1. “Simply re-write” is astoundingly naive. There are plenty of examples of re-writes (Lotus 1-2-3 switch from assembler to C, Netscape’s rewrite, KDE4, reference counting in S) that proved far costlier in time and effort than anyone dreamed.

2. Patches to improve the speed of R are a good thing, but not an end-all. I spend far more time developing algorithms, cleaning data, exploring data, trying models than I do in the actual processing time of the data. If you want to talk about optimization, “programmer time” is a much bigger bottleneck than “CPU time”.

3. Shaving milliseconds off CPU time is nice, but shaving hours of head-banging due to terse R documentation and too-cute-to-be-useful examples in the docs would be a far greater contribution to society.

4. I would argue that we need a smaller user base in Computer Science. Statisticians don’t need an elegant, modern, computer language. We need a modern, _simple_ “data analysis environment”.

Personally I would be much happier to see R move to a more permissive license, not more greedy. For my own open-source projects, I like to encourage people to use it however they want, whether that’s creating more open-source creations on top of it, closed-source commercial projects, simple usage as a library, or “stealing” the code for nefarious purposes (in quotes, because what’s given away can no longer be stolen. Once I give it away, it’s not mine anymore.).

I realize not everyone agrees with me about this, but as a community member I wanted to give voice to this – a lot of people at useR! seemed to feel the same way.