LuaJIT is great and it's very interesting. (In fact, I've been reading through its source code trying to understand the magic.)
My impression is that the developer is reading the important papers and putting the theory into practice. It's innovative in the sense of "new stuff that is made useful".
Guys like the SPUR team at MSR on the other hand are trying new ideas and writing these papers. That's innovation in the sense of "a change in the thought process for doing something".
Both are important and each one would be useless without the other.
My impression is that the developer is reading the important papers and putting the theory into practice. It's innovative in the sense of "new stuff that is made useful".
Guys like the SPUR team at MSR on the other hand are trying new ideas and writing these papers. That's innovation in the sense of "a change in the thought process for doing something".
Both are important and each one would be useless without the other.