ok, back to some real work... p3 has finished with "the art of the book". (proofing goes _fast_ at d.p. it's formatting, and then especially post-processing, that bogs down.) again, this book consists of roughly 2500 lines... ~500 were changed by p1, then ~100 by p2, and now roughly another ~100 were changed by p3... while this p3 count is _much_ higher than normal, reviews of the diffs show that most of the changes were on captions, which were typeset as all-upper. o.c.r. is notoriously bad on words in all-uppercase. so any post-processor would know that those lines needed to be checked closely anyway, so it probably doesn't matter if that close check is done during p3 or during post-processing. in fact, since p3 people just "leave a note" instead of _deciding_ the matter, it's probably less efficient to have p3 leaving a note. this is especially true if you consider the likelihood is high that _both_ p3 and the post-processor will do the "close check", and thus duplicate the effort... but again, this thread isn't aimed at d.p. efficiency. here are 2 actual _errors_ that were caught by p3.

Home has designed three founts, all of them Horne has designed three founts, all of them

so rich. Characteristic or his art is the headpiece so rich. Characteristic of his art is the headpiece

those were the _only_ o.c.r. word errors that p3 found, as far as i can tell. they caught punctuation errors too, but i didn't think to count them. but all in all, i'd say they probably caught a couple dozen errors p2 didn't... here's a place where p2 corrected a printing-error, but then p3 reversed that, _reinserting_ the printing-error, and then "left a note" that there was a printing-error.

It has been acquired I thas[**It has] been acquired

that probably seems silly to you... i'd guess it was done so the post-processor could note the change. ("note: a printer's error was corrected on page 206.") i've always found such notes to be counterproductive. ("why did you waste my time telling me that you fixed an obvious error? inform me if you _don't_ fix them.") in general, my opinion is that these _notes_ are bad. it takes time to _insert_ them into the text, and then it takes time to _remove_ them later. that's stupid... one of the innovations that roger tested on his site was _separation_ of the notes outside the text itself. he found that it worked well, saving time and energy, and providing general improvements to the product. of course, as it was with all of his findings, the d.p. "powers-that-be" chose simply to ignore it totally... *** oh yes, sorry to keep you hanging, p3 _did_ find and fix those 2 spacey doublequotes, so kudos to them for that. *** anyway... because of the large number of all-uppercase captions in this book, which we know require individual attention, this book probably is not a good candidate for my system. every system will do equally well (or badly) if you're gonna have to look at each line individually. don't need a system. my system _would_ do a good job on the hundreds of pages in this book which do _not_ have all-upper captions on 'em. many full pages went through all _3_ of the proofing rounds without a single character being changed on the whole page, so that was a bunch of time and attention essentially wasted. any system which focuses attention where it is _needed_ will be able to have a cost-benefit efficiency that is much better. -bowerbird