Google scanning 27'000 books per day - at least, says Economist

[Alexander Turcic]

The Economist runs an interesting story according to which Google is scanning the staggering number of 27'000 books on average per day:

Google will not divulge exact numbers, but Daniel Clancy, the project's lead engineer, gives enough guidance for an educated guess: Google's contract with one university library, Berkeley's, stipulates that it must digitise 3,000 books a day. The minimum for the other 12 universities involved may be lower, but the rate for participating publishers is higher. So a conservative estimate has Google digitising at least 10m books a year. The total number of titles in existence is estimated to be about 65m.

With the immense amount of digitized books, the author contemplates how people will read the book in the future.

As books go digital, new questions, both philosophical and commercial, arise. How, physically, will people read books in future? Will technology “unbind” books, as it has unbundled other media, such as music albums? Will reading habits change as a result? What happens when books are interlinked? And what is a book anyway?

At the end of the day, he doesn't believe that e-books will replace print books, but that, like paperbacks and audiobooks, they are a form that is here to stay.

[Steve Jordan]

"What is a book, anyway?"

Does anybody really know what a book is? Does anybody really care? If so, I can't imagine why... we all have read enough to cry.

Anyway...

Maybe if everyone didn't think about replacing print books, and thought about augmenting Literature instead, we wouldn't be debating the emergence of e-books at all.

[Azayzel]

That's a staggering number a day to be digitizing, wonder how much their cost expenditures are (esp. considering when they get into scanning really old books that have to be delicately handled)? Now let's just hope that the "books unbound" remain that way and will be updated to new formats as they evolve; i.e., don't get stuck in a format that's unusable/inaccessible in the future.

There are quite a few issues I can think of off the top of my head that will keep books in print long after new formats emerge/evolve, what would be nice is if they printed a set number of "master" copies of a given book on some highly rugged or nigh-indestructable medium that would ensure it will be around for generations to come. I mention this because I'm a mild book collector and I've quite a few in my library that it wouldn't be feasible to leave out for the young 'uns to play with. It would be a shame to see a work disappear due to aging... and back to the thread, Google will at least keep most of this accessible to people with online access and the right credentials.

[Steve Jordan]

I'm guessing that, because they have college libraries (and therefore probably students) doing the work, the cost to Google must be minimal. Still, if one library is digitizing 3,000 books a day, even with the latest and fastest scanning equipment (the slowest part of the process), that has to be multiple workstations and quite a number of students working on that project!

The article suggests they are simply making images of each page ("fingers are visible in the corners of many pages on books.google.com"), and that's a shame. If they're not being text-reco'd, they're missing a great opportunity.

In re-reading the article, I realize again they should have given that story to someone who actually knows something about e-books. In comparing e-books to CDs, the author says "The simplest difference is that transferring one's old music CDs onto iPods is easy, whereas transferring one's old books onto an e-book is impossible."

Really?

[igorsk]

Storing images of pages does not prevent them from being OCRed later, while storing just text loses quite a bit from the complete presentation, not speaking about possible OCR errors which can be hard to correct without checking the originals.
Anyway, if you do a search on books.google.com, you will see the search result highlighted on the image of the page. So I guess they store both the image and the text of it.

[NatCh]

Originally Posted by Steve Jordan

If they're not being text-reco'd, they're missing a great opportunity.

Doesn't Google's project allow searching the text of the books they've scanned? I was under the impression that it does, which would suggest ....

[The GreatGonzo]

...and how many of those books are actually scanned WELL?

I seem to run across quite a lot that have pages missing or so skewed that text is missing; some contain colored pages instead of black and white text; some files won't open or display correctly ... I mean, this is almost fully automated scanning we're talking about, right? Somebody jamming a bunch of pages into a sheet-fed scanner and uploading the result within minutes?

I've all but given up on Google books...

[CommanderROR]

I hope they really get serious about this project. Scanning all the books, correcting errors and then OCR and proofreading are a lot of work and only a really big company like google could even hope to manage that...

[Steve Jordan]

Actually, I never thought Google was right for this job. Given the apparent inconsistencies of the results, there should be a dedicated organization doing this... a group that will put more effort into properly scanned, reco'd and saved texts.

Don't ask me what organization. It should be the Library of Congress, but I don't think we can expect them to do it.