|
On the subject of OCR-induced errors ...
This isn't unique to ebooks! It's not uncommon for publishers who are issuing a reprint of an old title to obtain a copy of the previous dead-tree edition, razor it, feed it to a scanner, and OCR the results -- especially if the author can't provide an up-to-date corrected electronic copy of the original manuscript with copy edits added in. (As if!)
Worst case I've heard of ...
$PUBLISHER obtains the reprint rights to Big Name Author's out-of-print novel. The office intern is given the job of checking the output from the OCR program. Thinking they're smart, rather than actually proofreading the book, the intern feeds the OCR file to Microsoft Word's spelling checker and clicks "accept" on every suggestion. The results then go to typesetting, get turned into a Quark file, and end up in print before anyone senior looks at the book and realizes what's happened. Consequence: a pulped print run. (The bit that stuck in my head: every instance of the word "soldier" had been "corrected" to "sold her".)
|