In his keynote at EMCworld 2012 in Las Vegas, Pat Gelsinger proposed a tongue-in-cheek equation for the “physics of information technology”. He also used analogies drawn from the world of physics to talk about the transformations that are occurring in information technology — describing the transition from applications to data as a shift in the center of gravity, as well as discussing data in terms of concepts like structure, mass, velocity and half-life.
Pat’s analogy set me to thinking about the theoretical bases of cybersecurity. Physics is, of course, a fundamental discipline for computer science, from the latency of global data distribution to limitations in data density to use of quantum physics in key distribution. Certainly cryptography relies on fundamental mathematical principles of prime factorization, for example, and security event analysis draws on fundamental statistical analytical techniques of correlation, clustering and so on. But the most fundamental basis for security must lie in the interaction of attacker and defender. And the fundamental basis for understanding that interaction lies in the disciplines of game theory. Game theory provides important insights into battle strategies, perhaps most famously in Hayward’s analysis of the Battle of Bismarck Sea in World War II. Similarly, it can provide important insights into the security strategies for the “cybercrime battlefield”, as my colleague Uri Rivner calls the conflict of cyber attackers and defenders.
Though game theory is much better known in the domains of economics and military strategy than it is in cybersecurity, there has been a lot of research over the past ten years in applying game theory to understanding cybersecurity and developing effective cybersecurity strategies. There have even been conferences dedicated to this topic, particularly the Gamesec conference that started in 2010 and will take place again in November of this year in Budapest.
Some of the most interesting research being done in this area is by our own folks at RSA labs, in collaboration with Ron Rivest. In February of this year, Ari Juels, Alina Oprea, Marten van Dijk and Ron Rivest published a paper called “Flipit: the Game of Stealthy Takeover”. It’s a superb paper, insightful and challenging, yielding interesting conclusions not only about the interactions of attacker and defender but also about how game theory can be applied to cybersecurity. For example, the paper demonstrates that Nash equilibriums can be discovered that provide extremely useful insights into the most effective defensive strategies in exactly the conditions of continuous compromise that enterprises currently face.
We’re just beginning to realize the insights that game theory can provide about effective strategies for cyberdefense. But it has the promise of being very fruitful and pragmatic research. For example, in April I spoke at MIT about applying the insights from the Flipit paper to decisions about deploying key management systems for the hybrid cloud. Especially in the world of the cybercrime battlefield, a world where the fortress walls have been torn down and the moats have been dug up, game theory may just give us the insights and theoretic basis we need for effective cybersecurity.