Cass R. Sunstein’s recent book, On Freedom, is an engaging discussion of the relationships between individual freedom, welfare, and paternalism. In a previous book, Nudge (coauthored by Richard Thaler), Sunstein had called his views on these matters “libertarian paternalism,” because he thought that freedom to choose can be preserved by regulations that enhance individuals’ well-being as long as those regulations affect the contexts in which individuals make their choices without reducing the repertoire of choices available to them. For example, legal rules could increase employees’ long-term well-being by automatically enrolling them in a retirement plan unless they opt out. Such rules just “nudge” citizens toward choices that advance their perceived interests. As Sunstein puts it, “[f]or those who fear or distrust government, the principal focus should be on mandates, bans, subsidies, and taxes.”
Sunstein has long claimed that individuals are less free not only when physical obstacles, including the actions of other individuals, interfere with their actions, but also when cognitive biases prevent them from choosing the most effective means to their ends. He takes the fact that politicians are not exempt from the very biases that good nudges aim to counteract as a reason for making careful tradeoffs between market biases and political biases when it comes to deciding whether legal nudges are appropriate. Sunstein thinks that such tradeoffs should be governed by a utilitarian standard: nudges should maximize aggregate welfare.
In On Freedom, Sunstein argues that there is a positive correlation between freedom and welfare. Individuals who don’t know how to achieve their most cherished goals cannot “navigate,” as he puts it, their lives. As a result, their freedom is compromised. Here Sunstein seems to embrace what philosophers call a “positive” conception of freedom. On that conception, an individual is free to do or achieve X to the extent that she values X and can do or achieve it. In contrast, traditional libertarians tend to give pride of place to “negative” freedom, defined as the absence of interference, broadly understood to include coercion and punishment. Notice, though, that the negative conception of freedom cannot be wielded against Sunstein’s libertarian paternalism, since, as we saw, his proposals do not involve interference, at least to any significant degree. They just help individuals navigate their lives by giving them relevant information or enrolling them in arrangements from which they can easily opt out.
Sunstein points out that both markets and politics are sources of good and bad nudges: the potential for abuse exists in both sectors. But the plausibility of such remarks may distract us from the fact that there are market pressures towards good nudging, whereas politics is systematically biased towards bad nudging. Bad nudges are not hard to find in the private sector, as shown by Sunstein’s example of the way in which many restaurants arrange food lines, with unhealthy foods being more salient. Nevertheless, firms have incentives to offer good nudging. For example, clinics, health and fitness magazines, and other health-related firms have incentives to provide reliable medical information, and the spread of such information surely plays an important role in explaining the growth of the market for healthier food over the last several decades. Likewise, ad-financed websites enhance their reputations by providing prospective car buyers reliable information about the strengths and weaknesses of brands and models—a counterweight to the bad nudges provided by the display rooms in car dealerships. In the free market, the very existence of bad nudging provides incentives to sell good nudges (typically, relevant information), which in turn provide incentives to sell goods and services that meet the perceived interests of better-informed consumers.
Can politics offer any such self-correcting mechanisms? Sunstein suggests that government nudges are appropriate as long as they are “constrained by democratic requirements, including transparency, public debate, and independent monitoring (including continuing evaluation of how nudges work in practice),” and are “respectful of individual rights.” However, the structure of incentives under which public political deliberation takes place makes it unlikely that citizens will debate these issues fruitfully or support suitable constraints on the power to nudge, even if, by hypothesis, non-abusive nudging will benefit them. Voters have little incentive to spend time and money studying politics, given that each individual ballot has vanishingly small chances to affect political outcomes. Scholars have amply documented this “rational ignorance” effect. But perhaps no less importantly, politicians have incentives to appeal to, rather than challenge, political views that look reasonable in the eyes of uninformed citizens. As Fernando Tesón and I argued in Rational Choice and Democratic Deliberation: A Theory of Discourse Failure (2006), the rhetorical equilibrium that stems from the mutually reinforcing interaction between widespread political ignorance and political posturing is predictably at odds with the views that prevail in rigorous social research. Ironically, some of the cognitive biases described by Sunstein, such as the “present bias” that explains the resilience of self-destructive addictions, can only reinforce such discursive pathologies, as illustrated by the growth in deficit spending prior to elections—the “political myopia” predicted by economic models of political decision-making. In The Myth of the Rational Voter (2007), Brian Caplan describes various types of biases affecting voting behavior, and argues that such forms of irrationality are due to the low personal cost of being mistaken at the polling booth. These findings cast doubts on the hope that “democratic requirements” will prevent government from misusing the power to nudge.
