I wonder if they are re-hashing the older years' data with new algorithms? I thought I read somewhere that the granularity of their computations had improved by (don't quote me on this) something like a factor of 3 or something significant like that. So maybe they think we missed something back when we ran them in 2007/8? Computing power is certainly way higher now, so maybe that allows them a "finer tooth comb"? I'm not sure either, but just some speculation for you...
