Revisiting a few points:
1. Super computers simulating billions of individuals sounds amazing. As a matter of fact it bis becoming possible to use swarm intelligence techniques in multi variable optimization problems. The technique essentially consists of randomized individuals defined using a class, which have characteristics which are either genetically bred over generations (as in genetic particle swarm optimization) or are communicated to other particles (individuals) by means of a global best value (which stores the characteristic of the "fittest" individual) and also a particle best value (which stores the characteristics of each particle's best values). The "fitness" is decided based on some kind of objective function definition. Its fairly straightforward to implement this for a simple optimization problem for multiple variables, sometimes as many as 30 or 40. With psychohistorical problems, I can imagine the number creeping up to hundreds, thousands and possibly millions of variables. Here comes the paradigm of AI. AI can help us identify design variables since AI can be used in pattern recognition technologies. Indeed the human brain is far more effective than even supercomputers at pattern recognition and not number crunching, at which the human brain is ten million or so times slower than the average PC.
Mathematical models progressing from a swarm intelligence framework seem independent of the variables defined also; the technique is independent of the variables. This makes it applicable in all kinds of areas.
Please see these sites for more information:
www.swarmintelligence.org
en.wikipedia.org . . .
www.particleswarm.info
There are also other non-gradient optimization techniques like Simulated Annealing and Genetic Algorithms, but PSO has proved itself against both these techniques.
2. Mrkoconnell, I would like to learn more about your work on the object oriented approach and also the "grand unified theory of humanity". It sounds sufficiently ambitious to have piqued my interest! IMO the software engineering approach is a sound first approach, but the nature of organization in the model will have to continually upgrade, possibly get upgraded by itself. I propose that the computer itself generate code genetically, using successful bits of code which it can compose further iterations to itself.
3. Mental models. How does the Jungian psychology model (also used in the Kiersey Temperament sorter) work in constructing a mental model? I did some kind of small analysis to determine factors like sociability, intelligence, relative popularity, etc for a small set of individuals using the Jungian model, with normalised values of the four preferences, but didnt make any progress that time since the definitions were themselves vague. I need some literature on psychological prototyping, possibly a set of results of the brain as a neural net and its performance.
