Ilya Somin’s Democracy and Political Ignorance is a welcome gust of realism in the corridors of political theory, where democracy is too often considered in the abstract, isolated from the actual traits and tendencies of democratic decision makers—that is, voters.
Ever since the 1930s, empirical research has consistently found that voters in America, and often in other countries, are far less knowledgeable about politics and policy than most people assume.1 Once we recognize the breadth and depth of public ignorance, as Somin helps us to do, we can no longer take it for granted that elections or opinion polls express deeply considered popular “mandates.” Even if they did, the basis of a mandate in sparse (or faulty) information calls its own validity into doubt. Therefore, Somin concludes, we should shift power from collective into individual hands, where people are likelier to make better decisions.
While I wish I could endorse Somin’s book and its conclusions without reservation, they rest on an illogical and empirically discredited foundation: the theory of rational ignorance. Anyone trying to base small-government conclusions on rational-ignorance premises is building a castle on sand.
Rational ignorance theory is partly grounded in the mathematical truth that in a large electorate, the odds are extremely low that a single vote will decide an election. But this fact alone cannot explain voters’ behavior unless we assume that it is a fact of which voters are aware. Only if they know that their votes are highly unlikely to matter is it possible that they’ll decide not to invest in becoming politically well informed, since they recognize that this would be a waste of their time.
There is a libertarian payoff to this theory, since at first glance it seems that voter ignorance cannot be alleviated: no matter how hard we try to get people to pay attention to politics, it would be irrational for them to do so. Because ignorance is baked into democratic government, we should reduce the power of democratic governments, other things equal.
However, the libertarian conclusion does not follow from the rational ignorance premise. Rational ignorance theory blames public ignorance on the low incentive to become a well-informed voter. Raise the incentives and you solve the problem. One way to raise the incentives would be to make government far more powerful than it now is, so that everyone had a much higher stake in electoral outcomes.Another solution would be to turn state power over to highly knowledgeable experts who would be fired or even fined if their policies didn’t work.
Fortunately, such solutions are not advisable, because the theory of rational ignorance is false. Somin provides no evidence that many voters are aware of the likely insignificance of their vote, and there is overwhelming evidence against it.
Part of the evidence is on page 74 of Somin’s book, where it emerges that 70 percent of American voters think that their individual votes “really matter.” Unfortunately, by page 74 Somin is explicating rational ignorance theory and doesn’t step back to notice the devastating implications of this datum for the theory’s key premise: that voters know that their votes aren’t likely to matter. Similarly, the Citizen Participation Study shows that 89 percent of voters say that influencing government policy is either a very important or a somewhat important reason for having voted, which would hardly make sense if they understood the utter insignificance of each vote in a sea of tens of millions of other votes.2 Finally, if you don’t trust opinion polls, there is the fact that 100 percent of voters vote. Why would they do this if they thought their votes inconsequential?3
The fact that voters vote despite the astronomical odds against their votes being decisive is what political scientists call the paradox of voting. As a leading scholar of public opinion (himself a rational choice theorist) once put it, this is “the paradox that ate rational choice theory”—including rational ignorance theory.4 Voters who know that their votes are unlikely to matter should not only fail to inform themselves politically; they should also fail to vote. The fact that voters vote, along with the polls showing that they think their votes do matter, indicates that the premise of rational ignorance theory is false.
That is hardly surprising. We grow up in a democratic culture suffused with voting, starting in kindergarten, and at every step of the way we’re bombarded by propaganda assuring us that every vote counts. People tend to believe what they’re told unless it’s challenged and challenged early. But most people have never heard the challenge posed by economists and political scientists, such as Somin, who have calculated the infinitesimal chance that a vote will “count.” I’ve never taught a political science course in which more than one or two students had even heard of the idea that one’s vote may not be decisive (something that is hardly “intuitive,” as Somin claims: it took economists until 1957 to figure it out).5 It’s not that most students have “overestimated” the odds that their vote will be decisive;6 it’s that they’ve never even asked themselves the question. The whole issue of the decisiveness of their vote is an unknown unknown for them.
Somin’s solution to the paradox of voting only deepens the problems for rational ignorance. He constructs formulae showing that if a voter (1) attributes a high enough benefit to his party’s or candidate’s election, (2) multiplies this benefit by a high enough estimate of his chance of affecting the outcome, and then (3) subtracts from this benefit the cost of voting as compared to the cost of becoming politically well informed, he will conclude that going to the polls is worthwhile but that becoming well informed (which Somin assumes will be more expensive than the cost of voting) is not. So voters’ calculations, conducted at an “intuitive” level,7 tell them to vote, but to vote ignorantly.
