As Internet users know too well, the global information network is constantly plagued by hundreds of small failures that can bedevil wired companies and royally bum out more casual ’Net surfers.p. Yet such local problems tend to stay smalla result, some researchers now believe, of redundancy in the network that resembles what biologists have found in living organisms.
A new study released last week suggests that the Internet has become a peculiar creature, its vast infrastructure far less random or haphazard than some experts assumed.
Such research also carries a perplexing lessonthat after decades of watching the global network grow and evolve, the Internet’s designers now can scarcely comprehend their own creation. Some experts wonder if over time the line between computer networks and living thingsor even intelligent beingswill begin to blur.
“You wouldn’t think we’d have to study the Internet, because we created it,” said Albert-Laszlo Barabasi, a professor of physics at the University of Notre Dame and leader of the new study. “Internet researchers are like doctors trying to figure out what their patient looks like.”
Looking at more than 200,000 central physical connections within the Internet, Barabasi’s team found that the links form complex geometric patterns, with some crucial hubs connected to hundreds of other sites.
That’s different from the random pattern of connections that technicians often assume in designing new networks.
The report, published Friday in the Proceedings of the National Academies of Science, is one of the first detailed analyses of the massive network.
Some biologists hope that such studies may help them find clues to deeper mysteries of nature, including the interactions of living cells and genes. One prominent team led by Dr. Bert Vogelstein at Johns Hopkins University has even found similarities between Internet failures and the cascades of genetic foul-ups that cause some types of cancer.
“The next wave of biological research is going to involve systemswhen we get beyond individual components and see how it all fits together,” said Vogelstein, a cancer researcher at Johns Hopkins. “Systems engineers and people who look at systems of communication will probably be quite relevant to understanding how cells work.”
Barabasi’s team found that the Internet, like networks of proteins in cells, follows a “scale-free” geometric pattern, meaning that the Internet’s overall structure looks the same at the level of a whole country or a smaller region.
The network’s ad hoc arrangement has a deep effect on how the Internet can adjust to problems.
Most Internet nodes, also called routers, have only a few links to other sites, so losing them doesn’t affect the network as a whole.
Yet the reliance on a few central hubs also could leave the Internet vulnerable to terrorist attacks—what Barabasi calls the Achilles’ heel of the Internet.
The largest Internet hub in the world by some measures is a sprawling access point in downtown Chicago run by SBC Ameritech, with 125 major connections to other sites. Manager Tony Haeuser said for security reasons, the company does not disclose the facility’s precise location.
However the structure of the Internet arose, Barabasi and other researchers are finding similar patterns in far-flung sources. Ian Foster, a professor of computer science at the University of Chicago, said his group has drawn on Barabasi’s work in studying how scientists collaborate.
Working with pathologist Zoltan Oltvai of Northwestern University, Barabasi’s group last year published a detailed map of most protein interactions in a species of yeast. They found the same scale-free structure they saw in the Internet.
The more links a protein had, the more likely that its removal would kill the yeast. Yet most proteins have few links, allowing yeast and humans to withstand many mutations and other errors in protein production.
The similarity to living things goes only so far; Barabasi said the Internet is not truly alive because it cannot grow or function on its own.
October 6, 2002