Liberalism and Complexity: On Gerald Gaus’s "The Open Society & Its Complexities"

For most of us, if there is a possible articulation between liberalism as a general moral and political theory and the idea of complexity, it is to be found in the writings of Friedrich Hayek. With his book The Open Society and Its Complexities, Gerald Gaus has considerably updated the Hayekian articulation to give it a distinctive flavor and make it relevant to tackle some of the challenges that contemporary liberal democracies are facing.

Gerald Gaus was a professor of philosophy at the University of Arizona. He died while this book – the first one in the new “philosophy, politics, and economics” series at Oxford University Press – was about to be published. I’ve been reading Gaus work for a few years now, as I was at the time searching for fresh and ambitious theoretical approaches in moral/political philosophy less disdainful of social sciences (and especial economics) than one usually finds. My first encounter with Gaus was his huge The Order of Public Reason, followed by his more modest (in size, not in scope and ambition) Values and Justification, Justificatory Liberalism, and The Tyranny of the Ideal, as well as several of the many papers he has written and published. To my relative astonishment, I’ve realized that Gaus’s work, in spite (or because) of its originality has not reached a significantly wide audience among English-speaking professional philosophers[1] and is mostly ignored by social scientists, even among those with a form or another of philosophical inclination. He has nevertheless been locally influential and led to the emergence of a young generation of American political philosophers whose work follows Gaus’s intellectual path and is gaining growing attention.

Gaus’s philosophy is I think appropriately characterized as post-Rawlsian. There are at least two respects in which this label qualifies in this case. First, like Rawls (though this is something that is often either forgotten or criticized by philosophers), Gaus’s work largely builds on concepts, tools, and theories from the social sciences, especially economics. In the same way as Rawls was – not fully convincingly for many – using the economic theory of rational choice to argue for his principles of justice or relying on psychology to elaborate on his account of the stability of justice as fairness, Gaus made large use of social sciences models and tools to develop his philosophical account. But Gaus was far more systematic in this endeavor, making also repeated calls on the most recent experimental work in economics and psychology. This trait has become more and more prominent in his most recent writings, as his last book witnesses.

The second aspect that makes Gaus a post-Rawlsian philosopher is the general issue with which he was concerned, which I would formulate as the problem of normative justification in a morally diverse society. On this, Gaus’s (and also his followers’) work bears more on Rawls’s ideas developed after A Theory of Justice, especially as they appear in Political Liberalism. While Rawls is mostly recalled for his defense of justice as fairness, he was more focused during the last part of his career on what is now called the stability problem; in essence, how a conception of justice can gather a consensus in a liberal society where, because of its institutions, a “reasonable pluralism” prevails. For Gaus, the point of political philosophy is not so much to determine what is the right and the good than to characterize the conditions under which it is possible for people disagreeing on many issues to nonetheless agree on a moral code constituted by rules that everyone regards as justified. But as for the first point, Gaus’s work takes a more radical and systematic view: if we take the issue of diversity seriously, then we should not expect philosophers (or anyone else) to have the capacity and the authority to assess justice or moral views. Beyond, while for Rawls diversity and pluralism were seen as a risk for liberal democracies to be mitigated by an appropriate overlapping consensus, Gaus defends the radical view that diversity offers an opportunity for liberal societies to justify moral rules within some sort of dynamic equilibrium.

*************************

These two post-Rawlsian aspects are present throughout Gaus’s books and papers but are singularly accentuated in his last book. Indeed, they are so much accentuated that the result is very far from the Rawlsian model of liberal democracies as well-ordered societies. Instead, Gaus develops what he calls the open society model that analyzes liberal societies through neo-Hayekian lens. The book is indeed largely devoted to a critical assessment of some of Hayek’s ideas and claims about complexity, constructivism, and social order. But ultimately, there is few of Hayek in this book, except for the very general claim that liberal societies are complex systems in which moral rules are the product of social and cultural dynamic processes and which global behavior following an intentional intervention is largely unpredictable.

