Comments on Six Degrees and the Science of Networks
Professor Watts' survey of work on the six degrees conjecture (the tantalizing yet difficult-to-prove assertion that every person is socially separated from every other person by a chain of no more than six people in the intricate web of human relationships) is clear and captivating, as is his explanation of so-called small-world networks (consisting of nodes that are locally clustered yet globally connected to all other nodes) which he spent years researching. I am intrigued by the modeling of network-related phenomena in the universal framework of phase transitions describing behavior between the extremes of highly connected order and complete randomness. I also agree with the author that the analysis of networks offers (at least some) promise of providing the scientific and business communities with a better understanding of how and why social networks (including the financial markets) both work smoothly most of the time and fail miserably some of the time.
Optimism aside, however, I see the primary weakness of the "science of networks" as being its attempt to analyze social behavior with the precision of science without offering (at least to date, in my opinion) verifiable and useful predictions of significant real world phenomena. Without the rigor of hypothesis testing and an active interplay between theory and experiment, model-building ends up looking more like creative phenomenology than science. The author tries to rationalize this apparent shortcoming by stating that Darwin's theory of evolution also lacks predictability, but here I beg to disagree--the theory of evolution predicts (notably reaching beyond creationist theories) that beneficial mutations will, through natural selection, survive to chart the course of species evolution, and life scientists continue to uncover evidence that the mechanism of evolution remains alive and well today.
Instead of being an erudite treatise on a new science, then, Six Degrees is really more of a report on the trials and tribulations of academics involved in the interdisciplinary challenge of understanding the workings of social networks. The author's active participation in this endeavor and candor in relating his experiences make the book interesting reading, though I sense that the catch-all nature of the underlying subject--which dabbles into biology, epidemiology, computer science, sociology, psychology, economics, business and the financial markets--could leave its proponents holding an empty bag when the dust settles in a decade or two. However much I may wish otherwise, I fear that the ultimate fate of the science of networks will be akin to the academic community's silent dismissal of artificial intelligence as a worthy pursuit--like the now quaint AI of yesteryear, the new science of networks appears to be too long on hype and hope and too short on delivery of truly meaningful results.