It's not apocryphal. You can actually download the analysis they did and see the calculations for yourself. (I did; it was wasted on me.)

http://lesswrong.com/lw/rg/la602_vs_rhic_review/ (The link in Yudkowsky's essay seems to currently be down, but if you google the filename you find mirrors like www.fas.org/sgp/othergov/doe/lanl/docs1/00329010.pdf )