A review of Social Physics: How Good Ideas Spread — The Lessons from a New Science, Alex Pentland, 2014.
Pentland and his research group have appeared to have discovered a simple model of human behavior with great predictive power. By snooping on people's cell phones, they can reduce typical human interactions into a set of interacting finite state machines, and by noticing just how regular those behavioral patterns are they think they understand ideas. Quite evidently the populations they study live routine, predictable lives. (Perhaps we all do.)
My biggest criticism (& thus the non-recommendation for the book) is that the technologies for making these monitoring of human behavior that Pentland describes are dehumanizing and a grave insult to personal privacy. The authors vague promises of a personal data store that would broker our most personal information are unrealistic. If these plans pan out, we will always be watched over by machines that seek only to predict our patterns of behavior and exploit them. An algorithmic prison awaits us.
Every once in a while the author describes unexpected behavior by individuals thus modeled, and betrays an element of surprise that we might step out of our everyday paths into something his system doesn't contain. It's a glimmer of hope in a dystopian world that we might surprise our ever-present overseers and do something that their social physics does not anticipate.