Let’s start with a discussion of goals. The following paragraph is adapted from the proposal that we are showing to funding sources (http://www.globechange.org/ProposalIntro.html).
Our goal is to help bring the study of the Earth system into a modern scientific paradigm by using genetic programming (GP) as an umbrella for alternative methods of dynamic modeling, where the couplings among nominally distinct subjects are central. For a long but exceedingly incomplete list of couplings, see the Coupling Matrix at http://www.globechange.org/pdfs/03_couplings-general.pdf. In the current proposal, consideration of these subjects will be limited to just three: violent human conflict, economic phenomena and disease-transmission. For just one of many types of couplings, wars may directly impact health by disrupting the means of delivering health care and, of course, by causing death and injury. Though of varying coverage and quality, data are available in each of our three initial areas. We propose to create and validate scientific hypotheses and forecasts that test the usefulness of attending to all directions of causality among our three subjects, using the GP framework of methods described below.Â
This already raises lots of issues. How do we decide what are appropriate domains? How do we decide the connections to be represented, between and within domains? How important/feasible is it to insist on extensive empirical detail? Is there a hierarchy of goals that we should consider? Questions of method (genetic programming, vector autoregression, rational choice, …) are also important; are they contingent on the above issues? This blog is a place to share your comments.