Let’s set aside the implausibility of this scenario given that there’s no evidence that it has even occurred to most voters that their votes may be inconsequential. There is also a problem of pure logic. If voters can plug into Somin’s formulae even a vague estimate of the benefit of their party’s or candidate’s victory, then they must think that they know enough about this benefit to be able base their vote on this knowledge. Somin and other political scientists may think that voters should know a lot more than they do, but voters seem to think, even in Somin’s account, that they know enough that they can roughly guess who to vote for. And that’s all they need to know if they are to falsify rational ignorance theory, for, according to the theory, they should be deliberately underinforming themselves. But if they did indeed deliberately underinform themselves (by their own standards), then, of necessity, they wouldn’t be able to calculate the benefits of voting, because they wouldn’t think that they could predict the benefits of a given candidate’s or party’s victory.
It seems to me that this logical truth is fatal to rational ignorance theory, which not only assumes that voters know that their votes probably won’t matter, but also assumes that, in consequence, voters deliberately fail to inform themselves adequately—not adequately according to Somin, or adequately when judged against a standard of perfect knowledge, but adequately enough to be able to vote for one candidate rather than another.
Many voters doubtless realize that they are less than “fully” informed. We all know we could spend more time reading newspapers, etc. But that would be true even if we devoted all of our time and energy to following politics: in politics, as in everything else, there is always more to learn. What matters for rational ignorance theory is whether voters deliberately choose not to inform themselves enough (by their own standards of adequacy). But if they are able to choose one candidate, party, or ballot proposition over the others, then they must think they have information adequate to this task.
My argument could be falsified if voters flipped a coin after closing the voting curtain. Then we could reasonably conclude that they knew they were too underinformed to make a rational choice between the candidates. Surely on obscure propositions and minor offices, many voters do flip a coin—or refuse to cast a vote. But rational ignorance theory cannot explain what Somin is trying to explain: hundreds of millions of nonrandom votes on important matters.
Somin anticipates this line of argument by attributing another intuition to his supposedly ignorant voters (whose gut feelings seem to more than compensate for their lack of knowledge): the intuition that “to be an adequately informed voter, it might help to know the names of the opposing candidates, the major policies adopted by the government in recent years, and which officials are responsible for which issues.”8 Somin is saying that since everyone recognizes that they need to know such things in order to cast an intelligent vote, voters (who don’t usually know these things) must have concluded that voting intelligently isn’t worthwhile. But once again, survey data suggest that Somin is mistaken. For example, in a Pew poll taken in July 2012 (a full four months before the election), 90% of the respondents said they already knew what they needed to know about Obama in order to decide how to vote, and 69% said they knew what they needed to know about Romney.9 Clearly Somin’s criteria for being adequately informed are higher than most voters’ criteria. This means that they aren’t deliberately underinforming themselves.
One can easily confirm this fact by talking to ordinary voters, who are well aware that they aren’t omniscient but who have political opinions anyway. The same goes for Somin’s wholesale adoption of the theory of “rational irrationality”10 to analogize highly informed ideologues to sports fans who, for the sheer fun of it, willfully ignore unfriendly arguments and evidence. If one actually talks to ideologues, one finds find that they firmly believe they’re in possession of the obvious truth. The reason they dismiss counterarguments and counterevidence is that they think such arguments and evidence are implausible: they contradict things that any sane person knows to be true.
In both ignorant voters and dogmatic ideologues we have a good starting point for a truly realistic theory of politics. If voters don’t think they need to know very much if they’re to cast adequately informed ballots, they must think that our society is a mighty simple place, where it doesn’t take much information to be able to identify good policies and good politicians. The same is true of ideologues, who treat their views as reflections of obvious truths about society. (I know Somin would agree with me that modern society is more complicated than that, since he brilliantly demolishes as simplistic many decisionmaking heuristics commonly used by voters.)
That society is too complex to yield easy political answers is the most important fact of which most voters seem to be ignorant. But if so, it is not because they have calculated that it would be a waste of time to learn that society is more complicated than they think it is. It is because the complexity of modern society is, to them, an unknown unknown.
Notes
1 For a survey of the early literature, see Jeffrey Friedman, Editor’s Introduction to Political Knowledge, 4 vols. (Routledge 2012). The best compendium of evidence is in Angus Campbell et al., The American Voter (1960). The most succinct and lively analysis of the findings is in Philip E. Converse, “The Nature of Belief Systems in Mass Publics” (1964), republished in Critical Review, vol. 18, nos. 1-3 (2006).
3 The Civic Participation Survey shows that 95 percent of voters rate “civic duty” as an important reason for voting (ibid.). But a duty to vote for harmful policies makes no sense. Nor does a duty to vote randomly. That leaves only a duty to cast a helpful vote, one that will serve the public interest (or one’s own interest). Anyone who felt a duty to vote must therefore feel an equal duty to make it an adequately informed vote, since only if it were adequately informed could it be expected to serve either the public’s interest or one’s own.
4 Morris P. Fiorina, “Information and Rationality in Elections,” in Information and Democratic Processes, ed. John Ferejohn and James Kuklinski (1990), p. 334.
9 Pew Research Center for the People and the Press, “Most Voters Say They Already Know Enough about the Candidates,” July 24, 2012.