NDPR Discussion Forum: Jason Brennan’s Against Democracy

15 Replies to “NDPR Discussion Forum: Jason Brennan’s Against Democracy”

1. Thanks for the opportunity to discuss this further.

   Christiano leads mostly with an empirical critique, but that’s frankly not a promising route for him to take, and I don’t find any of the purported empirical criticisms troubling. Indeed, I think many of these criticisms were already answered in the book, while the rest are answered in the existing political science literature (much of which I cite). I’d recommend Christiano read Achen and Bartels’s book, to start. Here I’ll comments a bit on the empirical stuff. I’ll discuss the philosophical stuff tomorrow.

   1. There is a massive disconnect between empirical work on how voters, politicians, and democracy as an institution actually behave, versus what philosophical democratic theorists say about democracy and why it is justified. I’m not the first to say this–see, e.g., Diana Mutz’s *Hearing the Other Side* or Achen and Bartels’s *Democracy for Realists*, to name two of my favorite recent works on this issue. In general, philosophical democratic theorists provide normative arguments which show that democracy would be the best system, if only it worked differently from how it actually works. I worry that most philosophical defenses of democracy are not so much wrong but irrelevant.

   Achen and Bartels paint a similar picture to mine. Most voters are systematically misinformed. A small percent has a stable ideology, but this ideology is not generally based on careful consideration of ideas and evidence. The overwhelming majority of voters with stable party preferences vote that way not because of ideology per se or because of the information they posses, but because they build party loyalty into their identities. Aches and Bartels have a powerful critique of the shortcuts literature. They show the literature is often unrealistic, extrapolates from a few choice examples, and for the most part defends democracy by having implausibly low standards for what counts as good voting via shortcuts. (See also Ilya Somin’s book *Democracy and Political Ignorance*, which has a thorough critique of the shortcuts and heuristics literature, plus Somin’s response to Christiano here: http://www.tandfonline.com/toc/rcri20/27/3-4?nav=tocList ). Further, as Bartels, Caplan, Gilens, Althaus, and others have pointed out, we have extremely good evidence that people using heuristics and shortcuts do not behave as if they were high information voters.

   To illustrate: Parties reduce the demands on voters by reducing the number of ideas they have to consider. However, what parties push is also largely a function of what they think voters will support. If–as is in fact the case–voters are largely misinformed hooligans who care more about flag waving, mood affiliation, and expressive concerns than they do about facts and policies–then parties will push different agendas than they would if voters were, well, smart. But that point aside, as I show in chapter two, a very large percentage of voters lack accurate stereotypes of what parties tend to do. So, in short: parties dumb down the electoral choice, but dumb voters dumb down what the parties try to do. Yet, even after al this dumbing down, a good forty percent of voters (and even larger percentage of the electorate) still can’t identify what the parties want to do, which party controls Congress, etc.

   Or to illustrate again: Retrospective voting is extremely difficult. For voters to do it well, they’d have to know who was in power, what they had power to do, what they did, and what effect what they did had. The overwhelming majority of voters lack that information. They don’t know general social or economic trends. They’re even bad about estimating what happened to themselves. As Achen and Bartels joke, a lot of retrospective voting looks like people coming home and kicking their dogs because they’re angry there was traffic on the way home.

   2. Imagine I have a magic wand. When I wave it, it will make all voters even more systematically misinformed about politics. They will, henceforth, always support candidates like Trump or Erdogan. Would it be bad for me to wave the magic wand? If you think so, then you probably also accept a “micro-theory” in which voters’ behavior can make a positive or negative difference. Christiano accepts such a micro-theory; see his books.

   3. Imagine I have a different magic want. When I wave it, it will make all voters fully informed and make them believe the correct theory of justice, whatever that is. Would it be good to wave the wand? If you think so, then you probably also accept a “micro-theory” in which voters can make a positive or negative difference. Christiano accepts such a micro-theory; see his books.

   4. What is this micro-theory? Well, lots to be said here. On the Spatial Model of electoral outcomes (e.g., look up the Median Voter Theorem), voters fall on a variety of spectra in ideological space, but politicians recognize they can win votes by pushing an agenda in the middle of these spectra. Now, everyone agrees this model is too simple and in some ways wrong. On one hand, voters often don’t have a stable ideology and don’t really occupy the space. On the other hand, often preferences are multi-modal rather than being centered around one point, and so there isn’t a “middle” per se. However, on yet another hand, voters do tend to be loyal to parties which fall on a left-right spectrum and which cluster beliefs together, and so the median voter model has more power than we might expect if voters were voting purely on ideology. However, as Page and Gilens have shown, in fact politicians are more responsive to relatively rich and relatively informed voters than to the median voter. Again, I’d recommend reading Achen and Bartels, especially chapter two, for a thorough overview.

