Originally Posted by Shaggy
From what I've seen, some commercial publishers just scan/OCR a book and then sell it, without doing any error checking either. Some of the worst books I've seen with regards to OCR errors, have been commercial ebooks. It's pretty annoying when I'm paying pbook prices for an ebook, that I have to spend a lot of time "fixing" their book for them before I read it.
And the quasi-opposite seems to be true. By checking what can be seen in Amazon previews and Google Books I've found several instances of public-domain paper books with the same OCR errors (and sometimes lack of formatting[*]) of the Project Gutenberg books. It's quite obvious these paper books were printed from the unverified PG text. I check twice at least before buying a recent edition of a book that's available at PG...
[*] or plain wrong formatting as in using an opening single quote instead of an apostrophe in "'em" or "'tis".