So, everything's going to get digitized. I've always expected that, but never understood exactly how it would be paid for. I guess advertising is the answer, but that wasn't clear even a few years ago. Of course, more to the point, everything is going to be available (and fully indexed) on the web for free, assuming you don't mind a little advertising. Or at least everything without copyright encumbrances, i.e. everything up to 1923 in the U.S. And we have to imagine that, sooner or later, ways of getting much of the rest of library materials online will be worked out. Suzanne Pilsk asked me to meet with a group from the Smithsonian Institution last week that are involved in The Biodiversity Heritage Library Project (now part of the Open Content Alliance) which plans to get rights to digitize and display journal articles that are currently protected.
All this activity is showing that 'protection' has become close to 'hidden' for material that has limited income potential. I'm not sure what Google gets from advertising on a typical page view, but I'd guess it's probably less than a cent (although with a reported $6 billion in revenue maybe they get more than that). Library material that is not easily accessible online just takes too much effort to find and obtain for all but the most dedicated scholars, so it is effectively hidden.
Back to the original question: what will be the impact of having this material publicly viewable? It's easy to predict that it will be used much more than it has been for a long time, possibly more than it ever was. More people will look at a particular page online than will ever look at that physical page in all the copies in all the libraries in the world. That's clear, but what will the effect of those online viewings be on the need for physical access? Here's my prediction: seeing the page images online will result in more requests for the physical object, not less. I'll be very surprised if that's not true for the libraries that own the scanned item.
I'll even go further: I think we may see more use of equivalent items (e.g. at the FRBR manifestation or expression level) at other libraries, especially if we can get our act together and make those requests cheap and easy. I doubt if gate-counts go up because of Google-print and other activities, but we shouldn't be surprised to see access to otherwise forgotten items go up substantially.
One of the things the folks from the Smithsonian that I talked to wanted was a web service that would resolve journal abbreviations. Actually, what I think they really want is a service that can take in a cryptic citation and output a nicely formatted one with the full journal title, etc. There are lots of web sites that list various journal abbreviations, and even All That JAS which is a bibliography of those sites, but in my limited looking around I didn't find anything that quite does what they want, even for just the journal abbreviations.