We review, and extend, one of the classic dynamic models of conflict in economics by Richardson (1919) and Boulding (1962). It turns out that the stability properties of the model change if one takes a more realistic 'incrementalist' view, and that chance / friction can easily be incorporated into the standard model by defining a probability of (de-)escalation. This analysis is not just a study in the history of economic thought, but also relevant for the development of simulation models for the analysis of conflict dynamics.