Rom and Ori Brafman's Sway is a light, Gladwellesque traverse through the greatest hits of behavioral economics and social psychology, ranging from prospect theory to groupthink to the fundamental attribution error of allowing a label to overcome reality: http://www.swaybook.com/
Chapter Two, "The Swamp of Commitment," uses business school professor Max Bazerman's "auction a $20 bill" exercise as a parallel to LBJ's decision-making in Vietnam and, more briefly, to W's in Iraq. In the auction exercise, a class of students is given the chance to buy the professor's $20 for as little as $1. The nasty catch is that the second place bidder as well as the first place bidder has to pay the professor the value of his or her highest bid. The result obtained by Professor Bazerman is that two bidders, neither of whom wants to lose, both wind up bidding more, sometimes much more, than $20. The conclusion suggested by the brothers Brafman is that Vietnam and Iraq embody the auction exercise in real life, with US leaders initially hoping to win cheaply and then sucked in ever deeper by an irrational although behaviorally compelling unwillingness to let sunk costs be bygones and withdraw. Here--as is usually the case in their book--the Brafmans uphold rational choice; by expounding people's deviations from RC in practice, they hope to encourage rationality.
Okay...but here, as is often the case with RC arguments, some cautionary notes are in order. The world is full of models of how people deviate from RC, and it is often very hard to tell which if any model applies. Further, it is sometimes very tricky to tell what the normative implications of an RC model are, assuming its applicability.
Instead of the "auction $20" exercise, imagine a "Give me $5" exercise: Student A is given $20. Student B can ask Student A for $5 on behalf of Students B-Z. If Student A agrees, Student B must promise on behalf of the rest of the class that A will not be asked for more. This promise is not binding on B or the other students, though, as the professor explains. If it is broken by any of students B-Z asking A for another $5, another promise is made by the requesting student, and so on. If A refuses at any point, the game ends and A has to give all his or her money to B-Z if a coin flip turns up tails; if the flip is heads A gets to keep what he or she has left. The dominant interpretation of RC says that in the "Give me $5" (or "Munich" or "Falling Dominos" or "Bernie Goetz" game), a perfectly rational A who knows the other players are perfectly rational should refuse the first request for $5. In practice, I suspect many students would wait to resist, a la Neville Chamberlain, and that a majority would fork over the first $5 if the game were played with only A and B rather than with B-Z. To refuse the first request for $5, defensible as it is under RC, suggests that you are an untrusting person who would rather fight, with an expectation of only $10, rather than share and trust B or B-Z to keep B's promise and leave you with $15.
None of this is intended to debunk RC. The more RC stories one can tell and compare and contrast with an inevitably messier world the better, as long as one does not fall too much in love with any one of them and commit one's own RC version of the fundamental attribution error.
A VC story about how a society can reduce the likelihood and the magnitude of Chamberlain-style errors of appeasement on one side as well as LBJ-style errors of escalation on the other side: In the current political system in the US and most other wealthy nations, a center-left that is more worried about errors of escalation competes for popular favor with a center-right that is more worried about errors of appeasement. But the coalition of both sides as well as the group of voters in the middle contains numerous voters who not zealots for the party's tilt in favor of or against military force. Compared to a consensus system like that of the US in the early 1960s in which the use of military force does not divide the sides, vote-seeking politicians of both the more force-averse center-left party and the more force-favoring center-right party have stronger incentives to look with care at the costs and benefits of military force and avoid being trapped in an escalation game or an appeasement game. If there had been more of a division on military involvement between the parties under the New Deal system, Vietnam might well have been more like the comparatively manageable messes of Iraq and Afghanistan rather than the much deadlier quagmire it actually became.
How about a flipped VC story, in favor of the New Deal system and against the current more polarized division between the Democrats and the Republicans? There are plenty of those, some of which I find convincing. But I can't think of one as of now that seems as convincing on Vietnam vs. Afghanistan/Iraq as the one I just gave.