Amazon released its digital archive yesterday. WIRED has the story.
Briefly, Amazon had a few hundred thousand books scanned and OCRed and put them online in a database. You enter keywords in the search box as before, but now all searches include full-text search of all participating books. If a book had a full-text hit, you get to see a few lines of text around the matching keywords. You can then view the actual page (as a low-resolution image, but enough for it to be readable) and browse forward and backward through the book!!!
The resolution of the images is deliberately kept low enough that you would normally want to purchase the book if you were interested in actually reading it. Also, linking to individual books or pages appears to be intentionally crippled. (This is how Jeff Bezos sold the publishers on the idea in the first place.) But as a tool for finding books that relate to a given subject of interest, this is an incredibly useful tool. This holds particularly for expensive scholarly books --- even those of us who have US$50--$150 to spare (or to spend from their book budget) on a collection of recent research papers on some abstruse subject in the natural sciences would not like to waste their (organization's) money. Until now, people would rely on book reviews in scholarly journals, or (for older work) citations (via tools like CiteSeer or Web of Science), or on occasional opportunities at browsing (book fairs, exhibitions at very large scientific meetings, visits to national libraries with a "buy everything" policy and budget). Now one can actually do this from one's office or home, and much more efficiently than by browsing hardcopy.
The image views, incidentally, are particularly useful for books in the natural sciences, as mathematical formulas, chemical structures, graphs, diagrams of experimental setups,... do not lend themselves well (if at all) to OCR-ing.
Caveat lector ("reader beware") continues to apply, of course. Considering that Amazon will market virtually anything with an ISBN number, it cannot be ruled out that some searches will lead to material that is crackpot, repugnant, or both. But the same will apply to going to a large bookstore or the Library of Congress. Search engines are an exceedingly useful tool, but never a substitute for using one's necktop computer :-)