I would like to start by thanking Jeffrey Friedman, Heather Gerken, and Sean Trende for their many thoughtful and penetrating contributions to our discussion. Some of the points they raise are ones I will have to address more fully in future work. Jason Kuznicki of the Cato Institute deserves great credit for his excellent work in organizing and hosting this symposium.
Most of our discussion so far has understandably focused on our differences. So I would like to start this final contribution by emphasizing three important areas of agreement among the three commentators and me.
Three Points of Agreement
First and foremost, all of us agree that voter knowledge is a relevant criterion for judging the performance of democracy. Unlike some political theorists, my critics and I do not accept the view that the voters have an intrinsic right to make political decisions on the basis of ignorance. As John Stuart Mill emphasized, voting is not a purely personal right, but rather the “exercise of power over others.” It must be done with at least some reasonable degree of knowledge and judgment.
Second, none of my critics hold out much hope that today’s generally low levels of public knowledge are likely to increase significantly in the foreseeable future. They do not contend that political ignorance can soon be overcome through improved education, technological breakthroughs, or other means. In order show that political ignorance is not a serious problem, they must therefore argue that today’s low levels of knowledge are adequate. Sean Trende and Heather Gerken have argued for this position with great skill and eloquence. But it is a difficult case to make, given the enormous size and complexity of modern government on the one hand and the enormous depth of public ignorance on the other.
Third, none of the three commentators have made much attempt to show that ballot box voters’ knowledge and judgment is comparable in quality to that of foot voters. Even if we grant the validity of many of the points made by Heather and Sean, the fact remains that most people devote far more time and effort to acquiring information about where to live or which car to buy than they do about even the most important ballot box decisions. They also judge information about the latter with greater objectivity and less bias. Even if ballot box voters perform tolerably well, foot voters likely perform substantially better. That still amounts to an important potential advantage to making more of our decisions through foot voting.
Having gone over some areas of agreement, I would like to offer some final thoughts on the questions that still divide us: the magnitude of the problem of political ignorance (on which I disagree with Sean and Heather), and the extent to which political ignorance is rational (on which I differ with Jeffrey Friedman).
The Price of Political Ignorance
Throughout our debate, Sean Trende has argued that voter ignorance is not as significant a problem as I claim because 1) voters can often rely on small bits of information to make good decisions, and 2) on many issues knowledge may not have much value because even experts disagree over them. Notice the tension between these two claims: on the one hand the political world is so simple that voters need very little knowledge. On the other, it is so complicated that no amount of knowledge is likely to help very much. I don’t doubt that there are some political issues that fall into each of these categories. But a huge range of the issues dealt with by modern government fall somewhere in between: cases where knowledge can improve the quality of decisionmaking substantially, even if it doesn’t lead to perfect solutions. At the very least, as I noted in my last response to Sean, there are many cases where greater knowledge can help us avoid serious errors, such as those that exacerbated the suffering caused by the Great Depression.
Some of Sean’s specific examples help illustrate these points. For example, he continues to argue that it does not matter that two thirds of the public in 2010 did not know that the economy was growing rather than shrinking at the time. To my point that knowledge of trendlines was important to assessing President Obama’s economic policies, he responds that much of the public recognized that the poor economy was in large part the fault of Obama’s predecessor George W. Bush. But regardless of who was to blame for the situation as of early 2009, it was important to know how the situation had changed since then. Even if patients know enough to realize that a surgeon is not at fault for the condition of a sick patient brought to the hospital, intelligent consumers still need to know whether and to what extent his condition is improving under the surgeon’s care. This is not the only relevant fact, of course. For example, the fact that the economy was improving still leaves open the possibility that faster improvement would have been possible under better policies. But it is still an important one to consider.