   4. Against Democracy relies at most on a weak empirical claim: The quality of candidates on the ballot, and the policies and agenda they push, depends in a significant on the quality of the electorate. While there are debates about just how politicians respond to voters, I don’t know of anyone who denies that.

   That said, I lied. In fact, my argument doesn’t even require *this* much! After all, in the early chapters, I rebut a wide range of philosophical arguments which purport to show that democracy is intrinsically just. This leaves us with instrumentalism: we should pick democracy over other forms of government if and only if it turns out to do a better job delivering good outcomes, as determined by the correct normative theory, whatever that is.

   In chapter seven, I discuss a wide range of reasons why democracies seem to over-perform, in light of democratic ignorance. It’s not because wisdom is an emergent feature of group decision-making. It’s not because voters use helpful heuristics. It’s not because deliberation saves the day. Rather, there are a host of mediating factors: Leaders and bureaucracies have significant independence from voters, and sometimes that independence means they choose better policies than voters support. Politicians are also more responsive to informed voters than to uninformed voters, and so we get better outcomes than the Median Voter Theorem would predict. Etc.

   But one might think, contrary to points 1 and 2, that these factors mean that elections don’t really matter. Christiano seems to hint at that in his worries about my “micro-theory”. However, in response to that, I’ll quote from chapter seven:

   “Given that so many factors mediate between what voters prefer during an election and what governments actually do, one might then wonder whether the Competence Principle even applies to electoral decisions. I’ll answer this objection in the abstract, in the form of a dilemma. The Competence Principle applies only to high-stakes decision, decisions that can tend to cause significant harm to people, or to deprive them of life, liberty, or property. It doesn’t apply to low stakes decisions, such as what the national anthem or flag colors will be. Now ask, in light of all these mediating factors, that is, do electoral decisions count as high stakes or not? There are two possibilities:

   “1. Most elections are still high stakes. On this view, even though many factors mediate between what voters want and what governments do, voters still have enough power (in most elections) such that their decisions count as high stakes. If so, then, in light of empirical evidence on voter and electoral behavior we’ve examined over the course of this book, we should conclude that most electorates violate the Competence Principle. If so, then if it turns out that certain forms of epistocratic decision-making methods would perform better, then we ought to replace democracy with epistocracy. We should use epistocratic instead of democratic elections. (This assumes, of course, that the benefits of epistocracy exceed the costs of transitioning from democracy to epistocracy.)

   “2. Most electoral decisions are not high stakes. On this view, the various mediating factors are so significant that we can’t meaningfully call elections high stakes. If not, then the Competence Principle doesn’t apply to them, and it doesn’t matter from a moral point of view that electorates tend to make such decisions incompetently.

   “So, which is it, possibility 1 or 2? Do elections matter, or not? This is a big question. In a sense, thousands of political scientists have devoted their careers to trying to answer this question. I don’t want to do a hundred-page review of the empirical literature on all the various “mediating factors” here. I read this literature as showing that most major elections remain high stakes, if not as high stakes as a naïve fifth-grader might think. Elections of officeholders do not directly decide policy, but they significantly change the probability that different policies will be implemented. If I’m right, then we have presumptive grounds to view democratic elections with universal equal suffrage as unjust, even though this doesn’t mean every decision every democratic government agent makes is therefore unjust.

   “However, suppose I’m wrong. Suppose possibility 2 turns out to be correct. Suppose elections don’t really matter. Suppose that the post-electoral mediating factors are so significant that the typical parliamentary, congressional, or presidential election doesn’t qualify as “high stakes”. If so, then the Competence Principle doesn’t apply to these elections, and the facts about voter behavior we discussed in chapters two and three don’t give us reason to prefer epistocracy to democracy.

   “But if possibility 2 is correct—if elections don’t really matter—this should be little solace to most democrats. After all, think of their major reasons for preferring democracy to epistocracy. Most of their arguments rely in some way upon the view that elections matter, that elections empower the groups of voters, that elections with universal suffrage are necessary to make sure that governments respond appropriately to citizens’ interests, and so on. But if possibility 2 is correct, they’ll have a hard time making such arguments. Possibility 2 says that it’s only the post-electoral stuff that matters. If so, then it’s unclear why a democrat would prefer democratic elections (with universal equal suffrage) to epistocracies (which in some way have unequal suffrage). After all, the proceduralist arguments for democracy (which we reviewed in chapters four and five) don’t succeed. Possibility 2 implies that there’s no reason to prefer epistocratic to democratic elections or vice versa.

