Yeah, they'll get that, not exacly unexpected though, and I wonder what the marginal cost of running huge speculative mapreduce jobs on "spare" server resources add up too?

I honestly think there's some world-changing insight hiding in data lurking idly across Googles datacenters - which just needs someone curious enough to turn the right datasets round the right way and ask the right question...

What's that old science gag? Something along the lines of "Amazing discoveries are more likely to be preceeded by 'Hmmm, that's strange' rather than 'Eureka!'"