"In his view, it looks like the source ports are sufficiently random,
even though they are limited to a small range of ports."
The distribution graph appears to show Google resolver using random ports
between 32768 and 65535. While that's only half the ports available,
it's misleading to characterize it as "a small range of ports".
This topic has been visited numerous times. A particularly good article on theoretical computational limits appeared in Nature in 2000:
Lloyd, Seth. "Ultimate Physical Limits to Computation". Nature 406, pp. 1047-1053 (31 August 2000)
http://www.nature.com/nature/journal/v406/n6799/full/4061047a0.html
This is the single most coherent remark I've seen about this story. Unfortunately critical reasoning in the mainstream media doesn't pay the bills, so we'll never see it reported there.
"How to make a million dollars: First, get a million dollars." -- Steve Martin