Tag Archives: Westeros

Washington Consensus neoliberal types like Matt Yglesias think that Walder Frey’s actions were justified by Robb Stark’s breach of implicit contract. As a Washington Consensus neoliberal type myself, I’d normally agree. But Yglesias is wrong:

For a society such as that of the Seven Kingdoms of Westeros with its very limited formal capacity for legal administration, the availability of trust and custom is even more important. But as we saw on last night’s episode, sometimes the gains from a breach of trust can be temptingly large.

When considering this matter, I think it’s worth saying that Robb Stark’s actions are arguably even worse than Walder Frey’s. Whether he broke his original promise out of a misguided sense of honor (as in the books) or simply true love (as in the show), he’s actively underminingthe key social and political institution of Westeros. Arranged marriages run afoul of modern sensibilities and strike us as cynical. Turning such a marriage aside for honor or love strikes as us idealistic if perhaps misguided. But that would be a mistake. Marriage contracts are the only means that the major houses of Westeros have for forming alliances and ending conflicts. Walder extracted a high price from the Starks in exchange for his assistance in crossing the river, but the Starks were asking a favor of considerable value. The coin that Walder demanded—marriage arrangements for his children—was absolutely the standard sort of exchange for major political favor. A world in which one house promises another house a marriage alliance and then turns its back on the promise on a whim is a world in which houses are going to be perpetually at war.

Yglesias’ argument is circular, therefore it’s more an explanation of why Westeros is a shitty place to live than a justification of behavior thereof. See the emphasized: the only reason this is true is because nuts like Walder engage in extreme, violent, and arguably evil retribution. Had Walder accepted a fair price, Robb would not have broken the promise. And even if he did, if Walder reacted in a measured manner, hostility would have been avoided.

That the western world got rich from the rule of law and enforceability of contracts has been argued ad nauseam, and I have no intention of disputing that. However, fact is, some western societies have done better than others by allowing “easy” breach of contract. For example, in America bankruptcy is a lot easier than it is in Europe. That’s one reason why our recovery from the subprime crash has been rather more resilient.

Not all contracts, such as those made under duress, are fair and good. By legally enforcing such we would be perpetuating a Pareto-inefficient society. Yglesias sidesteps this by noting Westeros is devoid of a robust legal system and hence must rely on tradition. However, traditional norms become the imperfect legal moderator in such backward societies. To adhere to bad contracts extracted due to monopoly power would be just as ridiculous there as here.

But the most serious flaw in this defense of Walder is the application of game theory:

Walder’s decision to respond to betrayal in kind is extreme. But tit-for-tat is a viable strategy in the iterated prisoner’s dilemma and arguably represents a reasonable approach. We, with direct access to Stark/Tully perspectives, know that there Edmure Tully fallback marriage is a perfectly good faith arrangement but the view from the Twins is not so clear. You really wouldn’t want to elevate a new King who right from day one is betraying not only the Iron Throne but also his own bannermen. If you squint at it right, you can see what Walder was thinking. By contrast, the betrayal committed by Roose Bolton is pure ambition and cynicism with no justification whatsoever.

The “tit-for-tat” heuristic is highly robust in discrete, theoretical models. However it fails to explain dynamics in real-life, which are better approximated by continuous rather than step iteration. There’s a body of research on this, prominently from Peter Hammerstein (2003):

Game theoretic explanations of the evolution of cooperation in humans and other animals relies on assumptions — rational players should never cooperate, cooperative behavior is explained by direct or diffuse reciprocity, animals can do the mental bookkeeping necessary to reciprocate with multiple partners over time — that are not always or often borne out by data, necessitating new conceptual tools


Game theoretic explanations of cooperation involving tit-for-tat strategies and reciprocal altruism are not supported by a large body of evidence. Only a small number of animal examples have been found. Simple models of repeated games do not match the circumstances of evolutionary change. Partner switching and mobility counter the assumptions necessary for reciprocal altruism as a stable evolutionary mechanism. Reciprocity requires significant mental machinery “ how do organisms determine whether the actions of others are intentionally or unintentionally cooperative or uncooperative? Alternative conceptual schemas such as partner markets “ making it unprofitable for partners to switch “ offer alternative conceptual schemas. Emotions may play a role in mediating complex interactions in which intentionality and reputation play a part.

Indeed, the most devoted rational expectations scholars might beg to differ with empirics thereof, but I don’t expect Yglesias to fall in that stubborn set. More importantly, there’s very little “tit-for-tat” about Walder’s actions. Normal Prisoner’s Dilemma models look (something) like this:

P1/P2 Honor Breach
Honor 2, 2 -2, 3
Breach 3, -2 0, 0


Robb/Walder Honor Breach
Honor irrelevant -200000, 3
Breach 500, -2 irrelevant

My figures are rather subjective, yes (Note the point of the matrix is to show complete asymmetry between the two “tat” plays). But I think I correctly assume that the utility derived from falling in love with a pretty girl is less than that lost from:

  • Falling in love with a pretty girl and loosing her.
  • Watching your unborn kid get stabbed in his mother’s womb.
  • Watching your mother die.
  • Being killed.

Actually, I probably need to add a few more zeroes. I bet the dynamics play out a little differently. Furthermore, if Yglesias does intend to play the rational discrete expectations card, in which the “tit-for-tat” is the correct strategy, I would inquire why Walder chose a subgame imperfect strategy. Game theorists like to identify certain threats as “credible”. Inflicting pain on yourself, wife, and daughter to inflict even greater pain on your enemy would not fall in that category and fails to be rational.

Indeed, I would be incorrect if Walder was a “sadistic utility monster” in which the disutility of his enemies disproportionately contribute to his own happiness. However, I would argue being a sadistic utility monster is unethical, and hence Walder’s actions under that pretense is not justifiable in a court of law… Or Westorean tradition.