I found an interesting quote in a book I’m reading titled “What If?”, a collection of essays by historians dealing with counterfactual history. (I’m currently supporting a program developing techniques for counterfactual forecasting.). The quote by Geoffrey Parker is “Counterfactual experiments in history should always include two limitations: the “minimal rewrite rule” (only small and plausible changes should be made to the actual sequence of events) and “second order counterfactuals” (after a certain time, the previous pattern may reassert itself).” I agree with the first rule but I note that small changes can result in widely differing immediate consequences, several of which are described by other historians in the book. The second rule is interesting and I’m not sure that I agree with it. I think Parker is trying to say that broad trends will guide the development of history in the long run and will override the transient effects of the counterfactual. I’m personally not sure that this is always the case. I think I would rephrase this as saying that near term consequences of the counterfactual scenario are easier to predict because they will be dominated by first order effects. Long term consequences are more likely to be driven by second order consequences and those are inherently difficult to predict. I think there’s a class of consequences that isn’t covered but that is perhaps the hardest to predict. These are the medium term consequences as the historical system hasn’t reached a new equilibrium but there is a substantial mix of direct and indirect effects at play.