S is a language defined by its implementations. While many books and
articles describe the S language, the fact that there are differences
between versions of the two major implementations of S (both in R and
in S-PLUS) means that any description may be incomplete when written
and is soon outdated. This lack of uniformity may also be a barrier
to wider adoption of the S language for statistical analysis. In this
panel discussion, members will discuss implications of a canonical
standard for the S language in the style of Kernighan and Ritchie for
C. Points for discussion include:

* Differences in implementations of S in R and S-PLUS
* How could a standard help (or hinder) more widespread adoption of S
(in all its implentations) for statistical data analysis?
* What is core language, and what is library?
* Balancing a consistent definition of the language with the needs
for continued research and innovation
* Possible processes for defining and implementing a standard