The book takes three Hayekian theses as a point of departure, each motivating one of its three parts:

Hayek’s Thesis #1: Our evolved, tribal, and egalitarian sentiments are in deep conflict with the impartiality and inclusiveness that characterize liberal and open societies.

Hayek’s Thesis #2: Liberal and open societies are too complex for the practice of moral justification, i.e., there is no way moral rules can be justified to anyone, except for the fact that they are constitutive of past and current practices.

Hayek’s Thesis #3: Liberal and open societies are in many ways beyond human control and governance, such that any intentional “constructivist” attempt to change their functioning must fail and make things worse.

Gaus deals with the first thesis by sketching a natural history of morality that, at least in its spirit, shares some similarities with Rousseau’s Discourse on the Origins of Inequality. A significant difference of course is that Gaus has access to decades of multi- and interdisciplinary research on the history of human societies, their evolved institutions, and the biological and cultural traits of their inhabitants. Gaus abundantly relies on a wide theoretical, empirical, and experimental literature, from evolutionary anthropology (Boehm, Henrich), ethology (De Waal), cognitive psychology (Tomasello), to behavioral economics (Bowles, Gintis), and far more. The depth of knowledge displayed is impressive, if at times overwhelming. The elements gathered point toward a natural history that confirms some of Hayek’s (and Rousseau’s) speculations but ultimately reaches a fairly different conclusion. In substance, Hayek was right to point out that the conditions in which humans have lived some 40,000 years ago have favored the development of an egalitarian ethos grounded both on genetic and cultural aspects. This “Modern Egalitarian Moral Package” (Gaus’s term) has evolved from an initial evolutionary branching where humans, separating from chimpanzees, have developed advanced cognitive skills, especially for strategic reasoning and social learning. While the first humans were not egalitarians, the living conditions of the late Pleistocene have led to the evolution of behavioral traits and dispositions at the basis of egalitarian, non-hierarchical societies. As it is now relatively well-known, individuals living in hunter-gatherer societies developed highly cooperative practices based on reciprocity, as well as a sustained disposition for rebellion against any attempt to install hierarchical relations. Beyond a disposition for conditional cooperation, these societies also provided humans with capacities for moral reasoning, from the internalization of (moral) rules to the development of argumentative skills, which are crucial in the development of practices of accountability that ground any moral system.

If Hayek was right to identify the origins of the “egalitarian sentiment” in hunter-gatherer societies, his error has been to characterize it as an atavistic and persisting trait that makes modern humans unfit for the kind of complex societies where individuals mostly interact in an impersonal basis, following rules and submitting to social processes that they barely understand and even less control. While it is true that we have inherited dispositions that make us not optimized for the open and liberal social order, both our genetic and cultural endowments have made us sufficiently flexible to adapt our behavior and institutional arrangements to new circumstances. In other words, our moral attitudes display both “tribal” and universal/inclusive features.

The most important however is that our natural history has given us the key ability to justify our practices to each other, an ability which is at the roots of the development of any moral system. Hayek’s second thesis claims however that moral justification is beyond the capacity of humans in complex societies. That presumably implies (at least in the most conservative reading of this thesis) that the prevailing moral rules of a sufficiently complex society are out of the reach of any critical assessment, but also that any man-made intentionally designed moral systems cannot rival a spontaneously evolved one.