   “In short, if possibility 1 is correct, then the Competence Principle gives us presumptive grounds to favor epistocracy over democracy. If possibility 2 is correct, then the choice between democracy and epistocracy is something of a toss-up—in effect, it just doesn’t matter which one we pick. Either way, my argument so far in this book puts democrats in an uncomfortable position. At this point, they should either presumptively favor epistocracy over democracy, or they should be indifferent. At this point, when I say, “Let’s try epistocracy!” you should either be with me, or at least not against me, depending on whether you think the facts support possibility 1 or 2. With that, let’s take a look at some possible forms of epistocracy.”

2. Hi Jason,

   I know little about the empirical literature here but when I poked around and discussed your book in an interdisciplinary reading group I found people suggesting that the Achen and Bartels book is narrow in that it focuses only on the US – and that we should be hesistant to generalize from it (The Ilya Somin book looks like it is also focused on the US).

   Several people suggested there are broader studies with less pessimistic results and that might help us identify empirically what will lead to better results.

   For an example of the kind of response I have in mind let me quote from Neil Malhotra’s review of the the Achen and Bartels book: (http://www.journals.uchicago.edu/doi/full/10.1086/688172)

   “Another limitation of the book is that it is primarily US focused. This is in contrast to Raymond Duch and Randy Stevenson’s The Economic Vote (2008), in which they present empirical evidence from many countries showing that voters behave in a quite sophisticated manner, even anticipating the coalitional structures in parliaments. Why do books focused mainly on the United States and Europe, respectively, reach such different conclusions about voter competence and the model of retrospective voting? Perhaps institutions matter. In countries where the electoral rules and the organization of parties are more complex, maybe voters need to become more sophisticated in order to effectively participate in politics.

   Here is another review that provokes the same worry with more information about the Duch and Stevenson book and the positive view is purports to support regarding voters. It also pushes the need to look at broader data sets from different countries. http://www.journals.uchicago.edu/doi/abs/10.1017/S0022381609090616

   Can you point to pessimistic empirical arguments with broad data sets that include Europe and other places?

   Or are you unconcerned with these arguments for some other reason (like I say I do not know the in and outs of the empirical work at all!)?

3. In my paper “Democracy and Freedom,” I note that democracy is positively correlated with A) economic liberty, B) civil liberty (defined without including voting rights, etc.) and C) GDP/capita. Just why this is so is disputed in the literature. Some think that democracy and A, B, and C mutually reinforce each other. Others think the causal arrow runs one way or the other. Others think that society’s ideological commitment to liberalism causes both democracy and A, B, and C. Take a look at David Weil’s textbook on economic growth or Acemoglu and Robinson’s Why Nations Fail for some mainstream views on how institutions cause growth and prosperity.

   The standard economic view now is that the following institutions are important for long-term economic prosperity: 1) stable and honest government, 2) secure property rights without major threat of appropriation, 3) stable money, 4) open markets. Democracies are good insofar as they promote 1-4, though they often do not.

   See also Fernando Téson’s post here. I suspect that because it’s on Bleeding Heart Libertarians, people will assume this is a libertarian view, but it’s really standard textbook institutional economics: http://bleedingheartlibertarians.com/2017/05/democracy-and-prosperity/

4. Again, I don’t find any of Christiano’s empirical-based criticisms troubling; it seems he just isn’t versed in relevant empirical literature, and that he strawmans my position in order to argue against it. (The PRC is an epistocracy?) So, let’s turn the philosophical critiques, which perhaps are of some value.

   In chapter 5, “Politics Is Not a Poem,” I systematically critique “semiotic” arguments for democracy and semiotic critiques of epistocracy. Semiotic arguments for democracy hold that part of what makes democracy good or just is what democracy expresses, while part of what makes epistocracy bad or unjust is what it expresses.

   Many people think democracy is essential to express the idea that we are all equal in some way. But, I note, that seems to be a contingent social construct; it’s not written into the fabric of the universe that equal suffrage has that meaning. We could express a commitment to equally in all sorts of others way, for instance, by handing everyone a red scarf at age 18, or by implementing whatever institutions actually promote the most just outcomes.

   I argue in AD, as I also argued in “Markets without Symbolic Limits,” that we should judge semiotic codes by cost-benefit analysis. For instance, the Fore tribe of Papa New Guinea practice endocannibalism as a means of expressing respect for their dead. It really was a socially constructed fact that eating the brains of their dead expressed respect. But when they discovered this practice spread a deadly disease, they changed the practice.

   Similarly, suppose we discover the practice of expressing respect through equal voting shares causes bad outcomes, compared to, say, government by simulated oracle (my preferred form of epistocracy; see below for a description) or compared to futurarchy. In that case, we’d have similar reasons to revise our semiotic practices.

