Predicting Outcomes - Social Experiments and Simulations...
 

Being involved in product development, six sigma and research most of my life, I'm no stranger to experimental designs. The classical approach, onthogonal methods, and factorial all have there place and can be amazingly helpful at zeroing in on a property or performance of a given set of variables - both dependent and even independent.

Considering the vast knowledge of human, social and cultural behavior combined with serious computing power, I gotta believe that you could (with a certain amount of accuracy) predict the way in which various media, political and social decisions will play out on various groups of people including countries.

The folks in charge of world finance as well as national defense must be applying some sort of computer simulation before making certain decisions.

I know it is very complicated but with all that we know and our computing power its gotta be feasible - seems like it would take out many of the foolish mistakes that are made on a geo=political and social level. Isn't it fair to say that if more folks who were in power used these techniques less problems would occur?

You are SUCH an engineer!

You want a computer analysis that will predict outcome and behavior? 50% of the worlds population is female. Get busy, Mike. It's going to be a complex analysis. :D:D:D

Already being done.
Each "dramatic" story put out by the major news outlets are cross-referenced to incidents of key words on the net, with application of signifigance filters.

That's what search engines and supercomputers are for.

Goldman Sachs uses computers to predict the market..... Goldman Sachs Alpha to Fail? - Forbes.com

We mess with lots of simulations. Google psychsim, acats, jcats, urbansim, and on and on. Some things are easier to model than others. Sociocultural is a pita.

This will not be done for the same reason that politicians have no interest in seeking an optimal tax rate: it reduces the power of the politicians.

If their claims could be conclusivevly proven or disproven, they wouldn't be able to drive voters with emotional appeals. They also wouldn't easily be able to claim one motive while actually having another. (Claiming legislation is for one purpose when it is actually for another purpose.)

I assume some sort of averaging of individual human behavior comes out to slightly less than random. Multiplying just slightly less than random by the number of people comes out to just slightly less than random. I could say even more if I knew what I was talking about.

There's a good book on this butterfly effect that sums it up: "The Numerati" by Steven Baker.

Paging Mr. Seldon. Mr. Hari Seldon, please pick up a courtesy phone...

Man you're way too complex, just get 1000 typewriters, and 1000 chimpanzee's... you know the rest

Good one Moses!

OK, lets try. Women Model - Take the inverse of all reason, use that as a constant, add in a very predictable monthly cycle, subtract any and all conversations had with girlfriends then multiply emotional intensity of holidays, anniversaries, interactions with in-laws and the rating signature of Dr. Phil, Oprah and The View for that day and you may begin to establish a bias that an algorithm can be built from (provided you attach a random multiplier that has no dependency on anything yet an amazing amount of sway!).

There, simple...