Abuses of Power
Notice that these reservations about the prospects of preventing abuses of the power to nudge do not presuppose a slippery slope from libertarian paternalism to authoritarian politics. To be sure, nudges may set precedents that, through increasingly expanded interpretations of their underlying principles, will provide rhetorical ammunition for justifying mandates, bans, subsidies, or taxes, thereby jettisoning the libertarian component (i.e., the right to opt out) of Sunstein’s proposal. But I am not saying (or, for that matter, denying) that politicians endowed with the power to nudge are likely to start a process that ends up with them forcing citizens to make prudent choices. Rather, I want to suggest that, as long as particular government nudges presuppose that government has the power to nudge, the public justification of any particular nudge will collide with the anti-paternalistic stance to which supporters of liberal democracy, including Sunstein, are committed. Let me explain.
Suppose that a particular nudge is morally justified by, to use Sunstein’s criteria, its freedom- and welfare-enhancing consequences. To be intelligible, any such justification must also apply to the power to make that nudge: government cannot, as such, permissibly make nudges without the legal powers to make them. But if freedom and welfare justify such powers, then our above worries about imprudent (because of misinformed) voting behavior should warrant, too, governmental powers to nudge voters. Indeed, the libertarian paternalistic reasons for conferring governments such powers would be arguably stronger than the libertarian paternalistic reasons for realigning food lines in restaurants and many of the other cases of libertarian paternalism that Sunstein proposes, since widespread and serious harms and benefits are at stake in presidential and other major elections. Voters pick legislators and presidents empowered to declare war and command the armed forces, adopt economic policies that may bring about massive destitution, and make other major decisions on which the lives, liberties, and welfare of many—including those voters themselves—depend. This being so, properly framed referenda questions, tinkering with voting systems and the boundaries of electoral districts, and other measures aimed at securing electoral outcomes that promote freedom and welfare would meet libertarian paternalistic criteria. Or they would at least for as long as citizens remained free to vote as they wish, move around electoral districts, or otherwise opt out of the policies promoted by those nudges.
Now such uses of the power to nudge may in due course give rise to an illiberal or undemocratic regime, and indeed one that reduces overall freedom or welfare to a greater extent than the above discursive pathologies of typical liberal democracies do, even if each particular nudge promotes freedom and welfare. For example, the nudges on voters imagined in the previous paragraph may, over time, give rise to an electoral coalition capable of enshrining dictatorial powers through constitutional amendments, even if, as I assumed, each of those nudges enhanced freedom or welfare. This illustrates a point often made in discussions of moral consequentialism, including utilitarianism: a set of separately good-promoting actions may, qua set, do much harm. Consider this analogy: each dose of a painkiller may benefit a patient relative to a situation in which everything is equal but for the absence of that dose, yet many such doses may harm her relative to the kind of life she would have lived with no, or very few, doses.
Nudging as Manipulation
It might seem that Sunstein’s democratic requirements on nudging (see four paragraphs above) would forestall such misuses of the power to nudge. Why wouldn’t constitutional rights, independent courts, and other institutions and practices of a well-functioning liberal democracy block the slippery slope from bona fide libertarian paternalistic policies to illiberal or undemocratic ones? Notice, though, that I am not saying that nudges, no matter how defensible they may be when taken in isolation, set in motion a process that ends up in question framing, tinkering with voting systems and electoral districts, and other manipulations made by power-hungry politicians. While we should certainly be wary of such illiberal or undemocratic slippery slopes, the process that I have in mind is not a slippery slope. I assumed, arguendo, that the power to nudge is justifiable as part and parcel of the freedom- or welfare-based justification of particular judges. The problem is that the power to nudge involves the power to manipulate, in ways inimical to freedom of welfare, the aggregate outcome of separately defensible nudges. Sunstein toys with the idea that legislators may justifiably nudge citizens into saving for retirement, buying health insurance, driving more carefully, etc., given that those measures increase those citizens’ freedom or welfare. Now as this short list suggests, there are multifarious ways of exercising the power to make freedom- or welfare-enhancing nudges. Nudgers must then be endowed with significant interpretive leeway when it comes to setting priorities. The contestability of the conception of welfare governing those decisions would require further interpretive leeway, which in turn expands the power to pick winners and losers among the special interests that compete for self-serving regulation of the contexts in which consumers make their choices.