Sean also notes, correctly, that, in most elections, voters do in fact vote consistently with short-term economic trends – rewarding incumbents for recent improvement and punishing them for recent deterioration. This fact, however, is less than comforting. As I discuss in greater detail in my book, voters often myopically focus on only the most recent period just before the election, and they therefore often end up rewarding and punishing incumbents for short-term trends in the economy over which they had little control. In this respect, one could even argue that public ignorance about the state of the economy in 2010 was potentially fortuitous; otherwise, the Democrats might have gotten some undeserved credit for improvements they had little role in causing. Ignorance in one area helped diminish the impact of ignorance in another. But it would have been much better if most voters knew both that the economy was improving and that incumbents often have relatively little influence over such short-term trends. This would have enabled the electorate to make a better-informed assessment of Obama’s policies and the Republican alternatives, and possibly also have led them to put less emphasis on short-term economic trends in making their electoral decisions in the first place.
In discussing this and other individual examples of ignorance, it is essential that we not lose sight of the forest among the trees. Few if any individual bits of knowledge are absolutely essential to making good voting decisions. Most can potentially be offset by knowledge of other facts. The problem, however, is that most voters are not ignorant about just a few basic facts but about most such facts across the board. The cumulative impact of ignorance over a wide range of issues and a wide range of facts about the structure and organization of government is much more significant than ignorance about any one or two specific points. For example, voters’ ignorance about economic trends in 2010 might have been less important had they generally had a good grasp of economic policy and incumbents’ ability to influence it. Moreover, ignorance of some basic facts tends to be highly correlated with ignorance of others. A voter who doesn’t know whether the economy is growing or shrinking is also likely to be ignorant about other important aspects of economic policy.
Sean’s argument that some issues are so difficult that knowledge is of little value is similarly overstated. On a huge range of issues throughout American history, we have paid a heavy price for voter ignorance. Such grave policy errors as the persistence of slavery, Jim Crow segregation, persecution of gays and lesbians, protectionist trade policies such as the Smoot-Hawley Tariff, and many others were all to a significant degree caused, perpetuated, or greatly exacerbated by political ignorance. On these and other issues, more knowledgeable voters and policy experts on average realized the nature of the mistake far earlier than the average voter did.
Even on issues that are genuinely intractable given the current state of our knowledge, we have a better chance of finding solutions when decisionmaking procedures include stronger incentives to seek out relevant knowledge and analyze it objectively. In such cases, foot voters are more likely to find and effectively utilize new information than ballot box voters.
It is true, as Sean suggests, that there are different possible standards of adequacy for political knowledge levels. In my book, I consider several standards derived from different versions of democratic theory. The problem is that actual levels of voter knowledge fall below those demanded by even very forgiving theories, such as those that merely require voters to be able to punish bad performance by incumbents and reward the good. Because the public is so often ignorant of even very basic facts about politics and public policy, their knowledge levels are inadequate by almost any reasonable standard.
I am also not much moved by Sean’s argument that the supposedly more elite electorate the nineteenth century did not perform much better than ours does today. Since nearly all white male citizens got the vote by the early 1800s, it was not actually a particularly elite electorate – especially given that education levels were far lower in that era, and information was more costly. Even the education and knowledge levels of white male property owners (those allowed to vote before property qualifications were lifted) were probably not much higher than those of the average voter today. Many of the Founding Fathers – most notably James Madison – believed that the electorate of their day was often ignorant and irrational. Moreover, the primitive (by modern standards) level of social science at the time ensured that there was simply less useful knowledge available for analysis by even the most dedicated voter.
That said, I think a plausible case can be made that voters in the early republic made relatively fewer ignorance-induced errors than modern voters do, in the sense of errors that could have been avoided through better utilization of information actually available at the time; not because they were a more elite group, but because the relatively smaller size and complexity of government made it easier for them to monitor the relevant issues. At the very least, the presidents they elected – Washington, Adams, Jefferson, Madison,Monroe– compare favorably to the average president of the last fifty to sixty years. But a truly persuasive comparison of the early 1800s and the modern era requires much greater analysis than either Sean or I have been able to give it in the limited space available here.
The Rationality of Political Ignorance
Unlike Heather and Sean, Jeffrey Friedman largely agrees that voter ignorance is a serious problem. But he forcefully disputes my view that most such ignorance is rational.
