Re:Supercomputers for earth simulation are useless (Score 1)

Well it is sort of true that supercomputers are useless. This 4 week time span mentioned above is actually from 1960, when it was possible to forecast weather 2 days in advance but a month was thought possible. In 1974 this expectation was dropped to 10 days, and today, it is agreed (see Tennekes among others) that 7 days is the limit for accurate forecasting even as computing power increases without limit. So really, this big toy should take us a little closer to the 7 day forecast actually being right, but not much further.