   In the end, then, my point is that semiotic arguments for democracy do not stand alone and do not do the work people need them to do. *If* democracy is the best functioning system, the system that produces the most substantively good and just outcomes, then it’s fine to impute all sorts of symbolic value to democratic procedures. But if democracy systematically produces unjust outcomes, compared to other forms of government, then we should stop imputing symbolic value to democratic outcomes. The point is that semiotic arguments do not defend democracy from instrumentalist critiques.

   Tom seems to agree. He says:

   “By the way, the symbolic value of democracy as expressing the equal status of persons also depends on this instrumental value. Here Brennan stumbles; he seems to think that there are people who think that democracy can have value merely by having laws that assert that people are equal regardless of the effects on people’s lives (p. 128). But the arguments of Rawls and myself assume that the expressive value piggy-backs on instrumental value. The idea is that if having political power enables people to advance their interests, then depriving them of that political power expresses the idea that their interests are of little or no consequence. Now, one might ask: what do the intrinsic value and the expressive values add? They add something because there is a great deal of indeterminacy in determining how much people’s legitimate interests are being advanced, even though it is clear that political power does advance interests. The egalitarian intrinsic value presupposes the instrumental value but cannot be entirely replaced by it. The reason for the indeterminacy is another feature that Brennan’s discussion gives far too little weight to: the fact that there is a great deal of disagreement about what is a proper way to treat people as free and equal in the substance of policy. As a consequence there is not enough society-wide agreement to determine when people are being treated as equals or not. The way to resolve the society-wide disagreement is by giving people an equal amount of political power, which is known to help people advance their interests. It is no objection to this view to say, as Brennan does, that people are actually concerned with the common good and not with advancing their interests. We can all agree on this and that people have duties to advance the common good, but we can still recognize the ubiquitous facts of persons’ biases towards conceptions of the common good that are connected with their own interests and distinctive experiences in society. This is why, while everyone has a duty to advance the common good, they also have an interest in doing this in a context in which they have equal power. And it is why a system that fails to accord equal power is publicly treating some groups as inferiors.”

   Tom makes a number of empirical claims here which, again, are questionable. Actually, it’s not clear that the reason democracies respond to the poor is that the poor participate or have much power. That’s a surprisingly contentious view in the empirical literature. After all, forcing the poor to vote in high numbers (as in compulsory voting regimes) doesn’t lead to more substantively left-wing positions, more redistribution, etc. There’s also frequently a clash between what policies the poor want and what policies might actually be good for them. And, further, in general, most poor voters do not have a substantive ideology or set of policies they want to promote, and among those that do, there’s a lot of disagreement about what is good. Tom claims that people might aim for the good but be biased toward their own self-interest. But the self-interested voter hypothesis has been falsified over and over; people do not generally support policies which are objectively or independently assessed to be in their self-interest. They are instead nationalist sociotropes. Tom claims there might be bias, but if there is, the forty or so papers I cited on voter motivation didn’t find it.

   Some people subscribe to what I call the Naive Theory of Democracy. According to the Naive Theory, if you and people like you vote, then the government will promote your objective interests, regardless of how you vote. You can vote to shoot yourselves in the face, but the government will respond instead by helping you. Tom hints at the Naive Theory above, but there is little support for it in the literature.

   I’ll end with this passage from AD, which critiques the point about whether democracy is a way of providing evidence to voters that their interests are being promoted. I’m surprised Tom didn’t respond to this.

   “Christiano says that justice must not only be done, but must also be seen to be done. If fundamental political power is distributed equally, then citizens will tend to believe that everyone’s interests are being promoted equally. If power is distributed unequally, then citizens will tend to believe (or be suspicious) that the government favors some over others. If some citizens are granted the right to vote but others are not, people might be suspicious that the former’s interests are being promoted while the latter’s interests are not.

   “But such suspicions are not enough to ground a theory of justice in the distribution of political power. One problem for Christiano is that “to see” is a success verb. One cannot see a ghost in the shadows unless there actually is a ghost. One cannot “see” justice being done unless justice is actually being done. So, suppose it turned out—as it well may—that epistocracy is superior to democracy at promoting just outcomes. If so, then instantiating democracy over epistocracy would not cause citizens to see justice done—it would at best cause them to mistakenly believe they are seeing justice done. Christiano’s objection gets off the ground only if citizens’ suspicions of epistocracy are well-grounded, that is, only if democracy actually performs better than epistocracy in promoting all citizens’ interests equitably. In that case, the semiotic concerns would no longer be decisive. Instead, we should have democracy simply because it works better. But if epistocracy would perform better than democracy, then in order for citizens to see justice being done, they would need to see epistocracy.

