e-book typos

[groooooove]

anyone else been encountering them?

for my kindle app i've read three books thus far- two free ones (100% typo free, both bio/quotes about Beethoven) and one that i paid for. the one i paid for, (harpo speaks! by harpo marx) is littered with typos. not misspellings, obvious "my finger slipped" or drunk text-like typing.

i've also got a nook that i haven't read any full books on yet (twas an xmas gift) but a couple of the freebies i was scrolling through were filled with strange typos and lines of "*^%F)HD(&" type of stuff randomly (walt whitman leaves of grass i recall having this, and several others that i cant remember right now). i paid for one book (a $.99 copy of Da Vinci's notebook excerpts) for the Nook and it appears to be typo free.

anyone else see this? some of these books i've mentioned are fairly un-obscure and it really surprises me that theres so many errors. especially in harpo's book, they're terribly common- one every other page, at least.

[Deathblade Eric]

Yeah, I've noticed this too.

As far as I can guess, the problem is greater with books published in the 'pre-digital' era. These have to be scanned & then run through a text-recognition program. The next step ought to be to pass a spell-checker across it, but I don't reckon that always happens.

If you've ever messed around with any OCR package, you've probably seen what a mess they can make of scanned text; not just outputting garbage, screwing up paragraph breaks & ignoring hyphenation, but also generating perfectly valid words that don't make sense in context... 'arid' is a common result instead of 'and', so is 'nun' instead of 'him'. Spellcheckers will miss those.

To get decent output from OCR means proofreading the document - which I guess many providers of free versions will be loath to do as it means paying someone.

Theoretically publications typeset in programs like Quark & InDesign should be fine, but I found that generating an EPUB file from InDesign CS4 messed with the formatting & I still had to edit the output to get it into a usable state for my e-reader. Hopefully CS5 is better.

I guess it's the price we have to pay for being in near the start.

A quick Google for the Harpo book suggests it originally came out in 1961, so has probably fallen foul of crappy OCR.

Pete.