[duncanawoods]

> nearly all wars in history have not started as a result of rational actors making optimal decisions

Is this true? I see them often as the conflict between local / global goals where the real problem is that each step in the escalation from peace to war is tragically rational - each decision is a lose-lose choice where one path is a guaranteed small loss but escalation offers a chance to win despite the possible apocalyptic outcome. The whole matter is so bad because peace-time power-plays build up an unstable system containing things like interlocking mutual defence treaties deliberately designed to escalate the stakes of minor disputes.

My understanding of game theory may be weak because I don't see how things like the prisoners dilemma have an "optimal solution" instead they just illustrate choices whose desirability match different risk appetites. Your risk-appetite is like a value-judgement, taste or personality - judgements of what risks are rational is personal and relative instead of a mathematical absolute. Even on seemingly simple probabilistic problems where you can calculate expected-value seem fallacious when you only play the game once and you'll be dead within 40 years no matter the outcome. Can game theory prove "Live free or die" is irrational?

[0xcde4c3db]

> I see them often as the conflict between local / global goals where the real problem is that each step in the escalation from peace to war is tragically rational - each decision is a lose-lose choice where one path is a guaranteed small loss but escalation offers a chance to win despite the possible apocalyptic outcome.

This sounds a lot like the dollar auction game [1], which is used to study the disconnect between individual actions being rational and a clearly undesirable outcome. Basically, the game is a variation of an all-pay auction in which only the top 2 bidders pay. A dollar is auctioned off in bids of some fixed increment. At each turn, winning the dollar is a preferable outcome to losing your previous bid. Even after the price goes over $1, winning the dollar represents a smaller loss. In that sense, the rational thing to do is bid indefinitely high. It's supposedly a common outcome with human bidders that the winning bid is over $2.

[1] (PDF) http://www.math.toronto.edu/mpugh/Teaching/Sci199_03/dollar_...