So, for instance, you can leave _Applied Cryptography_ thinking that it's reasonable to use RSA to encrypt sporadic small semantically-meaningful messages, or to deploy a block cipher in ECB mode, or to base a protocol on Diffie Hellman without a trust anchor to break ties for MITM, or to fail to authenticate your ciphertext... and that's just the stuff you notice when you constrain your attention to the mainstream topics in the book, which obviously spends a whole lot of time touring marginal algorithms and protocols without providing clear signals about their unsuitability for deployment.
Another way to look at it is that _Practical_ is a book that is fighting against developers building their own ad-hoc cryptosystems out of things like AES, and _Applied_ is a book that is fighting against developers building their own block ciphers. And if "not replacing AES" is the battle you're fighting, you're already totally screwed, I guess.
When I think about Schneier's contributions to the field, _Applied_ and _Practical_ don't count; they're books, not papers. I'm suggesting instead that you look at his cite record. There's significant stuff there! But it's virtually always as part of teams of otherwise well-established cryptography researchers, and it's clearly not of the same kind as say Joan Daemen or Daniel Bleichenbacher or Don Coppersmith or Philip Rogaway. And I don't mean that in a "Schneier's no Rogaway" sense!
† (or, at the time, IDEA and DES-EDE)
To me, Applied places Schneider into a very separate category from ESR, whose open-source contributions actually were _very_ modest and not at all significant.