I think a book -- especially a popular book in STEM -- can be intellectually useless/misguided without making any specific factual errors. To be interesting to a general audience the book has to focus on a small number of exemplars so you are relying on the author not merely to correctly cite the facts but also to pick exemplars that happen to illustrate a true regularity in the complete data.

It's really easy to write a popular book that supports the narrative that religion is the cause of a a huge fraction of suffering in the world and equally possible to write an equally factual and persuasive book arguing the opposite.

Ultimately, that was my sense of Gun’s, Germs and Steel. Its thesis might or might not be true -- I've got not truck with the can’t think about things racists might like idea -- but the book doesn’t actually give me much reason to believe it.

It's the same sense I get from books in areas I am an expert in which -- despite being factually accurate -- can give a completely wrong metaphor/theory or a great one.