At the aperture used (f/2.5), the Nikon 50/1.8 does struggle outside the middle 60-70% of the frame on full frame digital in my experience, so I'm not so sure Steve had a bad copy. At f/4 and beyond, it is a very, very sharp lens.
My main technical criticism of the comparison is that taking two cameras with AA filters (E-P2 and D3s) and giving their files the same amount of capture sharpening as the file from a camera with no AA filter (Leica M9) is going to produce misleading 100% crop comparisons. I've no doubt that the M9 would produce more detail regardless, but this is not an even playing field.
It seems pretty obvious from all the disparities that he gamed his test methodology to produce the M9-favoring result he wanted, and then ignored any contrary data (e.g. the severe color fringing and false-resolution edge artifacts in the M9 crop.)
But from the comments, that seems to be what his fan base wants to hear, so everybody's happy.