Your Scribd example is a bad one because that's an example of O'Reilly putting a sample of that book on Scribd to tempt people to buy the real thing.
Basically, I think she's going to have a really hard time proving any harm. She says her book was downloaded 100 times. Now prove that that equals 100 lost sales, or X lost sales for any value of X.
My assumption is that if my book was downloaded from Scribd 100 times that I've lost way less than 100 sales (if any) which at royalty rates on a book means I might have lost a couple of dollars. For that reason I wouldn't sue Scribd.
The fact that the content is in "digital" even makes checking for notices of copyrights that much more trivial. We're all geeks here, so lets be honest:
http://www.scribd.com/doc/15490856/Java-Pocket-Guide-by-ORei...
"Copyright (c) 2008 Robert Liguori and Patricia Liguori. All rights reserved."
If you are taking a digital format and converting it to the scribd format, how trivial is it for you to write a function that greps for the "Copyright" and "all rights reserved" phrases in the text?
The Olive Oil business: (A protection racket. Organized crime. etc.)
I hope she gets her pound of flesh.