I don't quite see how it is useful to apply the requirements of an internet search engine to all types of data processing. Adam Bosworth's only concern seems to be how huge amounts of poorly structured data can be narrowed down to a few hundered links. That's what we get from google today (thanks by the way) and what we got from Altavista many years ago in roughly the same way.
The technology is actually so primitive that I can't even find a book written _by_ a particular person without finding all the rubbish people have written _about_ this person. I would be very interested to hear how RSS is going to solve this problem. And I would be very interested to hear how people are supposed to analyse their sales figures using text queries on top of Atom.