Linked
by Albert-László Barabási
The middle consonant in László is pronounced S, not Z. Just had to get that off my chest.
This seems to me an important book. Barabási is a leading researcher in the science of networks and a clear and forceful writer. He concerns himself here with the formation of systems that fit the "small world" model and behave according to "power laws." It might appear that his subject overlaps what Mark Buchanan wrote about in Ubiquity and Nexus, and the appearance is right, but Buchanan is a skilled science writer while Barabási is a creative scientist.
When I read Buchanan's books I found it frustrating not to learn why or even how small-world networks come about. The notion that "any two people in the world are connected through just a few links," with the added claim that the number of links is six, has to do with these networks, in which for example you and your friends form a cluster, a rock 'n' roll knight and his friends form another, but the world also includes someone with a connection to both clusters. "I guess it's a small world after all," you say on discovering that one of your buds knows a traveler who has a cousin whose address book lists George Harrison's barber's next-door neighbor.
Buchanan's fine works (I wrote some notes about them) describe and diagram these networks but don't really say what process gives rise to them. Since they occur so frequently, it could be important to understand what creates them. That's one of the strengths of Linked: Barabási can show you the network and also demonstrate how it gets built.
You set initial conditions (consider a few points all connected by lines) and make a plausible assumption about what happens when the next point is added (the probability it links to an existing point depends on how many links the old point already has, and maybe also on some other factor like distance), and then you turn the system on. Let it run for a few cycles and make a chart showing how many points have 1 link, 2 links and so forth. You get some version of a small-world network. The result doesn't depend on whether the points represent telephones or movie actors or predator/prey species, although details will vary.
So the answer to my question lay in the fact that networks aren't hooked up at random but grow by laws. You can look at an existing network, like that of film performers, and reconstruct how it came about; you can relate the process to the statistics in the network (the average Bacon number is 2.79); but you can't understand the results without doing the experiment of casting a new movie or releasing zebra mussels into American waters. This, I believe, is what was missing. I should say this is what I was missing.
There's more than movies and molluscs in Linked. The author himself is highly linked in a growing community of network researchers, and in fifteen rather exciting chapters plus an epilog written for the 2003 reissue he presents his own findings and those of many, many other scientists. Networks provide a revealing way to look at physical, ecological and social systems, and the work described in the book has already led to predictions, some of them pretty sobering: The Internet has no good defense against massive denial-of-service attacks, and al-Qaida has no "head" you can sever to destroy the enterprise. While Mark Buchanan described the nature of these networks and showed that they appear again and again in society, Albert-László Barabási tells you where their study is likely to go in the next few years. You will need a pencil and a doodling pad to follow what he says, but no math beyond about fifth grade. This is a terrific book and one you'd regret not reading.
| Approved Ben Teague web site |
 |
Books and plays:
Science and technology:
Linked