Bryan Caplan has a series of blogs in which he attacks Levy and Peart on analytical egalitarianism. Caplan's main point is that people in fact differ and Levy and Peart are wrong because they fail to recognize this. Read his argument here. We use the opportunity to try to summarize our position on analytical egalitarianism.
As we defend it, analytical egalitarianism [AE] is a claim about model construction. There are a class of objections to AE which presume the modeler is guided only by the desire for the truth. These are variations on the theme that
“E isn’t true so AE can be rejected.”
We challenge this inference because we don’t believe the philosopher is guided only by the public good of truth. Without AE, we wonder whether truth-seeking is incentive compatible.
Obviously, we are indebted to the early work in public choice which applied the homogeneity insight to those in politics: if selfish agency is assumed in markets, then selfish behavior ought to be assumed in politics. If, conversely, selfishness is not presumed in markets, we ought not to presume it in politics.
Suppose we call this motivational homogeneity. The step we take that others haven’t (yet) taken, is to suppose that those who model economics and politics have the same structure to their preferences as the people they model. We wish to get beyond the idea that the public choice theorist is motivated by the truth, or some statistical equivalent to the truth, from which the theorist has a vantage to look at people’s behavior.
What if the modeler has the same mix of selfish and sympathetic desires as anyone else? Then, immediately, we run into incentive compatibility issues in the space of models. AE is an attempt to block a class of issues in which the incentives seem to get in the way of truth. Consider models with agents of different fixed types. Suppose a modeler proposes to pick who is in the “better” and who in the “worse” class. If the modeler can do this and policies follow from the exercise, the modeler may benefit. That’s one incentive issue. We consider rewards from both material sources and applause.
How do we get evidence about how modelers actually work? The history of economics and statistics seem reasonable data bases. One case we consider at length in The 'Vanity of the Philosopher' is how Galton and Pearson obtained their “results” that Jews were inferior. The history of statistical practice in the eugenic period is full of episodes in which one “race” was treated differently than another “race.” Impartiality might suggest that there is really only the larger group, but the partial sympathy we all feel suggests that “our group” will be viewed differently than “their group.”
Our point isn’t so much about whether “malevolent conclusions” “follow from the premises” of difference or hierarchy. Instead, our case is that once we allow for difference to creep into the analysis, the incentives are asymmetric: the theorist gains more by showing difference than similarity. And despite what Caplan thinks about the usefulness of knowing what such real giants in economics past or present as Adam Smith and J. S. Mill wrote about incentives and truth seeking, the past provides a rich set of data for us to examine their hypothesis of incentive incompatibility.
One source of approval is adherence to the standards of the community as articulated in the textbooks. So, getting closer to truth brings approbation. In our work on statistical ethics we make the case that a temptation to deviate from these standards comes from the interaction between an expert and “his client” which offers a second source of approbation.
Ordinary people fight temptation by moral constraints. Like Lionel Robbins, we see AE as the philosopher’s moral constraint, a constraint against the temptation from heterogeneous sources of approbation.
Though we are most interested at the moment in AE as a presumption of motivational homogeneity – of the theorist and subjects, the street porter and the philosopher – Caplan seems more interested in whether all people are really the same or different. We have been asked several times, first by James Buchanan and more recently by David Warsh (in a review that is quite friendly to Vanity, overall), what we would say if Plato were correct, that there are fundamental differences among humans? Setting aside the question of whether Plato’s politics are consistent with Platonism, we would look first at the history of statistical studies of human capacity to see what variation in results have been obtained. The Galton and Pearson work on the Jews is technically fascinating. Is there anyone in the Platonic camp of natural differences who defends these pieces or the general thrust of Jewish “inferiority”? If not, why not? One is unlikely to find more skilled statistical workers than Galton or Pearson.
Suppose we pre-commit to some inheritable sort of capacity – which Pearson and Galton clearly did not do – and we find that one group is less capable than another. Then what? If capacity is a vector of characteristics we need to consider the various dimensions. If there are characteristics in which groups persistently excel and characteristics on which they fall short, then it is a simple prejudice to call them less capable in some scalar sense. If the group is less capable in all dimensions then we are close to Hume’s other rational species problem which we talk about a good deal. We obviously accept Smith’s answer that our capacity for sympathy equalizes; and A. R. Wallace’s answer that “natural” selection stops at the doors of such unfortunates.