A systemic approach to change
In twelve pages, he makes a convincing argument that WWI's trench warfare was not an example of the prisoner's dilemma. Furthermore, he suggests that the lesson from that trench warfare is that we might better achieve positive change by setting up immediate benefits from cooperation (or desired behaviors) and watch for other forces that might resist such cooperation, rather than relying on long-term motivation for good behavior as being sufficient to overcome short-term penalties for such behavior. In other words, don't expect people to trade off long-term benefits for short-term pain, when you can provide both short- and long-term benefits.
If you're interested in more such insights, check out his blog, Statistical Modeling, Causal Inference, and Social Science.