What is the problem that a driverless car is going to fix?
To paraphrase Henry Ford, it sounds to me like google is actually trying to build a faster horse,
Uh - maybe auto accidents and deaths for a starter [1] ? Computer driven cars are much more ikely to be safer than manually driven ones in aggregate.
To flip the tables, lets use a computing analogy for cars: Imagine if each TCP-IP packet (or connection) were hand-driven or managed. Lots of collisions and traffic jams. Some packets/connections would have unbelievable latency/throughput. Others (most) would be stuck in traffic that was inherently preventable assuming some rules were in place that would need special permissions to override.
Now compare with our Internet (as sucky as it is, buffer-bloat and all) - it's a goddamned paradise in comparison to the above.
Now imagine the flip side analogy - cars "routed" by algorithms, protocols and, where applicable, user intervention. That's Google's vision - it's not a new one, just one where they're building it out. Actions >> Words.
I would love to commute to work not actually doing any of the driving (secretly I'd prefer public transport, but only if it were nearly as convenient as point-to-point driving that I can do now). A driverless car is a great idea - sure my commute might take a few min longer as "the system" routes me, but the likelihood of traffic incidents and the like would probably be lower, preventing those 2-3x longer commute days.
Sign me the fuck up.
[1] http://www.csmonitor.com/Busin...