As the financial markets battle the fallout of years of poorly regulated unwise greed, the language of analysis is revealing. Commentators talk of “contagion spreading“, financial “gears jammed“, and “turbulent” markets. This is the language of non-obvious connection, where it’s theoretically possible but impossible in practice to predict the future state.
Listening to This American Life’s new episode on the spreading financial market failures brought this home. It talked about Credit Default Swaps (CDS) and how hedge funds would both buy and sell these so that they were making money from the interest even as their clients were covering each other’s principals. This worked fine for a while, but because hedge funds were each others’ clients there arose these long lines of dependencies—if A failed to pay B, then B couldn’t pay C, who couldn’t pay D, and so on. When the markets started to choke on rotting mortgages, the CDS chains began to unravel.
The CDS game was a bugger because it was easy to check one fund’s books and say “yes, we are covered”. But there was no way to identify these chains of dependency: the connections between funds were invisible. And, as it turned out, the connections between funds were so much more impotant. Just as one person dying isn’t as important as the fact that they can infect others, so too one hedge fund going belly up wouldn’t have been anything like the disaster of the dependencies. With globalisation, these connections can span markets, borders, currencies, and languages.
This newly-realized importance of the network reminds me of biology, where we thought we could understand an organism by mapping its genes. Now we realize an organism is a complex mixture of manufactured and transformed chemicals and even other organisms, and the genetic blueprint is necessary but not sufficient for understanding. You can no more understand how an organism works by reading its DNA than you can understand how San Francisco works by reading its phonebook. This “whole organism” multi-level integrative approach is called systems biology.
Nodes often aren’t as important as the connections between them. Reductionist science and analysis from the 19th and 20th centuries focused on nodes. I believe 21st century science, economics, political science, and computer science will use more complex systems theory to understand the interactions between chemicals, speculators, nations, and users.
Social network mining, exemplified by Duncan Watts‘ six degrees work, is just the tip of the iceberg. I think this kind of modern network analysis can make the world a safer place, our futures more secure, and even our bodies more comprehensible. In 100 years time, historians will say that this century’s Einstein, Watson, Crick, and Feynman were students of network analysis, and that this was the Connected Century.