State of R

R is terrific and very successful

Fostered and engaged a large community providing (mostly) good software for cutting-edge statistical techniques.

Development of R itself is stagnating and intellectually less interesting and attractive than it was 10 years ago.

Many opportunities to do much better
but large user base, different goals, difficulty of coordinating and personalities makes this less likely.

Many obstacles to changing R through patches, etc.

How to get the excitement about fundamentals back into stat. computing and face the challenge of handling large data fast and doing qualitatively new things?

I've wrestled with this for over 10 years, and the following is a new compromise strategy for consideration!

The goal is to enable some good people to do low-level work that can extend R by working around parts of R, but leveraging many of the good parts.