Re: War & Poisson distribution

[Tuan V Nguyen <t.nguyen@garvan.unsw.edu.au>]

Hi Anh Huy and folks,

>Do you mean
>
>           0.69^x\times e^{-0.69}/x!,
>
>ie a Poisson dist with mean 0.69, instead of   
>
> P(X = x) = 0.60 x e**(-0.69) / x!?


	Yes. Sorry for the typo. Boy! it is hard to pass the 
eyes of a statistician :)

>Fixed rate on average? No. World population increases between 1500 and 1931
>and it becomes easier to mobilise 50000+ men, so the rate of 50000+ man war
>should increase. For 1500-1931 I don't think that this is cancelled out by 
>people less likely to resort to war and by fighting war with smaller 
proportions
>of the population. I think this is enough to say that the fit is a fluke
>or a fiddle.

	Interesting thought! 

	But given the goodness of fit of the model, I can 
hardly think that it is a "fiddle." Now, if the number of 
new wars initiated each year is constant (as we assumed), 
what does it mean? it may mean that society has a constant 
level of hostility. Years ago, I read an article somewhere 
which says that the number of wars ending in a given year 
can also be well described by the Poisson's law. Consider 
two things together, one may say that the average constancy 
in which society maintains is a reflection of a wish for 
changes, not for war or peace. 


>Independent? The wars in WWI were related, the Prussian wars were related, 
the 
>Napoleonic wars were related, the Louis XIV wars were related, the Swedish 
wars 
>were related, etc, wars by a powerful Ming or Ching emperor were related,
>wars after the Japanese modernisation were related etc. But it is reasonable 
to 
>think that in the old days in any one year clusters of related wars in 
different 
>regions of the world are quite independent. But how reasonable?
>
>Random. Quite reasonable.

	Do not you think it is contradictory (or I 
misunderstand your saying): The wars were related, but were 
random. I do not think they are random events, because, to 
my thinking, anything that can be modelled (by whatever 
law) can not be considered to be stochastic.

	Cheers,


	Tuan