Networks are everywhere: from social networks and terrorist networks linking people through the World Wide Web and beyond to biological networks communicating within a cell and from linguistic networks describing how words relate to each other to networks tracking how diseases spread globally. What’s remarkable, says physicist Albert-Laszlo Barabasi—who studies such networks using the methods of statistical physics—is that these seemingly very different systems have many similar properties. This observation has changed the way he views many things. “Ten years ago it would have been easy to say that the cell and the World Wide Web have nothing to do with each other,” says Barabasi. “Yet using some of these network ideas, we were able to show that the architecture of these systems is very similar.”
Growing up in Transylvania, Barabasi wanted to be a sculptor. But strangely, he found he kept winning physics contests, and eventually realized he had more talent for physics than for sculpting—though he still greatly appreciates the arts.
Barabasi first became interested in networks when he worked at IBM as a postdoc, just after earning his PhD in 1994 from Boston University. He had some freedom to study what he wanted, so he decided to learn about networks.
He had also made an observation: “At that time I lived in New York City, and walking through the city I realized there are so many cables under the pavement: Internet cables, phone cables, electricity, water, gas …all those things.” As he thought about these utility networks, it struck him that there must be enormous complexity to these networks. But all the literature at the time treated networks as totally random entities. “There’s no way this could be completely random,” he thought. There had to be more to it, and this observation led him to become more interested in studying networks in more detail.
He wrote his first paper on networks in about 1995. It was rejected by several journals, including two physics journals, he says. “Nobody said it was wrong; everybody said, ‘why do we care about this?’” Which was a legitimate concern at the time, he admits.
In 1995, Barabasi joined the physics department at Notre Dame University. Although he was interested in studying networks, he figured those kinds of networks wouldn’t get him tenure, so he focused his research on materials science at first.
After several years, Barabasi turned his attention back to networks. In 1998, he and colleagues discovered that a large number of networks are what is called “scale free networks.” Classifying nodes in a network based on how many links that the node has, Barabasi found that while the vast majority of nodes have only a few links a few nodes, the hubs, have a very large number of links. The distribution of links follows a power law, Barabasi discovered.
The discovery that many diverse systems exhibit this property helped spark an explosion of interest in network science in the late 1990s, says Barabasi. Physicists, especially, have been attracted to the study of networks. While some networks are small enough to map out, others are so large that one can only trace a small fraction of the nodes and connections, says Barabasi. This is where the tools of statistical mechanics can be useful, he says. Statistical mechanics tries to predict properties of collections of particles, such as molecules in a gas. One can’t possibly track the behavior of the individual molecules, but using statistical mechanics, one can still calculate properties, such as the temperature and pressure, of the gas as a whole. “Networks are not so different in many ways,” says Barabasi. “It turns out that you can say things about the behavior of the whole network even if you just know about a few nodes. This is very similar to what happens in statistical mechanics.”
Some of Barabasi’s recent work involves biological networks. With the human genome sequenced, scientists are realizing that we need to understand how genes interact with each other in networks. Barabasi offers an analogy: “Just having the list of parts of the car will not actually teach you how the car works. If you just have a big bag with the parts of your car, you’re just lost. You need to know which piece is going with which and how they interact together.” One of Barabasi’s recent projects on biological networks involved studying how networks can help us understand diseases related to genes; another project focused on how metabolic networks and their properties can help us learn how to design drugs rather than find them through a trial and error process.
He is also studying the dynamics of networks- not just the nodes and links themselves, but the events that happen on the nodes, and the timing of those events. Barabasi has noticed a variety of interesting phenomena through his research. For instance, he and colleagues recently found that the number of people accessing news stories posted on the internet follows a power law, peaking within a few hours after the story is posted, and decaying significantly after about 36 hours. Barabasi has also written a popular book on networks, called Linked: The New Science of Networks, which came out in 2002.
His work on networks has even influenced the way Barabasi views the world and interacts with others. For example, in his own social and professional networks, he says he pays particular attention to making good use of hubs, those well-connected people who can help him get what he wants. But he has also learned to use the hubs wisely, because those people are often overworked. That’s just one example of how his work has given him a different way of looking at things, he says. “I think that once you adopt the network view, you will have a bit different perspective about the world.”