tl;dr: Norvig argues that language is best modeled statistically, Chomsky's model is deterministic. He also disagrees about what makes good science (usefulness for engineering is a valid criterium & gathering facts over developing theories).
[1] http://en.wikipedia.org/wiki/Artificial_Intelligence:_A_Mode...
But Chomsky's contribution to linguistics in general and computational linguistics in particular goes well beyond anything like that, e.g. [1]. I'm not sure, but I suspect that linguistics and computational linguistics where unrelated fields before Chomsky tied them together.
(that does not mean that Norvig may be wrong and Chomsky right on that point you are quoting. It seems you're suggesting that Chomsky is not a great linguist because he disagress with Norvig - that would be a mistake).