I think this would be a good time to bring my participation in this discussion to an end by summarizing my take on the debate over rational ignorance theory in three points.
1. No answer has emerged to the “paradox of nonrandom voting.”
To its great credit, Ilya’s book, unlike other rational ignorance treatises, squarely confronts the “paradox of voting”: If voters deliberately underinform themselves because they know their individual votes are unlikely to matter, why do they bother to vote? However, Ilya’s answer is that while voters intuit that the odds against their vote counting are small, they also recognize that the chance is greater than zero, and they roughly calculate that the great benefit that would flow from casting the decisive vote for their preferred party or candidate justifies the low cost of voting (but not the higher cost of becoming politically well informed).
Yet as I said in my first post, Ilya’s solution to the paradox of voting produces a fatal problem for rational ignorance theory. If voters think that the victory of their preferred party or candidate will produce a great (or even a small) benefit, they must think that they are well-enough informed about politics to be able to make this judgment. Otherwise, they could not be confident that in the unlikely event that they ended up casting the deciding vote for their party or candidate, great harm instead of great benefit would not result.
The general form taken by this problem is the paradox of nonrandom voting. Any voter who thinks he can justify (to himself) voting for a specific candidate or party, rather than voting randomly, must think he is well-enough informed to prefer that candidate or party to their competitors. Therefore, it is logically impossible for such a voter to have deliberately underinformed himself—according to his standards of adequate information.
2. In light of the paradox of nonrandom voting, rational ignorance theorists’ complaints about voter ignorance simply display their disagreement with voters about how much information voters need.
As I said in my second post, it is striking that Ilya must constantly impute to allegedly ignorant voters knowledge—not just knowledge that their vote probably won’t matter, but knowledge that they are underinformed. His latest post continues this practice. He attributes to voters the knowledge that it is “blatantly obvious” that “the facts about whether the economy is shrinking or growing are highly relevant” to a vote cast on the basis of economic factors. But Ilya is failing to take ignorance seriously. A voter may very well think he knows that the economy is “shrinking” because he has heard reports about unemployment and low labor-force participation. A voter may not even know that there is a difference between “economic growth” and employment rates. Or a voter may disagree with Ilya that “economic growth” per se is important compared to high rates of unemployment. Suppose that Ilya is right, and economic growth rates are tremendously important. Then this voter is ignorant of an important truth. But we know from the paradox of nonrandom voting that such ignorance cannot be deliberately chosen, but must instead be the product of the voter’s belief that the economy is (as far as he knows) in the tank.
One explanation for the failure of rational ignorance theory is that neoclassical economics (of which rational ignorance theory is an offshoot) cannot deal with genuine disagreement. In neoclassical economics there is just “knowledge,” and it is assumed that everyone knows not just that this stuff is available but that it is important. In reality, the importance of a given “datum” is never self-evident, but depends on an interpretation of what is important, e.g., an interpretation based on a tacit or explicit theory of economics.
What it boils down to is that Ilya’s complaints about voter ignorance—however justified they may be in a given case—are disagreements with voters who think they know what they need to know. But Ilya doesn’t accept that these are disagreements. He insists that voters agree with him about what they need to know because it is “glaringly obvious” (to him!) what they should know. So it becomes puzzling that they don’t go out and get this stuff that they must know they should have—after all, this stuff is free for the taking.
I think this whole line of thought displays a failure of sympathetic imagination. In this respect rational ignorance theory is much like “rational irrationality” theory, which basically says that since the policy conclusions that economists think are glaringly obvious are also (by sheer assumption) glaringly obvious to the voters, the voters’ disagreement with economists’ conclusions is evidence of voters’ deliberate choice to be “irrational.” There’s a certain lack of perspective here. Doesn’t it occur to the rational-irrationality theorist that what seem to him to be glaringly obvious truths only seem that way to him because he has been studying economics all his life? Similarly, only a policy wonk of long standing would think that the “large bodies of political information” that are “out there” are self-evidently relevant to casting an informed vote.
I must say that as a political scientist who’s thoroughly familiar with these “large bodies of political information,” I disagree with Ilya about their relevance to voting. But this is a topic for another discussion, and Sean Trende has said much that I would say in such a discussion.
3. Rational ignorance theory is a barrier to thinking about the relative merits of markets and politics.
I think Ilya is right to point out the different behavior of consumers researching a major purchase and voters not researching their vote. But because of the paradox of non-random voting, we cannot reach for the easy explanation for this difference: the relatively high incentives for consumers to do research. In any case, the need to do consumer research is not intuitive knowledge (rational ignorance theory justifies all the knowledge it attributes to people by calling it intuition). This knowledge is based on experience and feedback from past mistakes. Not everyone does research, and they often live to regret it. Sometimes they learn from the experience that even though they thought they knew enough, they didn’t, so they should be more careful next time. Sometimes people learn this through other people’s bad experiences, or their good ones. (“How did you know Subarus are so reliable?” “I read Consumer Reports.”)
Now a crucial question is whether political decisions produce similar learning opportunities through negative feedback. After all, bad policies fail, this produces bad results—so people should (in principle) be able to learn from their mistakes in this sphere as in the private sphere. Indeed, voters think it’s easy to see whether a policy is needed and whether it is a success.
But perhaps we disagree with voters about this, too. If policy failure and success were so transparent, then why do policy professionals so often disagree about whether a policy failed or, if so, why? And—crucially—once we have an account of exactly what it is that might make it harder to get accurate perceptions of policy failures than consumer product breakdowns, might this reason for the opacity of policy effects hinder policy professionals’ ability to perceive policy effects clearly?
In that case, maybe technocracy suffers a knowledge deficit, not just the relatively easy-to-fix problem of the “incentives” of the technocrats. (Again to his credit, Somin allows that voters are not self-interested. Why not extend the same charity to technocrats, or at least provide evidence—not theories about incentives—showing that real-world technocrats are corrupt?)
Answering such questions requires hard thinking about the philosophy of social science and the nature of modern society. That is a research agenda made possible by understanding voter ignorance as the result of voters’ simplistic view of social and economic problems. If we think modern society is more complex than voters assume, we have to explain precisely what makes it complex and precisely what makes human beings ill equipped to penetrate the complexity.
But if we think that “the facts” about modern society announce their importance to human beings intuitively, so that it’s “glaringly obvious” to voters what they should know and, once they know it, it’s glaringly obvious to them what conclusions they should reach, then we cannot possibly undertake that research agenda. Why? Because we have bought into the very same simplistic worldview that explains voters’ conviction that their low levels of knowledge are adequate for making potentially consequential voting decisions.