Wouldn't another factor in the lessening of books on sql be that it's a mature technology, which hasn't changed all that much in 30 years?

I mean really, select-from-where-groupby is the bulk of it. Sure there's a jillion fine points like clustering, replication, and what-not. And numerous incremental sexy additions like merge, etc.. But the bulk of what's needed for day-to-day work can be perfectly well picked up from a sql book 30 years old, and a month or two of hands-on cussing.

Just one factor among several, I would think.