With respect to this second thesis, Gaus largely builds on his previous work according to which justification proceeds from diversity and convergence on rules that everyone finds better from the moral point of view than a state where no rule prevails. The argument in this part proceeds in three steps. First, Gaus develops a thorough critique of “Millian liberalism” and “progressivism” on the one hand, and social contract theory (in its rationalistic version) on the other. Against the former, Gaus argues that it reflects ultimately an intolerant and sectarian stance that excludes, on the ground of some rationalistic faith, a wide range of beliefs from the realm of public justification, starting with religious beliefs. Because these beliefs are nonetheless part of human culture and at the bedrock of most societies, this makes progressive liberalism unamicable to our rule-following nature:

“To interrogate all rules – to insist that what cannot be understood is to be distrusted as superstitious or prejudice – is directly opposed to our nature as a cultural species. Our truly distinctive trait… is not individual reasoning but social learning. The progressivist program is a recipe for drastically reducing social learning (aka imitation), throwing us back on our cognitive capacities. That, however, is in turn a recipe for undermining ultra-social cooperation, and would probably make any significant system of social rules dysfunctional. Indeed, I shall presently argue that progress itself depends on our ability to copy” (p. 101-2).

So Millian liberalism has the conception of progress wrong and is noncompatible with human nature as it has evolved from the Modern Egalitarian Moral Package. The critique against rationalistic social contract theory is similar. Social contract theory builds on an impartiality requirement that viewed from the outside share some similarities with the kind of impartiality that we find in moral reasoning as it has evolved. But once we look closer, it appears that this impartiality – as well exemplified by the way Rawls’s models it with the veil of ignorance – is one that is aiming at stripping moral reasoning from all sorts of diversity and partiality. Hence the stability problem that arises in Rawls’s and virtually any other contractarian/contractualist moral account. Principles are derived from an artificial situation where parties are thought to be homogenous. Once these principles are situated in the actual diverse society, it appears however that individuals may lack the proper motivations and reasons to abide by them. Sophisticated and convoluted arguments are then needed to show nonetheless that these principles will be stable. The problem is however unsolvable and lies at the core of the social contract approach. It stems from not acknowledging the reconciliatory nature of morality in complex and diverse societies.

The second step is precisely to make it clear what is a complex and diverse society. Here, Gaus builds on well-known formal models of complexity, such as Brian Arthur’s and Stuart Kauffman’s. These models emphasize the significance of increasing returns to scale and autocatalytic processes: in a complex system, diversity begets diversity by increasing the number of adjacent possible states that can be explored. This is well-documented in technological innovation and Gaus suggests that this also applies to moral innovation. In a system with a relatively dense network of interactions, complementary mechanisms of exploration (innovation) and exploitation (imitation) tend to lead to an explosion of diversity. Complex systems that result from these autocatalytic processes tend to self-organize around rules that have not been intentionally designed. In the case of human societies and their morality, Gaus argues that self-organization leads to the emergence of rules related to the rights of persons and property defining a system of natural liberty. In such a system, everything that is not explicitly prohibited is implicitly considered as being permissible, thus favoring further moral experiments and innovations. The morality that evolves is focused on fairness, freedom, and harm. In other words, the morality of open societies is the WEIRD (Western Educated Industrialized Rich Democracies) morality as it has been characterized in the anthropological work of Joseph Henrich and others. This morality forms the minimal core “necessary to functional human cooperation” (p. 137).[2] From the Modern Egalitarian Moral Package, we have thus evolved to the Open Society Moral Package, basically the WEIRD morality.

The third step of Gaus’s account of moral justification in complex and diverse societies is to demonstrate that moral diversity is not a threat to justification but quite the contrary makes it more likely. This step illustrates what I said above about Gaus’s commitment to using tools from the social sciences as it appeals to agent-based modeling. Gaus presents several simulations of a population of moral agents having to choose which moral rule to adopt in a context where their “moral utility function” depends on (i) an intrinsic and more or less intense preference for one rule over the others and (ii) a disposition for reconciliation according to which an agent prefers, everything else equals, endorsing the same rule than others. The population can vary depending on the diversity of the intrinsic preferences for rules (in all the simulations, only two rules are available) as well as regarding the types of agents, which depend on the weight they give to reconciliation. The main lesson that can be extracted from these simulations is that “sometimes adding more diversity makes agreement more likely”. This answers Hayek’s second thesis about the impossibility of moral justification in complex societies.

