The initial work on R by Robert Gentleman and I produced what looked like a potentially useful piece of software and we began preparing it for use in our teaching laboratory. We were heartened enough by our progress to place some binary copies of R at Statlib and make a small announcement on the s-news mailing list in August of 1993.
A number of people picked up our binaries and offered feedback. The most persistent of these was Martin Mächler of ETH Zurich, who encouraged us to release the R source code as "free software".
We had some initial doubts about doing this, but Martin's arguments were persuasive, and we agreed to make the source code available by ftp under the terms of the Free Software Foundation's GNU general license. This happened in June of 1995.
At this point, the development of R was a relatively closed process. Robert and I (soon joined by Martin) would get bug reports by e-mail and from time-to-time release updated versions of R. We quickly noticed that there was no real forum for users to discuss R with each other and so we began maintaining a small mailing list.
As interest in R grew (mostly by word of mouth) it became clear that manually maintaining the mailing list was not an effective option. Worse than that, at Auckland we were paying for e-mail, and the cost was beginning to become noticeable. Eventually Martin volunteered the use of facilities at ETH Zurich to establish automated mailing lists to carry discussions about R and R development. In March of 1996 the r-testers mailing list was started. Roughly a year later this was replaced with three newsgroups: r-announce, r-help and r-devel.
As R developed and people began porting applications to it, it became clear that we needed a better distribution mechanism. After some discussion it was decided a formal archive mechanism was desirable. Kurt Hornik of TU Wien took on the task of establishing the archive. In addition to the master site in Austria there are a number of mirror sites, including StatLib.
With the introduction of the mailing lists, development on R accelerated. This was partly because we obtained many more reports and suggestions and partly because we also began to receive patches and code contributions. The contributions ranged from fixes for typos through to changes which provided substantial increases in functionality and performance.
The level of contribution was such that Robert, Martin and I couldn't always make changes at a rate which was satisfactory to those asking for changes. As a result, in mid-1997 we established a larger "core group" who can make changes to the source code CVS archive. This group currently consists of:
Doug Bates, Peter Dalgaard,Since all work on R is strictly of a voluntary nature, the organisation is very loose, with members contributing when and as they can.
Robert Gentleman, Kurt Hornik,
Ross Ihaka, Friedrich Leisch,
Thomas Lumley, Martin Mächler,
Paul Murrell, Heiner Schwarte,
and Luke Tierney.
When Robert and I started work on R, we were hopeful that we might be able to produce something we could use to teach our introductory data analysis courses. Had we continued to work strictly on own it is likely that this is precisely what we would have achieved.
The decision to make R free software has enabled us to set rather higher goals, because it has given us access to a large pool of very talented individuals who have been willing to invest significant effort in the project. Indeed, one of the very best things about having worked on R has been the chance to work with such a great group of people.
In addition to the core group listed above, I would like to acknowledge the following individuals who have made significant contributions to R.
Valerio Aimale, Ben Bolker,(I apologise for omissions here. Our record keeping has not been all that it could be).
John Chambers, Simon Davies,
Paul Gilbert, Arne Kovac,
Philippe Lambert, Alan Lee,
Jim Lindsey, Patrick Lindsey,
Mike Meyer, Martyn Plummer,
Anthony Rossini, Bill Venables,
Gregory Warnes, and
mward@wolf.hip.berkeley.edu
In addition a host of other individuals have made contributions.
R is still under active development and there is still some work needed before it can be considered ready for widespread use. In particular, some changes will be required to support moderate to large-sized data sets. More importantly, there is an almost complete lack of introductory documentation, although much of what has been written about S directly applicable to R.
Despite this, it seems that R is beginning to reach the point where it is stable enough for regular use (at least under Unix). I am hopeful that during the next year we can release a complete R version 1.0 package as part of the Free Software Foundation's GNU suite of software.