> None of these hashes are cryptographic.

I never said they were.

> Crypto hashes don't have a problem of being slow.. they are designed to be slow.

No they're not. Only password-based key derivation functions are designed to be slow, every other hash tries to be as fast as possible within their security budget.

> The conclusion tries to tie weak/fast hashes to crypto/secure hashes (the https mention), but they're worlds apart.

Actually, the specific link I made between cryptographic strength universal hashing and maximum-speed universal hashing is precisely because they're not worlds apart. They use the same techniques, and stand on the same mathematical background.