With the exponential (or near exponential) growth in computer processing power and speed, the tendency to look to silicon for answers about the future increases. Particularly in the case of complex systems, with many drivers, variables, actors, inputs, and outputs, computational power can make the linkages and interactions seem more apparent and predictable. And, with small, understandable, and therefore eminently mapable systems (like a business, or even an industry), modeling continues to serve as a useful tool. It is considerably less useful, however, answering questions about the future when it attempts to grapple with truly massive and dynamic systems.

For example, it is an oft noted truism that one of the most difficult things to model is weather. We understand – in a physical sciences sort of way – the weather system. There are many variables, many of them dynamic, and modeling dynamic variables leads one immediately down a nonlinear path resulting in torrential downpour on the one day you were convinced by the weather guy to leave the top down on your convertible. That’s the weather, now imagine something vastly more complex, where we don’t even know what all the components are, where we’re unable to predict interactions and reactions except maybe at the grossest levels, and you’ve got climate change.

Yet, our reliance on fast silicon convinces us that more processing power will allow us to really understand what’s going on with climate change. Certainly, the folks at DOE think that the Berkeley Lab’s ESNet project (superfast connectivity, massive parallel processing) can get us pretty far down that path. Despite the fact that we may be able to model storms in the Atlantic over the last 1000 years, until we understand how humans – then and now – interact with that system, how oceans store and release carbon over millennia, how solar radiation affects upper atmospheric change, and a host of other issues, the models aren’t entirely useful. We still have much more fundamental work to do that has nothing to do with gigantic models before we’re remotely in the position of being able to say that we not only understand the climate, but how it is changing, and most importantly, what it’s going to look like in the future.

While the power of models to accurately describe a future state is wishy-washy at best in describing physical systems (where at least in theory all of the components can be identified and characterized), when humans are added to the mix things get really unstable. The problem, of course, is that humans are notoriously unreliable variables in unconstrained systems.

Modeling the specific actions of a floor manager in a factory is possible because there are only a select set of actions permitted by the system. Modeling the actions of an individual going about his or her day is something else entirely The latter example is one of a relatively unconstrained system). Now add 4 billion more individuals, throw in a variety of competing, overlapping, or contradictory social, political, and economic systems, some diverse and varied geography, as well as specific understandings and expressions of self-interest, and you’ve got a very different animal indeed.

Modeling at that level is only meaningful in the very broadest of terms. Understanding requires something else entirely, and something that sits at the very heart of useful futures work: imagination. That’s where we’ll pick up next time.