The Art of Google Books
February 11, 2012 9:13 AM   Subscribe

Here's a stunning gallery of failures and book oddities in Google Books. You can see camera-sensor moire, content detection algorithm failures, condom-covered fingers, failed dewarping, "dewarped" images, failed dithering, pages blurred by turning, and the scanning environment itself, with binder clips to hold book covers and checked rules for absolute scale. My personal favorite: A shot showing the Google page-turning mechanism (employed when they're not using a guillotine and ADF/sheet-feeder). A project by Kristina Wilson.
posted by fake (19 comments total) 26 users marked this as a favorite
Pay no attention to the man behind the curtain!
posted by Meatbomb at 9:17 AM on February 11, 2012

I realized the moment I fell into the fissure, that the book would not be destroyed as I had planned. It continued falling into that starry expanse, of which I had only a fleeting glimpse. I have tried to speculate where it might have landed, and I must admit however, such conjecture is futile. Still, questions about whose hands might one day hold my Myst book are unsettling to me. I know my apprehensions might never be allayed... and so I close, realizing that perhaps the ending has not yet been written.
posted by radwolf76 at 9:23 AM on February 11, 2012 [12 favorites]

Failed Dewarp is the name of my new band.
posted by twoleftfeet at 9:27 AM on February 11, 2012 [1 favorite]

Metafilter: a stunning gallery of failures and book oddities.
posted by michaelh at 9:30 AM on February 11, 2012 [2 favorites]

condom-covered fingers

More properly called "finger cots."
posted by jedicus at 9:32 AM on February 11, 2012 [6 favorites]

Wonderful stuff. I particularly like that the book shown in one of the samples above is Rules for Regulating the Subscription Library at Stamford (1787).
posted by WPW at 11:12 AM on February 11, 2012

More properly called "finger cots."

Because they make your fingers fall asleep uncomfortably?
posted by Sys Rq at 11:22 AM on February 11, 2012 [1 favorite]

I was doing a lot of google book reading a few years ago and I joked that I was going to screen cap all of the accidental fingers I found. Damn. Coulda made the front page!
posted by Biblio at 11:28 AM on February 11, 2012

Ron Swanson's grandfather had a bad case of The Spots.
posted by gjc at 12:28 PM on February 11, 2012

The failed dewarping is the one that confuses me. How can the algorithm get *that* confused by text?
posted by gjc at 12:30 PM on February 11, 2012

It's a hard problem, actually. My community has developed a series of dewarping algos, and depending on what features you choose to generate your line (which you then straighten) it is pretty easy to jump from the top of, say, a "t" to the next line's descending "j" and then back again.

Dewarping is far from trivial. We have threads on it and there is also a yearly dewarping contest.
posted by fake at 12:44 PM on February 11, 2012 [7 favorites]

That there's a page dewarping contest is completely awesome.
posted by GuyZero at 12:52 PM on February 11, 2012

"Dewarping contest" sounds like something they'd have at teen evangelical retreats.

Meantime, do any of these links tell you who to contact when you find a Google Books fail?
posted by yoink at 4:52 PM on February 11, 2012

Now somebody start a similar study for Internet Archive, and may the best team win.
posted by Twang at 5:19 PM on February 11, 2012

Internet Archive does no dewarping at all, so many of these problems don't exist in their collection. They reflect a different approach which is more about copying the visual totality of the book than a searchable computational derivative of the book.

Also their scanners do not wear finger cotdoms, rather gloves, so I suspect you won't catch any pink surprises amongst their stuff.
posted by fake at 6:21 PM on February 11, 2012

Maybe it's just some geolocation oddity, but why the hell is a 136 year old book still marked as being under copyright?
posted by ymgve at 7:09 PM on February 11, 2012

I think that while the text itself has certainly entered the public domain, the image of the specific book can still be considered owned by the library in question. At least that's what I recall learning over a decade ago while working on a pre-Victorian women writers project. We'd mark everything up in SGML to describe various features (including layout) and would include brief descriptions of engravings, but images of and from the books couldn't be included in what was made available to our subscribing institutions.
posted by nobody at 7:31 PM on February 11, 2012

I hope that this blog gets made into a book which is then scanned by google.
posted by moonmilk at 8:34 PM on February 11, 2012 [2 favorites]

fake: "you won't catch any pink surprises amongst their stuff."


Thanks for that chilling mental image.
posted by subbes at 7:26 AM on February 12, 2012 [1 favorite]

« Older A tourist's guide to Belbury   |   Thanks for Playing. As Usual You've Been a Real... Newer »

This thread has been archived and is closed to new comments