Reply #4 to other comments posted about my proposal; February 1998

----------------------------------------------

Are you asking if the model I've described can predict the future better than existing models, and adapt dynamically the way it makes such predictions, based on whatever change in the model it encounters at any time?

One way you can look at this is like a weather or economic forecasting (as I suppose them to be; anybody remember the mass psychologists in Asimov's _Foundation_ series, who tried but couldn't predict "The Mule"?): You build a program that allows various actors (individually and/or as aggregates) to interact with each other in ways that are too complex to handle in your head. You provide a dataset representing the initial state, then the ongoing interactions of the modeled actors change the internal state of the various actors as they affect each other, respond to those effects by acting further, and the effects cascade and multiply; you let it run for a time, and you check the results.

The big difference here is that I'm not contemplating specialized models for weather, social studies or anything else. I'm suggesting building an engine that parses all the text available to mankind, uses the information contained therein to extract nouns as well as attributes and descriptions of behaviors of those named concepts (in a systematic way) and to build therewith a computerized model of *everything* known. To the extent that we can gather data on human beings (individually and in groups) those would fit into the model as readily as any other data. Having gathered the "state of the universe at a given time" and the "principles that govern it", we can then "predict all things, barring outside factors"--very roughly and haltingly at first, and constrained by the limits and self-contradictions of our own knowledge.

So the "rules" the model uses to answer questions would be, at any point in time, simply the behaviors of things that had been observed in the real world; based either on the parsed text, or on the updates to the model provided by humans as they "experiment" or "work" within it.

Remember that this is still a concept--while I can envision clearly in my mind how this would be built, it hasn't been done; some of the tools are a year or two (or more) from general availability (as far as I know and as far as I can get info from them on it), and it would take a crack software engineering team to make it work. (Maybe 3 years to get a useful system going; maybe 10.) Of course I don't want to make it sound impossible at all--I believe it will happen and am definitely anxious to participate--sooner than later if possible.

I hope this helps some. Can you elaborate further on your last message; especially on what you mean by considering both 'rational' and 'irrational' in this context (is this the rational/irrational behaviors of humans, or the irrationality of unpredictable business environmental change?), and on what you mean by "best response" (is that the best response to a business problem, like that of mapping strategy in the changing environment?).