The egregious metadata of Google Books is almost all date related problems. Anyone programmer who has done metadata-related work before will tell you that human compiled metadata always needs to be treated carefully, and date related metadata is usually "wrong" when the metadata comes from multiple sources.

Don't believe me? Crosswalk your data to dublin core and then tell me what date it was written, first published and the date the edition you have was published. I bet all those field go to dc.created and in standard dublin-core you have no way to identify which is which.

Now tell me why Google's data is so bad again?

The truth is that for subject-related data it is much, much better that library catalogs, for Author data it is pretty close and for date data the library catalog wins at the moment.

(Speaking as someone who spent a long time working in the field, and has had papers presented at the world library congress and at dublin core metadata conferences)