predicing future political developments using math models...?

[Hans de Vries]

AS in topic title - is the political world order deterministic enough so that at some point in the future we'll be able to predict what course world politics will take? Like you know, predicting with >90% certainity that 5 years from now country X will go to war with country Y and win after Z months of combat or that country U will have such and such foreign policy vis a vis country V etc Is that even possible?

 

Political "scientists" often make predictions about future events but these predictions are not based at mathematical modelling. So they may "predict" that by 2030 US will still be majority white and world's most powerful country or that China will face problems while India will be increasing in power.

 

Can we even be sure that in 2030 US will be a superpower and not a failed state embroiled in civil war?

[dimreepr]

Hari Seldon tried it through psychohistory, in Isaac Assimov’s “Foundation series” the problem (that Assimov solved with telepathic overseers) is the same that the met office face in trying to predict the weather and very well explained in your climate thread.

[swansont]

 

Political "scientists" often make predictions about future events but these predictions are not based at mathematical modelling. So they may "predict" that by 2030 US will still be majority white

 

I'd venture to say that population dynamics/demographics predictions do involve mathematical modeling and statistics, and the calculations are probably not carried out by political scientists.

 

http://en.wikipedia.org/wiki/Population_dynamics

http://en.wikipedia.org/wiki/Demographic_analysis

[MigL]

While you could model the combined behavior of a large number of people in the same manner that Statistical Thermodynamics does with particles, you would not get valid results.

Particles interact with each other in a specified way, all the time; people, on the other hand, interact in totally unpredictable ways.

 

You could, for example, have a lot of computing power dedicated to analysing the voting habits of very small demographics, and then doing the equivalent of finite element analysis to obtain the combined outcome.

But then, one person goes crazy and shoots the president...

[michel123456]

One way to make accurate predictions would be to establish a mathematical model of what should be done and then chose manually the exact opposite.

[Delta1212]

While you could model the combined behavior of a large number of people in the same manner that Statistical Thermodynamics does with particles, you would not get valid results.

Particles interact with each other in a specified way, all the time; people, on the other hand, interact in totally unpredictable ways.

 

You could, for example, have a lot of computing power dedicated to analysing the voting habits of very small demographics, and then doing the equivalent of finite element analysis to obtain the combined outcome.

But then, one person goes crazy and shoots the president...

Totally unpredictable is a bit of a stretch, but yeah, the possibilities are many and they branch quickly enough that even if you can predict general trends over X period of time, it's very easy for small difficult-or-impossible-to-predict things to have very large effects and drastically throw off your predictions.

 

That doesn't mean that you can't make predictions, just that the longer term you go, the less specific you can be with any hope of accuracy.

 

Predicting the winner of a war that won't start for 5 years is difficult. Predicting the number of months of fighting in that war is practically impossible.

[MigL]

That raises the interesting scenario where you predict that you will lose the war so you make sure you don't get into one.

[michel123456]

Although predicting that the war will begin at spring is not a difficult case.