I'm glad that Google is in the world, and I imagine that most people would agree. Sure they're practically taking over the Internet, but so far they seem to mostly benevolent about it. Sure they have a few lapses in judgement like China and (more recently) Dell, but mostly they're good guys. I think of them as the BDFL's of the Internet... if they hired Guido they can't be all bad. :-) They set a very high bar for software quality, just like (if I may say so) my employer, Apple. They have an office here in Santa Monica just a few blocks from mine, so I feel a certain connection with them.
That's why I feel obliged to make a suggestion: I've noticed that on Google Maps, the little green arrow is often not quite accurate. It's usually on the right block, but it's often not on the right building... often three or four doors away from the right location. I think it'd make a good project for the "one day a week" that Google engineers are supposed to spend thinking about new, cool stuff. I mean, it's no MentalPlex, but I bet there's some way to analyze a set of data comparing where Google Maps thinks an address is to where it actually is (as manually indicated by human data entry), correlate the errors to some other variable(s), and construct an algorithm that could be used to correct whatever approximations are inherent in the geolocation data sets that they're using. I mean, how cool would that be, if you could type an address into Google Earth and it showed you where the front door was, with an error margin of 3 feet instead of 300 feet?
So that's my idea. I don't have any illusions that Google researchers haven't already thought about this problem, but maybe with a little gentle prodding they'll look at it from a fresh angle. Consider this post a friendly "open letter" to my esteemed neighbors. :-)