Good points, all. But I think we may be underestimating the book to byte 
transition.

Currently, the chief problem with search engines is that they search 
relatively indiscriminately once given their commands. Neither the 
search engines nor many of us, apparently, are able to differentiate 
between the useful and the useless "hits" that come our way. What is 
needed now is some way or ways to sort through the 100 or 1000 hits and 
identify the 10 or 20 really useful ones--a "reliability" index of some 
sort, maybe several of them (e.g., good enough for this, good enough for 
that) would help.

Of course, once that's done, we're just a few software generations away 
from the next obvious step, which is for the computer to provide the 
user with a precis, report, or summary--according to user-supplied 
specs--of the material that is found on the Net about topic a or b. This 
would probably be a commercial service at first, maybe forever, but it 
also might be the end of "research as we know it."

Not to worry, though. It'll still be a while before computer-generated 
reports are sophisticated enough to discourse on the many meanings of 
"cool." The  human brain is much more complex than computers (I think 
that's still a true statement!) and definitely quirkier, so we shouldn't 
expect to be supplanted any time soon.... But in ten or twenty years? 
Who can say?

Doug Deal
History/SUNY Oswego