   “Now, suppose Christiano modified his view to say that it’s important that people believe justice is being done, and in some cases, it might be more important that people falsely believe justice is being done than that justice actually be done. For instance, suppose it turned out that people are terribly stubborn. Even if we had overwhelming proof that epistocracy produces more substantive justice than democracy does, they would still regard epistocracy as unjust, and, as a result, epistocracy would be more unstable than democracy. If the instability is bad enough, perhaps that would be outweigh whatever substantive benefits epistocracy would bring, and would be reason to favor democracy over epistocracy. But notice here that we’ve moved away from a semiotic argument for democracy to the instrumentalist question of which system performs better, all things considered.”

5. Brad, good questions. Aches and Bartels focus on the US in part because we have far better data and empirical work on American politics than anywhere else. They think most of what they say applies elsewhere, but some of it does not.

   Off the top of my head, here are some thoughts:

   1. Levels of voter knowledge are roughly the same around the world. The Swiss may know more than the French, but overall, in most large democracies, average levels are low.

   2. Party tribalism is much stronger in the US or the UK than in certain proportional voting parliamentary systems. Part of the reason for that is that First-Past-the-Fence voting tends to lead to there being only a small number of viable parties, while proportional voting systems make it possible for there to be many small parties (which have to form coalitions with one another). People in, say, Sweden have a larger range of viable choices, so they may shop around more than in the US or UK.

   3. Switzerland is unique in many ways. They have a better educated and better informed voting populace than normal. They keep many issues local, where the epistemic demands are much simpler, and where people have a stronger incentive to figure out the truth and do what works. For national-level referenda and elections, they have very low turn out, and the people that turn out tend to be the epistemic elite.

   4. The worries about heuristics and shortcuts, about sociotropic voting, and so on, are more or less the same in every major democratic country.

6. Hi Jason. Thanks for this. As you might expect, I’m sympathetic to Christiano’s view and skeptical of your thesis. However, I do not write in this area, so my views are unsettled. Nonetheless, I’ll hazard the following defense of what you describe as the Naïve View. What I say here might be in the spirit of Christiano’s critical comments and theory of democracy, but I certainly don’t offer any of this as a defense or restatement of his view.

   As you formulate it, the Naïve View of Democracy is as follows:

   “[I]f you and people like you vote, then the government will promote your objective interests, regardless of how you vote.” 

   Here’s the Less Naïve View:

   Because democratic governments extend the franchise to the demos, they do a better job promoting justice and the objective interests of the demos than extant forms of non-democratic government.

   Christiano’s review alludes to elements of the mechanism that would support the less naïve view. In short, if each member of the demos has more or less an equal formal say, then politicians and political parties will have an incentive to package their policy agendas in terms of the interests of each member of the demos. Moreover, politicians and political parties will have an incentive to explain why their rival parties’ policy agenda does not serve these interests. In this same vein, other organs of civil society, interest groups, think-tanks, unions, and so on, will have an incentive both to form and to engage in this debate on these terms. By contrast, in societies in which the franchise is severely circumscribed, the incentive to package and critique policies in the terms of the interests of the demos will be removed (or at least much reduced) and deliberative civil society will be less robust.

   The key claim that I don’t think you have engaged in your comments is the following: In a sense, voting doesn’t matte on the less naïve view, because democratic forms of governance secure policy options at the policy agenda setting stage that are systematically better than the policies that non-democratic governments implement. According to this less naïve view, there is little incentive in extant non-democracies to package policies in terms of the interest of the demos and there is a relatively undeveloped deliberative civil society. As a result, the policies that these non-democracies implement are systematically less conducive to the public interest than the policy options produced by political parties in democracies prior to the act of voting.

   To be clear, the less naïve view does not address the relative instrumental value of epistocracy. However, it does bring some of the promise and costs of epistocracy into view.

   To be sure, on the naive view, the competence of the electorate matters, for we would expect an even closer alignment between the political Overton window and justice/the public good if the democratic electorate were better informed and hence less ripe for manipulation. Epistocracy promises to deliver a better informed electorate, and this might have positive effects on the political Overton window and hence the resulting policies. However, epistocracy also removes a key incentive to package and frame policy discussions in terms of the interests of those excluded from the epistocracy. Moreover, the more demanding the rule for inclusion in the epistocracy, the larger the number of persons whose interests can be safely ignored by the epistocracy’s policy agenda setting mechanisms. In short, epistocracy promises to secure better selection of policies within a given political Overton window; however, it might very well do a poorer job than democracy in aligning the political Overton window with the public good.

   To be clear, I don’t think these comments settle the issue. But hopefully they bring to the fore a more powerful argument in defense of democracy vis-à-vis epistocracy than you have considered.

