Another key point.

Answering an issue raised in another followup, I do see regulatory-driven use of SAS as one of the large remaining market holdouts, together with the small (but significant) sites with a large, established SAS code legacy which needs to be maintained. However even in areas as regulation driven as the pharmaceutical industry, alternate tools, particularly S-Plus and R, are making inroads due to their advanced statistical functionality, extensive user-developed libraries, and the ease with which complex analyses can be developed. The write-only nature of the code is a problem (if you thing statistician-written SAS data steps are ugly, try S-Plus code....). Initial adoption appears to be by individuals, and particularly academically-associated affiliates. Organized evaluation programs are beginning to appear.

> Again, it is not impossible to imagine the 'big bang' of SAS unveiling > the code. Indeed, this would be an impressive marketing strategy, and it > would give great chances to improve their products and revenues. > I don't think this would depress their finance. Big corporations > will continue to pay for support, while thousands of young users will join > us.

While I see our project as a challenge to SI, I also see it as something which they may wish to evaluate as a potential business strategy moving outwards. There's a dog and a tail here, and I'm not sure who's wagging whom.

The rumblings from Cary over the past few years have been that SI is looking to become much more a services company. As such, releasing source would be a big plus. Would independent SAS consultants *really* like to go head-to-head with a SI Consulting arm which has exclusive access to source and developers? I think we've found *that* primrose path leading to Redmond. And, frankly, it would be very useful to have SI's development talent behind the project. While we haven't actively approached SI -- and given the last two years' campaign to get a Linux version out it would likely take fer freekin' ever -- we've kicked the idea around.

> Unfortunately, SI interest in meeting statistical science needs is > not clear enough at the moment.

SI have become somewhat de-focussed in the past decade, yes.

> Hope this will add 100 Italian Lire to the discussion.

Thanks much.

> Ciao > Fabrizio > > =============================================================================== > Fabrizio Carinci, Statistician > Head, Unit of Statistics and Information Systems, > Department of Clinical Pharmacology and Epidemiology > Consorzio Mario Negri Sud, Strada Nazionale-66030 S.Maria Imbaro (Chieti)-ITALY > Web: http://www.cmns.mnegri.it > ------------------------------------------------------------------------------- > TEL: 39-0872-570265 FAX: 39-0872-578240 E-MAIL:carinci@cmns.mnegri.it > =============================================================================== > > ------------------------------

-- Karsten M. Self (kmself@ix.netcom.com) What part of "Gestalt" don't you understand?

Scope out Scoop: http://scoop.kuro5hin.org/ Nothin' rusty about Kuro5hin: http://www.kuro5hin.org/


Back to: Top of message | Previous page | Main SAS-L page