It takes only four characters to state Newton’s first law: F=ma. But are such simply expressed theories possible – or even desirable – in an age of large data sets and complicated biological and social systems?
SFI Faculty Chair David Krakauer and Google Engineering Director Graham Spencer, an SFI Trustee, debated the question to begin a day-long joint meeting at Google headquarters in Mountain View, Calif., in late September.
David argued that simpler theory is important because it facilitates communication between scientists and “interoperability” between fields, and large data sets and powerful computers might stand in the way of that. He cited an example from early physics: Tycho Brahe explained planetary orbits using a system of interconnected circles, but Johannes Kepler used the same data to develop a simpler explanation: ellipses. The problem, David said, is that the descriptions are mathematically equivalent, and computational methods might not find the simpler answer.
Graham countered that simpler theory is OK, provided it offers a means of quantifying complex phenomena. He adopted cybernetics researcher Norbert Weiner's convention of measuring the complexity of a system based on the algorithmic complexity of some observable
property of the system. This, of course, captures only a small part of the system under observation. It might still be possible to nd simple models, he said, but that might mean stumbling on something that happens to be simple.“Why are you looking for your keys under the lamppost?” he asked. “Because that’s where the light is.”
SFI External Professors David Ackley and Walter Fontana continued with a discussion of novel forms of computation, including Ackley’s “infinite” computers and Fontana’s concept of a programming language to describe biochemical reactions. SFI External Professor Jim Crutch eld and Google’s Joseph Smarr spoke about network fragility and the changing nature of social and economic networks.
Informal side discussions were an equally important feature of the meeting, says SFI VP for Administration Chris Wood, who manages the Business Network. “There will be lots of individual follow-up, and that’s where much of the value of the meeting will play out,” he says, pointing to SFI postdoctoral fellow Bela Nagy as an example. Bela spent several days talking with members of Google’s Public Data Explorer Team and spoke about the need for data preservation on the web.
“The kind of people interested in Santa Fe Institute science – there’s a big overlap between those people and the kind of people who work at Google,” Graham said, and he wanted to get the two groups together and talking. “If other stuff falls out of that, that’s all the better.”