Evan gave a talk on this at Le Web 3.0. It was short and sweet. I liked it, it makes the point well. Found the video online: http://light.vpod.tv/?s=0.0.392057
I agree that this is the potentially "hottest" statement from the story. Yet only two examples are given: Google and Twitter. Taking away the clutter around the search box was a good idea, but wasn't Google's real innovation something like counting links to pages to judge their significance? The Twitter example is better, but one example is not enough to make the case. Are there others?
Before Google, most search engines only examined the keywords on a given page to determine relevence.
Google improved on this by ranking results according to the "PageRank", which is a probability distribution that represents the likelihood that a person randomly clicking on links will arrive at any particular page.
With few exceptions, PageRank doesn't really capture any semantics associated with the links.