   Also, one final point. I assume that democrats and epistocrats would agree that the ideal political system would comprise a demos of epistocrats, for such a demos would be a better guarantor of justice and the public interest than an oligarchy of epistocrats. And, I assume that all other things being equal, the closer a system approximates this ideal the better. It seems then that an important question is as follows: What form of political organization holds greater promise for achieving or significantly moving extant political systems closer to the ideal demos of epistocrats: one that codifies into law the distinction between those with and without the franchise or an initially democratic system in which we have a shared interest in educating one another?

7. Just a minor point, Jason. How can you characterize Teson’s argument as “textbook institutional economics?” You don’t mean the original Institutionalists, who would, in fact, tend to associate popular economic advancement with democratic constraints on capitalism.

   Of course, Coase represents an assimilation of institutional concerns by neoclassical theory, but Coast is not an Institutionalist in the original sense. So what did you mean by “institutional?”

   See https://en.m.wikipedia.org/wiki/Institutional_economics

   Have you engaged with the institutionalist tradition in any of your related work?

   Thanks for your response!

8. I don’t want to take up a lot of space here because it is important that other people have a chance to say something.

   Let me start by saying that the discussion above does not give me the sense that Brennan has grasped the fundamental problem with his approach or the magnitude of the problem. The same holds, I should add, for works such as Ilya Somin’s or even the Achen and Bartels’ work. The whole picture that Brennan draws starts from a particular micro theory of the voter. What I mean by micro theory is something along the lines of the theory of consumer choice, labor supply or the behavior of firms. What is distinctive about the micro theory of democracy, and here I think Brennan provides a valuable reminder, is that the economics of information plays a large role here. To be sure, the economics of information is becoming more prominent in microeconomics. But the basic texts are pretty weak on low information rationality, which is surprising since this is as ubiquitous an issue in economic life as it is in politics. The classic and still the best work on micro theory is Anthony Downs’s An Economic Theory of Democracy. And I mean the whole book, not just the median voter theorem materials but also Downs’s extensive discussion of the economics of information of the voters.

   Let’s start with the basics. What is the voter’s utility function? How do we characterize their rankings of states of affairs? This is a surprisingly neglected question or it is at least never deeply explored. Downs, for instance, seems mainly to punt here. He just says that voters have rankings of social states with the highest argument being what he calls the ideal social state. He seems to recognize that the narrow self-interest approach won’t do here since he sees that people vote. But he does think that people want to economize on information gathering costs. Somin simply produces a complicated utility function that has the precise implication that people will vote but they will not become informed. This seems to be a leap of faith on his part because no serious argument is given for this and other things that he says about people’s propensity to altruism seem to contradict this. Brennan says that people are interested in promoting the common good. But this answer is extremely vague since what we need to know is how much they are interested in promoting the common good and how this concern trades off with other concerns they have. Furthermore, Brennan seems to abandon this perspective when he describes actual voters who he says are either hobbits or hooligans. Certainly hooligans are not commonly associated with a concern for the common good. Admittedly these are ideal types but the theory should work with persons who fit exclusively the ideal types. This part of the underlying micro theory is essential to the whole picture of democracy. Without an adequate account we either have arbitrarily selected stereotypes of voters or we have a model that predicts nothing at all.

   Here, just as an aside, I was alarmed in the book and the response to my review by the extraordinary misinterpretation of my admittedly limited micro theory. I claim that people are interested in promoting the common good but that their conceptions of the common good are biased towards their own interests and experiences. This is a perfectly commonplace observation about people’s biases when they are trying to understand something difficult. Brennan takes me to task for suggesting that people are motivated by their self-interest alone, a view that has been conclusively refuted. I fail to see how one can get the self-interest view out of my view. It is a statement of bias not of secret machinations and it no way suggests the self-interest view. People do have a concern for the common good it is just that they are limited in their abilities to grasp it.

