Despite my love for a great many works of historical research, and despite my abject dependence upon such works for the information needed to formulate and evaluate models of historical change, I never really understood what position historical works have in with relation to the behavioral sciences (economics, psychology, sociology, etc.). Historians are well trained in the scientific method, but they do not have any "theory" to go on in deciding what material to collect and how to interpret their findings. In a way, this allows historians to "just find out what really happened," without trying to force the facts into a preconceived paradigm and without using theoretical prejudices to decide what "facts" to collect. But then I learned that there are no "facts" independent from a theoretical framework that makes sense of the facts. And this appears true, at least if the "facts" involve higher-level constructs, such as "power," "purpose," "culture," and the like.
Well, now that I have found out that society is a complex adaptive dynamical system that will never be fully modeled, and have learned that the human mind is full of partially filled-out "templates" that can be deployed and refined given a new set of data, I no longer am uncomfortable with the historian's musings. Now Avner Greif comes along and tries to convince the reader, rather successfully, that historical research can after all be undertaken fruitfully and synergistically with the social sciences. But, and this is a rather giant but, only if we accept a broad notion of the social sciences in which transdisciplinary arguments are routinely made and upheld, and when a reasonable effort is made to adjudicate the differences across disciplines by developing novel arguments as syntheses.
The great merit of this book is to have largely accomplished this feat. Perhaps we are witnessing the death and rebirth of history as a discipline. Just a philosophers have come to the realization that knowledge of science complements their studies, perhaps historians---at least those who deal with big themes and broad expanses of time---will realize the value of integrating their discipline with traditional, rigorous, social-scientific studies.
Greif argues that the late Medieval Maghribi are best represented as a "reputation-based private-order institution." Greif's basic trade model is that of a merchant in one country hiring an agent in another country to carry out the many tasks involved in protecting and selling the merchant's goods in the agent's country. The critical point in this classical principal-agent model is that the agent has an incentive to cheat the merchant. If the volume of trade is sufficiently high, and if the agent can be harshly punished for a transgression by being fired, then the merchant can use the threat of dismissal to induced the agent to behave honestly (for a general model of this type, see Samuel Bowles and Herbert Gintis, "Walrasian Economics in Retrospect", Quarterly Journal of Economics 115,4 (2000):1411-1439; and "The Revenge of Homo economicus: Contested Exchange and the Revival of Political Economy", Journal of Economic Perspectives 7,1 (1993):83-102). However, in many cases, including those of medieval trade, a more sophisticated social process was required, in which the agent had to sustain a general reputation among all merchants as being honest, so that the report of cheating by one merchant would lead to the perpetual ostracism of the agent from dealings with other merchants for the foreseeable future. The construction of this model brings Greif into a significant game-theoretic model-building exercise, from which the careful reader will learn much about the general value of game theory in elucidating social research.
One of Greif's greatest achievements is to show convincingly the value of a model in which individuals obey both the canons of rationality so beloved by the economists, and at the same time exhibit the moral inclinations, emotions, social attachments, and internalized norms so beloved by the sociologists and social psychologists. For a general treatment of how these aspects of human sociality interact creatively see my book The Bounds of Reason: Game Theory and the Unification of the Behavioral Sciences (Princeton University Press: 2009). I there argue that contemporary sociology's descent into theoretical oblivion is due precisely to its rejection of economic reasoning, and contemporary economics' descent into theoretical aridity is due to its rejection of sociological reasoning. I have an ally in Avner Greif, who combines the two modes of reasoning quite adeptly.
Indeed, Greif goes to great length to convince the reader that this is not simply a standard exercise in economic reasoning, because the "perfect information" assumptions of his model are not historically plausible. How can a merchant be sure the agent cheated him? Why can't a merchant blackmail an agent into giving him a better deal by threatening him with ruining his precious reputation? Indeed, why should a merchant report that he has been cheated at all? Does he do this out of love for justice, out of the desire for revenge and retribution, or because he cares about the well-being of his fellow merchants? Why do Maghribi trade networks exhibit strong ethnic regularities, but not strong kinship regularities? These are all obvious questions that a traditional game-theoretic analysis cannot handle, but can be addressed using a modicum of standard sociological analysis.
Greif deserves a careful reading, and I consider this a great book, because the author has gone where few have had the guts to tread. However, this book is a beginning and not an end. There are many faults, all of them associated with the books great novelty. The books should be at best half the size, and the social theory should be much more sophisticated and better integrated with the formal models. For instance, Greif alludes to several areas of social theory that might help answer the questions I posed in the previous paragraph, but he does not in fact answer them head-on. Why does the merchant bother informing on a cheating agent? How does a merchant determine that he has been cheated? What is the status of ethnic ties? I would like to have seen a much more pointed analysis. Instead, we get very general high-level theorizing that leaves the key questions "sort of answered," but not really.
The following is typical: "The Maghribis and the Genoese were constrained by the same technology and environment, and they faced the same organizational problems. But their different cultural heritages and political social histories gave rise to different cultural beliefs. Theoretically, their distinct cultural beliefs are sufficient to account for the diverse institutional trajectories of the two groups. Cultural beliefs may thus have had lasting impact despite their temporary nature." (p. 300) Statements such as these are both theoretically trivial and at the same time hard to accept. Are not culturally beliefs codetermined with other aspects of social life, rather than being autonomous and determining?
In sum, this exceptional book has a theoretical tentativeness that marks the opening of a new era in social theory. The aspiring social theorist and/or historian will see the future in Greif's adept reasoning, even if through a glass not darkly, yet not fully illuminated, thus leaving work to be done by standing on the giant's shoulders.