The utility of better weather forecasts is clear – and far beyond personal benefits like planning picnics and wearing the right coat.  Farmers can use the information to choose the right days to plant and harvest their crops.  Shipping and transportation companies of all types choose which routes to take and whether or not to shut down lines based on impending weather conditions.  Construction companies use forecasts to decide what to build when, and energy utilities rely on weather predictions to decide how much fuel to buy and when to turn on extra generators.  Perhaps most importantly, foreknowledge of impending storms, floods, and droughts can significantly lessen their impact, saving lives and property all over the world.

What are the implications of new computing power enabling forecasting ability to continue to increase at a slow but steady pace?  Not much, at least in the scheme of things.  More accurate forecasts will save each of these companies slightly more money each year; farmers will be slightly more productive; and some extra lives will be saved.  There will continue to be unexpected weather events that catch forecasters off guard, but such events will slowly decline.  However, a lot of money will continue to be spent on meteorological research and development because globally, even these incremental improvements can save a lot of money.  Currently, for instance, “weather forecasts save America’s airlines around $500 million a year.”[69]  Probabilistic forecasts in particular (the “70% chance of rain”) can alert organizations to possible threats and provide the basis for economically efficient responses based on the given levels of uncertainty.

Weather has been explored here as a case study of a complex adaptive system with important ramifications for humanity.  The trends observed above most likely apply to computer modeling and prediction of other complex and chaotic systems.  For example, directly related to weather forecasting are models of air and water pollution that predict what happens to released pollutants.  Also closely related are climate change models mentioned earlier, as well as related (and often embedded) models of ocean processes[70], ecosystem change, river flow, forest growth, land use, etc.[71]  Finally, it seems possible that useful models of the dynamics in social systems might emerge, for instance giving campaigners new ways to predict the outcome of elections.  For the same reasons as in weather forecasting, tipping points or major surprises are unlikely to result in any of these fields.

As for the utility of these models, it should be kept in mind that most of them give us little to no assistance in figuring out how to actually lead the world towards a favorable course of development.  They predict with increasing accuracy the buildup of pollutants and locations of new deserts, but they are not designed to tell us how to avoid those things.  In the case of weather, we don’t know how to change it anyway.  In the case of many other models, what not to do is already obvious:[72] don’t spew sulfur into the air, don’t destroy wetlands, don’t clear old growth forests, etc.  The main utility of the models is to inform us in as explicit terms as possible the ramifications of failing to achieve what we already know we should do.  They can also provide forecasts of relative impact, telling us where to pollute if we have to do it somewhere.  This sort of “optimal polluting” is certainly less effective than reducing pollution or other harmful activities directly, but it allows decision makers to find the most cost effective solutions.

Thus probably the most important use of complex adaptive system computer models will be for increasingly numerical scenarios.  If the computer models are realistic enough, their numerical, probability-based predictions may be more useful and more appealing to decision-makers than the more subjective types of scenarios described by Hammond and others.  If an organization can decide how much money they want to spend or how much pollution they want to curb, such models can conceivably allow them to virtually try out a lot of possibilities and find the most cost effective option. 

Predictive models will likely be an increasingly important tool of a market economy struggling to transition to sustainability in as cheap a way as possible.  Yet the same models could just as easily be used to determine how to pollute in a way optimal for “fortress” communities while ignoring the rest of the world.  In a similar duality, incremental advancement of forecasting accuracy means that prediction centers around the world have kept pace with each other, but on the other hand prediction centers are only possible where the economic and political situation allows buying leading supercomputers.  Ultimately, humans remain responsible for choosing the direction of our future; predictive computer models are simply a tool that will allow us to follow that path more intelligently.

 
References

Footnotes

[69] “Predicting the micro-weather,” 24.

[70] Holland and McWilliams, 57.

[71] Mahlman.

[72] Mahlman.