Serraphin: But you have to allow someone to cache your site to let the www in its current incarnation work :)

True, but this is an argument about fair use: The web--and especially search--would be useless if people couldn't make copies and keep them on a server for a few months. (That's what Google does; they make multiple byte-for-byte copies and store them until the next scan.)

But what's the connection between copying being good for the web and copying being legal?. Not much, unless you believe in fair use. Fair use allows small infringements if they benefit society enough.

So why should we, as a society, allow Google Print to fudge copyright rules enough to scan books, index them, and show 30 words at a time? Because it would cost the average author little, it would increase sales of the backlist, and--most importantly--it would make most of human written knowledge available within a two-second search.

Huge benefit, little cost.

Again, this is an argument about defaults. I think that any author who objects should be able to (1) send a form letter to Google (and their competitors) telling them to leave certain ISBNs alone, or (2) post such notices in a common database. But if we actually require Google to perform copyright searches and secure individual permission for each book, we'll loose about 80 years of backlist, and a lot of minor works in the frontlist.