Google search, from very early on, considered it a success metric when users went quickly to a result. I have no idea how that factors into the current surely hideously complex ranking algorithms, though.

As for the parsing of queries, that's probably based on how most users use search. Not everyone is familiar with keyword -based search. I expect they've done tons of A/B tests to determine what kind of query interpretation makes most users get better results. We're just not "most users".