My question was actually simpler...
What I was asking...
were you using defect density (and all the underlying statistics and math that go with that)
OR good die/wafer
OR percent yield?
Based on your answer I assume you really did mean the first... and I think you are all wet. As has been pointed out, core 2 has a higher percentage of cache, which thanks to redundancy yields better than random logic. The main thing Bose-Einstein adds to the mix is process complexity, and netburst and core 2 have the same process...
It may be that on the first mask set for core2 they messed something up and yields were lower than they should be, but that is something that would be quickly corrected.
--Alan