The University of Michigan has made publicly available 684,597 of its bibliographic records. They're published under a Creative Commons 0 license, which puts them in the public domain.

Why don't we all do that? In part it's because for many libraries, the provenance of the information in their bibliographic records cannot be known with certainty and may well include information received under licenses that forbid re-publication. It's a knotty, irksome problem.