Moved from D1 forum

[kandrathe]

Economists need to spend less time reading Smith and Marx and looking at history and more reading Johnny von Neumann and Oskar Morgenstern and looking at game theory. Especially the results from studying and modeling the prisoners' dilemma which are coming very close to explaining what we call "ethical" behavior, including ethical economic behavior. Not to mention explaining why we are so much ruder on the highway (automotive or informational) than in person.

My sense is that economists spend almost zero time with Smith these days, and less with Marx, but quite a lot with game theory.

What they don't do so much, is go back and reapply the basic insights gained from game theory to the big picture ideas of the old economists, and instead spend their days designing ever more esoteric mathematical contraptions with it. There seem to be sharply diminishing returns to the insights gained from such formalizations, but since the win for the economist is in publications rather than understanding per se, I'm not holding out much hope.

I watched an interesting show on Hawking recently, in which he described the elegant manner in which imperfections cause folding in various aspects of our reality.

In general, humans are pretty simple minded creatures, and so we cling to our new toys and theories. Hopefully, we are in an aftermath of an era which is characterized by over reliance on what computers can do for us in terms of allowing us to create predictive models.

What is frustrating about software construction in general, and I extrapolate this to other modeling in general, is that you can describe a great deal of the normal behavior of a thing with a small amount of concise and elegant design. But, then, the devil is in the details. As you factor in exception, after exception, what started out as a simple elegant algorithm quickly becomes a Frankenstein of a twisted logic (structure).

Enter Chaos Theory... Or, the non-linear equation. The problem this raises in my field is that in general, humans are pretty simple minded creatures. We rely on a process of unraveling complicated things in an ordered process of decomposition. Based on that decomposition, or simplification, it is easy to draw a set of logical (linear) conclusions based on the examination of parts of the complicated problem. However, the world is probably not so much described by linear concepts, but rather remains and will remain predictable in special circumstances where brief linearity exists or is enforced.

Really though, because of the increasing velocity of change, it is becoming very, very difficult to use computer modeling, or statistical methods to make predictions. These methods can give us insight and a brief glimpse of possible conclusions based upon the limited reality that is modeled (e.g. our discussion on predictive climate change models). I believe and tend to trust more of a Delphi method, gathering many predictions/estimates (that use reasonable methodologies) and then based on all that data, make a reasonable justifiable prediction.