   The second part of the micro theory has to do with the economics of information. Given the voter’s utility function and given the social context in which they find themselves, we need to know how voters come to the kind of information they have in order to vote. We certainly have reason to think that people economize on information gathering costs, as Downs says. This is a ubiquitous feature of human life. People simply don’t know a lot about the political world they live in or the economic activities they engage in. As I write these words, it is a complete mystery to me how the words appear on the page and how they are then disseminated to lots of people. Fortunately, they are. The institutions that do this are mostly trustworthy. What we all rely on all the time is a very complex division of labor which includes a very complex cognitive division of labor. Everything we do depends on information held by other people. And it is a good thing that this is the case because the opportunity costs of people becoming really well informed about a significant amount of what happens in their lives would be the massive slow down in the productive capacity of the society. Here, I fear that many thinkers have been misled by a confusion of normative and empirical claims. To say that people depend on a cognitive division of labor does not imply that the division of labor always works well. In some cases it may work badly as in the case of people who put their faith in financial advisors and bought subprime mortgage loans at variable interest rates. This is also an information shortcut depending on a division of labor. Brennan mentions Somin’s and Achen and Bartels’ dismissal of information shortcuts. But it is clear that these authors have not deeply thought about these information shortcuts and it is clear that their discussions confuse normative and descriptive accounts. Their discussions are cursory and dismissive (although Achen and Bartels do a lot better). But about the descriptive fact, there can be no doubt that this is a fundamental feature of voters’ activity in economic and political life. So here is the other part of the micro theory of the voter. We need to know what kinds of information shortcuts they face and how they aggregate many of these pieces of information. Presumably this will depend on the networks of people they find themselves in either as friends, family or coworkers. We need to know about the larger system of information such as newspapers, television, internet, education and social media. We need to know about the political institutions such as political parties and interest group associations. These are just a few of the elements of a division of labor that people commonly rely on. Downs has a longer list and though I don’t agree with all of his observations (such as political parties simply want to maximize votes) he give a pretty thorough discussion.

   To neglect this complex web of information is a disaster for any democratic theory that claims to explain what voters are doing. This is all the more extraordinary in the light of the enormous investment that democratic societies put into this information system.

   Now lets think about the normative question here. My thought is about a possibility, which I think is partly actualized but could be improved. The first thing to note is that it doesn’t follow from the fact that a person doesn’t know certain crucial facts about who they are voting for or what they are buying that they are not acting on good information. Everything depends on whether the links they are depending on are trustworthy. Again, if I buy a car based on my wife’s reading of consumer reports, normally I buy a good car based on good information, even though I couldn’t say much that is true about the car’s essential properties. And my wife has probably forgotten them after a few days as well. So here is one main lesson. The surveys of opinion that occur after elections, which display such seemingly alarming ignorance among voters do not imply that voters are not acting on good information. It simply doesn’t follow. These results are completely compatible with action based on good information. I am not saying that people are always acting on good information or that everyone usually is in modern society. The point is that the data can be explained by people acting on good information based on a division of labor. This is the point that people like Somin and Achen and Bartels completely miss. Partly, I think they are relying on a mistaken conception of knowledge, which requires that one know that one knows in order to know. They are thinking of the kind of person who might have existed before the development of agriculture and thus an extensive division of labor. Partly they are confusing normative and descriptive concerns.

   From here I have two normatively relevant ideas that I can only partly defend. One, in contemporary societies the division of labor people rely on in order to make democratic choices are somewhat effective in getting people to act on good information. Two, since people’s economizing on information gathering relies heavily on the division of labor, whether the information networks are capable of getting people to act on good information much depends on the structure of the institutions in society.

   In defense of the first claim I do unashamedly put forward the empirical results Brennan is so disarmingly unconcerned with. The facts that democracies do not go to war with each other, are the creators of international institutions, are by far better protectors of human rights, are much better at promoting public goods, avoid famines, enhance economic growth, lessen poverty for their members are all well established. And these facts compare very favorably with self-proclaimed epistocracies such as Iran, the Soviet Union, or Britain in the 19th century. These facts compare actually existing epistocracies with actually existing democracies (of high quality, say above 8 on the Polity IV scale). That suggests that the most vulnerable populations in society do better under democracy than under epistocracy. And one of the key differences between these societies is that the most vulnerable populations have a say in one and not the other. This cannot be conclusive since we need to look at other evidence and we need to rule out confounding factors, but I think it is a good start. Simply to not be concerned with such information is to suggest that one is not serious about instrumental evaluation. This is the best we have. Furthermore at least in some of the cases such as protection of human rights, when one attempts to decompose the democratic state in order to ask what feature of it has the most explanatory power it is the majoritarian element and not the division of powers or the rule of law (see Christian Davenport, State Repression and the Domestic Democratic Peace.)

   The second normative thought is that one should pay attention to the institutions that people depend on to act on good information. The idea is that some institutions will be better at enabling people to act on good information and others worse. And we should make some effort to try structure institutions so that people from all walks of life can have the ability to act on good information. There are numerous institutional proposals for this kind of thing. Campaign finance legislation is meant to limit the superiority of the wealthy in disseminating their views and enhance the capacities of others. Here the values one has are going to play a role in determining what the best institutions are. A highly egalitarian approach will attempt to bring about institutions that enhance everyone’s capacity to act on good information. This may sometimes imply limitations on freedom of speech or freedom of contract (if we take seriously the idea that unions can disseminate what Downs calls free political information to workers).

