Open warfare erupts in the world of mathematical biology, as Lior Pachter of UC-Berkeley writes three blog posts attacking two papers in Nature Bioscience, accusing one of them of being "dishonest and fraudulent": The Network Nonsense of Albert-Laszlo Barabasi
, The Network Nonsense of Manolo Kellis
, and Why I Read the Network Nonsense Papers
. Kellis (MIT) and his co-authors respond
"the scale-free network modeing paradigm is largely inconsistent with the engineered nature of the Internet..."
For a decade it's been conventional wisdom that the Internet has a scale-free topology
, in which the number of links emanating from a site obeys a power law
. In other words, the Internet has a long tail
; compared with a completely random network, its structure is dominated by a few very highly connected nodes, while the rest of the web consists of a gigantic list of sites attached to hardly anything. Among its other effects, this makes the web highly vulnerable to epidemics.
The power law on the internet has inspired a vast array of research
by computer scientists, mathematicians, and engineers.
According to an article in this month's Notices of the American Math Society
, it's all wrong.
How could so many scientists make this kind of mistake? Statistician Cosma Shalizi
explains how people see power laws when they aren't there: "Abusing linear regression makes the baby Gauss cry."