In his most recent response, he emphasizes the claim that rational ignorance cannot explain the phenomenon of nonrandom voting. Voters who know they are ill-informed should not believe they know enough to cast a vote. I believe I addressed this in my previous responses. But perhaps I wasn’t sufficiently clear in stating my position, a mistake I will try to remedy now.
Nonrandom voting is entirely consistent with rational ignorance because people apply different standards of informational adequacy to decisions that have only a small chance of making a difference than to ones where that likelihood is higher. In many aspects of our lives, we routinely make perfectly rational and nonrandom minor decisions on the basis of small bits of information. Very often, we know that we might make better decisions if we sought out more information or analyzed it more carefully. But we don’t do so because the decision isn’t important enough to make it worthwhile to spend the additional time and effort. When I need to get gas for my car, I rationally devote only a small amount of time to searching for the cheapest and most efficient gas station. But that doesn’t mean I choose gas stations completely randomly, or that I am unaware that a more thorough search might turn up a better or cheaper option. I noted in an earlier post my rational conclusion that The Hunger Games was the best movie of 2012, despite recognizing that greater knowledge might lead me to reach a different and better-informed conclusion. Rationally ignorant voting is structurally similar to rationally ignorant judgments about movies and gas stations. The difference is that individually rational behavior in the former case causes vastly more social harm than in the latter.
Jeff’s view implies that voters would not bother to acquire significantly more information if, for example, they could personally select the next president or the next governor of their state. This is both psychologically implausible and at odds with lots of empirical evidence. For example, jurors work harder at acquiring and analyzing relevant information than voters do, in part because they make decisions in small groups where each individual vote is likely to matter greatly.
I agree with Jeff that one of the reasons why foot voters have knowledge advantages over ballot box voters is that they learn from mistakes, both their own and those of others. But ballot box voters, of course, can also potentially learn from mistakes, including those made by other nations’ governments. On average, however, foot voters learn from error more effectively than ballot box voters do because they have much better incentives to gather information about such failures and analyze them rationally. That is one reason why ballot box voters have yet to learn the lesson of centuries of failed protectionist policies and why it took decades for them to realize that policies such as racial segregation and oppression of gays and lesbians caused more harm than good.
The fact that political ignorance is rational does not by itself prove that limiting and decentralizing government is the right approach to the problem. Other solutions are also possible, including Jeff’s suggestion that rational ignorance could be overcome by properly incentivized technocrats. But I am skeptical that the problem of technocratic incentives is as easily overcome as Jeff suggests. In addition, technocrats have information problems of their own, albeit somewhat different ones from those that bedevil voters. Be that as it may, in evaluating competing solutions, it is important to understand the nature of the problem.
I do not imagine that either my book or this symposium comes close to fully resolving the longstanding question of what we should do about the problem of political ignorance. But I hope we have at least advanced the debate over these issues, and led more people to seriously consider the possibility that the harm caused by ignorance could be alleviated by limiting and decentralizing government.
 For my critique of claims that knowledge can be greatly increased by these or other methods, see Ilya Somin, Democracy and Political Ignorance: Why Smaller Government is Smarter (Stanford: Stanford University Press, 2013), ch. 7.
 Ibid., pp. 100-102.
 For a discussion of this kind of fortuitously beneficial ignorance, see ibid., 59-60.
 Ibid., Chapter 2.
 For example, Madison believed that most of the voters who cast ballots selecting delegates to the Virginia ratifying convention for the Constitution did not understand the issues at stake. See Ilya Somin, “Originalism and Political Ignorance,” Minnesota Law Review 97 (2012): 625-68, 643-44.
 Choosing gas stations is a foot-voting decision rather than ballot box voting. But it is still an example of rational ignorance because the benefits of seeking additional knowledge are low relative to the costs. In this case, they are relatively low because the decision itself is unimportant. In the case of ballot box voting, the benefits of knowledge acquisition are low relative to costs because the chance of influencing the outcome is extremely low (even though the outcome itself may well be very important).
 I discuss the technocratic alternative in greater detail in the book. See Somin, Democracy and Political Ignorance, 183-85.