A new simulation on the consequences of inequality has been released http://www.sesync.org/sit... and is the subject of intense debate – see for example:

http://blogs.discovermaga...

and http://www.theguardian.co...

Without commenting on the adequacy of the HANDY model in its specific formulation and assumptions, I feel compelled to jump into this debate in defense of simplicity in modeling and simulation. If we accept the view that models must replicate reality in all its complexity, we are doomed to misunderstand, and thus lose, all the value that models bring to the scientific process.

Models do not prove hypotheses, nor do they replicate reality. Rather, models are useful because they give us some insight into possible mechanisms and possible consequences. These new insights, then, feed into the iterative process of hypothesis-testing that underlies good science.

There is no problem with models that are “simplistic” – they are supposed to be. A model is an abstraction from – a simplification of – reality. The objective is to see if you can understand the essential mechanism(s) that drive the system, and what some possible consequences of this mechanism might be. If the model results give you valuable insights into certain real-world trends, then maybe you have managed to capture the essence of the mechanism. If not, then you probably haven't, and you will need to either revise your model or add components to it.

There is no particular value to making a model complex...you add just enough complexity until you are convinced that you have captured the essence of the mechanisms you are trying to understand. Meanwhile, it is fully acknowledged that there are many other things also going on in reality, that are not captured by the model.

For example, a demographic model that predicts population increase based on higher fertility rates does not need to include behavioral (social or economic) responses to large family size or depletion of resources. It is simply designed to tell us what the demographic consequences of higher fertility would be, absent behavioral responses. Then, understanding this, scientists and policy-makers can recommend responses.

Likewise, a demography/resources model that predicts collapse given certain relationships between the population and their resource base does not need to include the complexities of political, economic and social adjustments made in response to the situation. It merely shows us the possible outcomes, and leaves it to the archaeologists, anthropologists, social scientists, political scientists, economists, and policy-makers to debate the actual and hypothetical responses to these possibilities, and how they altered (or will alter) the outcomes.

In 2011 my Stanford colleagues and I published the results of an agent-based simulation of human demography and resource which explored whether inequality affects the relationship between resources and demography. http://www.plosone.org/ar... The simulation was a 5-year effort which included extensive sensitivity-testing, precise definitions of instability and extinction, three separate formulations of inequality (2-class, 5-class, and Pareto distribution of resources to individuals), 10 levels of inequality, with and without accumulation of wealth, in both stable and fluctuating environments, with and without the ability to migrate.

Human labor was not factored in, nor was any behavioral adjustment to population density or resource depletion (other than migration, in certain simulation trials). Nonetheless, to our great surprise, greater levels of inequality resulted in greater levels of instability in the system, and a higher probability of population extinction. At first we thought there was a problem with the simulation code, because we had predicted quite different results. After extensive analysis we came to realize that the inequality in access to resources had the effect of breaking down feedback in the system. Because mortality was sequestered in the lower classes, populations with higher levels of inequality in resource allocation tended to overshoot carrying capacity, become severely resource-limited, and then crash. Thus populations were less able to achieve an equilibrium state, where population was in some balance with available resources, as mediated by fertility and mortality rates.

Note that this was all in the absence of behavioral adjustments - it was purely a demographic response. When populations were allowed to respond behaviorally to worsening conditions through migration, the unequal and unstable populations ended up spreading outward and capturing all available living space. Once there was no more unoccupied frontier into which to expand, unequal populations remained dominant through a continuous process of instability-driven outward expansion into sites with other unstable populations.

The HANDY model finds that unequal populations collapse, although they apparently attribute this to the lack of labor as the working class dies off. Again, without commenting on the adequacy of the specifics of the HANDY formulation, I find the simplicity of the model useful and the results plausible.

If we want to criticize the HANDY model, and by the same token our Spread of Inequality model, let’s focus on the specific mechanisms that are postulated – not on the simplicity of the model, and not on the lack of exact parallels with past events.