Please, no more! It's still measuring different algorithms (memoizing, iterative) and measuring them in different ways (including/excluding startup costs).
This says nothing, and does no one any good.
I'm usually okay with these comparisons as long as the author lists his intentions out, as well as his environment/hardware and how he runs each test.
I agree that the past two blog posts have been kind of pointless, but fun anyways :)