The availability isn't the problem, it's the execution. Google could quite easily license the already accurate information from the participating libraries or from OCLC but didn't. And while the article is from the perspective of a scholar, the majority of users are likely to be students, a user base that may not be aware of or understand the implications of the errors until after their paper is finished. And remember that these books are mostly coming from research libraries. These will mostly be scholarly books of primary interest to students and faculty.
The fulltext search of millions of books is certainly useful and, as a reference librarian at a research institution, I often encourage students to use it to discover if "there is a book out there on that subject" but the inaccuracies do really matter! A small example of why is the case of books with only a snippet or less available in fulltext. Our students, and anyone with a public library nearby in the US for that matter, can request the book through interlibrary loan (almost always for free). But to make a successful request, you need accurate information about the book you want to see. See where the Google's lack of metadata accuracy could be a problem?