9. David,

   Thanks. What I mean is institution economics as studied in neoclassical economists such as North, Acemoglu, Weil, Robinson, and so on, not the heterodox school from way back when.

10. Tom makes a number of important points. Democracy is the best system we’ve tried so far, and it’s clear that it does certain things extremely well, compared to the other systems we’ve tried. Just why it does so well is not clear, though it is clear that it over performs compared to what we’d expect if the most naive version of the median voter theorem were true.

    Tom is right that my approach is essentially comparative. I do think democracy is to some degree intrinsically unjust, because elections violate the Competence Principle. But in the end, what I’d like to do is experiment with certain forms of epistocratic “fixes”, and if these work, scale them up.

11. Pingback: NDPR Discussion of Against Democracy - Bleeding Heart Libertarians

12. Tom, thanks for engaging with this piece. I think my earlier posts come off as more arrogant or cranky than I intended. I’m sorry for that. I didn’t mean to take a strident tone.

    I have a question for you: In your earlier books, you worried that voters were not informed enough to choose good means to their ends. You though they were capable of deliberating about and choosing good values, but you thought that they were not sufficiently well-informed to choose policies (including party platforms) that would realize those values. You advocated a democratic system in which voters chose ends, but in which technocratic parties would then be tasked to realize those ends. This seems to me to a be a significant departure from what happens in real-world democracy, where parties always run on a mixture of goals and policy-platforms.

    But in your critique you seem to have backed off from your earlier skepticism about voters. Is that right? And if so, why?

13. ‘In my paper “Democracy and Freedom,” I note that democracy is positively correlated with A) economic liberty, B) civil liberty (defined without including voting rights, etc.) and C) GDP/capita.’

    I am in no doubt that this is the case now. But it is no real arrgument for causation. I suggest that in fact rather than democracy causing the good things in your list the truth is that the good things caused democracy. In other words rather than economic liberty et al thriving in a democracy, democracy has thrived because of economic liberty et al. Britain and the US are the two great examples here. In the 19th century was not a democracy, but a monachic opligrachy with minimal suffrage. The same was true of the US in colonial times. But both had your A, B and C. In the US this led to the wish to break free of supposed repression of the mother country and retain A,B and C. In the UK the call for more democracy grew as the country became more and more successful in terms of A, B and C.

14. Hi Jason, thanks for the discussion. I want to come back to something Tom Christiano said. I agree that the self-interested voter hypothesis is false. But, while it is false that people vote in their self-interest, it seems more plausibly to me that voters (like everyone) suffer from powerful self-serving biases. Of course, these biases have been extensively documented by psychologists and you consider some of this evidence in you book. If voters suffer from self-serving biases, this may lead them to discount or ignore the interests of people unlike them, even if voters are otherwise highly educated.

    Take a look at Chris Blattman’s slides on democratization here: http://chrisblattman.com/files/2017/05/9-Democratization-wrap-up-1.pdf

    I’m struck by the evidence that the expansion of the franchise really does seem to change policy in what seem like beneficial ways. There seems like plausible evidence that this effect is causal.

    One hypothesis: while people don’t vote in their self-interest, they still ignore and discount the experience of other people unlike them and certain issues that affect out-group members are just not salient to them. Expanding the franchise to some extent counteracts or counterbalance these biases. This would help explain why expanding the franchise seems to cause good things, like improvement on health indicators, less corruption, etc.

    I admit I’m not quite sure why expanding the franchise seems to have good results given that that the self-interested voter hypothesis is false. But I worry that limiting the franchise even to more competent people would have bad effects. Let me just ask: what do you think about the studies that Blattman discusses?

15. Jason,
    I haven’t backed off the skepticism entirely. I still think the primary aim of the citizens is to provide direction to the political system by the choice of values. Of the process is complex since the political parties are trying to persuade citizens of what the right values are. But I have changed my mind on two things since The Rule of the Many. One, I have given up the really strong skepticism I once had about citizens’ understanding of policy. And this is because of closer study of Downs’ work and because of the ubiquitous reality of shortcuts. This doesn’t mean that there cannot be major failures, but I do have what I take to be a reasoned confidence that society can be organized so that all citizens can have access to good shortcuts. Two, a philosophical change. I thought that the conviction expressed in Rule of the Many that all that democracy required was direction was mistaken. I now think that the idea of democracy requires not only that the society receive direction from the citizens but also that they have reasonable confidence that the system is acting in good to pursue it. This is where I think a good system of institutions that give citizens good shortcuts can be important. I’ve tried to articulate how this can work in my paper “Rational Deliberation among Citizens and Experts” in Deliberative Systems.