Suppose rulers exercise their legal powers in conformity with textualist understandings of those powers. The slippery slope critique of government nudging would then be misdirected. Yet, if I am right, the cumulative consequences of such non-abusive exercises of powers might give rise, in realistic scenarios, to forms of manipulation that adversely affect individuals’ freedom or welfare. On Sunstein’s criteria, then, we should not grant such powers to anyone in the first place. Of course, if we suspect, as I think we should, that politicians would abuse their power to nudge, we should be even more reluctant to grant them that power. But my present point does not turn on that kind of cynicism.
We can therefore agree with Sunstein that particular legislative nudges are freedom- or welfare-enhancing, while being wary of government nudges—even freedom- or welfare-enhancing ones. For government nudges come with the power to nudge, and the power to nudge involves the power to manipulate.
Sunstein is right that private firms sometimes nudge consumers to the detriment of their freedom or welfare. However, in a free-entry market such firms face competitive disadvantages that have no parallel in political decision-making. Thus, the very existence of pre-formulated standard contracts self-servingly written by private firms gives competitors incentives to offer better terms (and to warn consumers of their competitors’ fine print). Likewise, consumer-report websites have incentives to sell reliable reports to their subscribers. Contrast those settings with those in which laws or regulations automatically enroll employees in a retirement plan from which they can opt out. While such measures might separately promote freedom and welfare, they presuppose, as we saw, legal powers that engender perverse incentive structures. Economically rational citizens will ignore such dynamics for the same reasons that lead them to underproduce public goods.
The current public controversy over web search engine bias illustrates the potential and limits of nudges as engines of freedom or welfare. Some conservative commentators complain that Google searches of politically relevant topics prioritize non-conservative viewpoints. Now searchers can often bypass such suggestions just by spending a few moments scrolling down the results or searching for other descriptions of the same topic. Should alternative search algorithms lead Google users to better results in terms of, to use Sunstein’s criteria, their freedom or welfare, there would be room for libertarian paternalistic regulation of those algorithms. But even if particular regulations could be justified along these lines, the power that they presuppose would be hard to justify on those criteria. There is much at stake when we consider who is to possess that power, not just because those wielding it may be affected by the cognitive biases that Sunstein describes, but also because algorithm designers can capitalize on the cognitive biases of ordinary citizens to garner support for public policies adverse to their freedom or welfare.
I do not mean to suggest that these considerations suffice to adjudicate between advocates of regulations of web search nudging (and the power to regulate that this presupposes) and advocates of unregulated market nudges. But I do think that any reasonable verdict on these issues should take notice of these two facts: (i) in a free-entry market, technological innovation either upsets inefficient monopolies or gives rise to a sequence of provisional monopolies that is dynamically efficient, and (ii) the power of the state to use freedom- and welfare-enhancing nudges in ways that are cumulatively counterproductive reinforces itself.
To be sure, Sunstein advocates case-by-case assessments of the costs and benefits of government nudges. But the previous discussion suggests that such assessments will be misleading if they overlook the cumulative effects of freedom- and welfare-enhancing tinkering with the contexts in which individuals make choices. With this caveat in mind, I highly recommend Sunstein’s book to anyone interested in a readable and well-argued study of the relationships between freedom, welfare, and the law.
 For evidence and discussion of the gap between popular beliefs about the most basic features of the economic system and those prevailing among academic economists, see Bryan Caplan, The Myth of the Rational Voter: Why Democracies Choose Bad Policies (Princeton: Princeton University Press, 2007), pp. 23-93.
 For a recent example, see Ohad Raveh & Yacov Tsur, “Political Myopia, Public Debt, and Economic Growth,” OxCarre Working Papers 200, Oxford Centre for the Analysis of Resource Rich Economies, University of Oxford, 2017.
 Glenn Whitman and Mario J. Rizzo develop the slippery-slope critique of nudges in “Paternalist Slopes,” NYU Journal of Law and Liberty (2007).