The good news, I think, is that Lifehacker references this blind test by Tyll Herstens. I think that blind testing of audio equipment is a good thing. Unfortunately Tyll Herstens does not perform a double blind test or "ABX" test.
Also I wonder about the results. He chose correctly 13 times out of 19. I wonder if that is statistically significant. I wonder also why he didn't do the test 20 times, it would have been easier arithmetic
As others have pointed out I think also that the big problem with testing for burn in is that it is difficult to rule out manufacturing differences between one pair of headphones and another.
However I will say that the whole thing is a huge improvement over the usual anecdotal evidence reported as fact.
Personally I think that break in in headphones is quite likely but not very significant. But I'd love it if this would be clarified by a good scientific approach to listening tests.
I do agree with the comments from the Grado. I suggest that in general, people should stop fetishising their equipment and listen to the music instead.
I would love to see Hi Fi publications (paper & web) adopt thorough double blind testing of equipment and reduce the amount of "reviews" which are just impressions possibly very coloured by suggestion and auto-suggestion.