I think you mean "less-than-C" performance, no? I'd say it's pretty close to optimal performance - it's just that the optimum is not that high compared with a stricter language, or indeed a compiled language, which are able to exploit far more invariants (especially in the latter case).

Anyway, I think his point was fundamentally correct, but it just seemed like an odd choice to single out one of the world's best interpreters - and one which is fairly competitive even with fast compiled languages - as your example. If he'd chosen CPython, I'd have been with him all the way, haha. (As someone who's written a faster Python interpreter, ...don't even get me started on that one.)