[tsotha]

Real evidence would require a double-blind test, but there is already such a large productivity difference between programmers you'd need a pretty large sample.

And what possible metric could you use to determine if the code is "better"? That's a really subjective thing.