Predicting Social Systems – a Challenge!

After Gary Polhill’s talk on prediction at Social Simulation 2019, there was a lively discussion about whether there are any simulation models (particularly agent-based simulations, but this was not entirely clear) that have predicted aspects of complex social systems.

Thus we (Gary Polhill, David Hales and I) are issuing a challenge to those who think there are such examples to send us the documentation about these predictions. We can then better assess the extent and manner of such predictions. Suggestions do not have to be limited to one’s own research, but could be any published work. Details at:



So far, only one person has presented me with a possible example, and that turned out to be more explaining known data rather than predicting anything known. I have started to survey examples in the literature, but not yet found any convincing cases of prediction-before-the-fact

Obviously there are now a host of COVID19 models that aim to predict in the widest sense (see but (a) many of these teams say they are not forecasting and (b) none has a good track record so far, but have been continually adapting their models to fit current data which makes this more like an explanatory project.