I have a high opinion of this book, and the reason for that opinion is why I was not able to finish it. Greg Cashman seems just short of exhaustive in his analysis of what causes war and—though of course this is no fault of his own—this necessarily results in a frustrating litany of "...on the other hand" caveats. That is the state of the research. Fortunately, the work has at least started to inform any supervised or unsupervised machine learning I do on conflict datasets to try to categorize armed conflicts and/or see into the future a bit. For instance, I have been made especially aware of the dyadic nature of foreseeing conflict, and of the potential importance of changes of power balance over short time periods, which would possibly put a dynamical element in any models I come up with. In any case, while I am not going to go all the way through this book, I will keep it on hand as a useful reference.