Wow! I had the same idea over 10 years ago. I'm glad it's being done.<p>My idea was using a giant P2P indexbase. However, I read the tech page, but it doesn't explain enough. For example, for this to work, there would have to be some kind of "information routing" whereby queries within certain information domains would be routed to machines. I can't see how this is being done.<p>Because you can't have the whole index on each machine, therefore, how does it know where to route a given query?<p>Additionally, peer nodes would have to crawl to a certain criteria; either subject based, or perhaps geographically. Otherwise you'd just end up with the same stuff everywhere.<p>If there's some more explanation, please link.<p>thanks.