The third thesis discussed by Gaus will of course be familiar to all readers of Hayek. It is related to Hayek’s well-known attack against “constructivism”, the idea that it is possible to intentionally control a complex system like a market or a society to achieve specific outcomes. Gaus does not spend any time recalling and discussing Hayek’s specific arguments sustaining his attack. Though the underlying idea is Hayekian, the discussion goes in a very different direction. In substance, Gaus asks about the type of self-governance that can be implemented in a complex system. He opposes two broad approaches. The first he calls the “Legislative Theory of Democratic Self-Governance”, according to which self-governance proceeded mostly through legislation by influencing behavior through the threat of punishment. Gaus is very critical of this view, arguing (not always convincingly, as I shall mention below) that not only the law is most of the time powerless in enforcing behavior when it goes against prevailing social norms, but also that sometimes it may “strengthens the very norms and attitudes it sought to displace” (p. 177). Against this view of self-governance, he pushes for the “Process Theory of Democratic Self-Governance”, according to which governance “is a process by which the democratic state interacts with non-state actors and informal norm networks to pursue its designated social goals/values results” (p. 178).

This pursuit of goals and values corresponds to the first dimension of self-governance. A second one, largely studied by economists and political scientists, is the management of “strategic dilemmas”, i.e., coordination problems, assurance problems, and suboptimality traps (prisoners’ dilemma and the likes). The third and final dimension is the direct action on the rules and the structure of interactions. These three dimensions of self-governance (pursuit of goals, strategic dilemmas, action on rules) can operate at three different levels: the macro, the meso, and the micro. Gaus proceeds to the detailed discussion of this “space of self-governance”.

To overly simplify the discussion, Gaus’s view can be summarized as follows. In the spirit of Hayek’s thesis, Gaus is highly skeptical about all forms of macro self-governance, both about the control of goals and the action on rules (sensibly, he does not consider the macro governance of strategic dilemmas). The argument Gaus offers in support of this view is a relatively standard one according to which the macro dynamics of a complex system is essentially impossible to predict. He is more positive with respect to meso self-governance, especially regarding the management of strategic dilemmas. Gaus relies heavily here on the work of Elinor and Vincent Ostrom and the polycentric approach to governance that it has inspired. The core of the message is that polycentrism helps to focus governance on the resolution of well-identified and pressing problems in which goals and perspectives are largely shared by individuals. Gaus’s positive outlook is however largely attenuated for meso governance about goals. It is difficult to implement in open societies where individuals will not agree on goals, even at the meso level. The only sensible possibility is to imagine a system of interconnected communities competing to attract individuals, in the spirit of Tiebout’s model in public economics. Ultimately, it is not clear however whether these communities can remain open. Gaus also invokes Philip Tetlock’s work on expert forecasting to argue (partially against Tetlock’s own views) that control of goals is made delicate by the fact that expert knowledge is most of the time not sufficient to predict the course of a complex system, even at a sectoral level.

Unsurprisingly, it is therefore the micro-form of self-governance that attracts Gaus the most. More than elsewhere, Gaus relies heavily on the social science literature to make his point, from agent-based models of the COVID pandemic to randomized controlled trials that are now standard in development economics. Gaus has in particular a very interesting discussion on norm change policy, opposing the “community-driven” approach of public justification to what he calls the view of the “impatient moralist”. In substance, Gaus (following notably the important work of Cristina Bicchieri) argues that changing norms cannot be done through a top-down process. It requires discussion and debate within the targeted population until public justification for the new norm is reached. Most of the time, the successful conduct of this change requires a deep knowledge of the community and its network of interactions, especially to identify “trendsetters”. This is a particular occurrence of Gaus’s more general dismissal stance against the authoritative (authoritarian?) pretense of moral and political philosophers to command what are the correct moral principles to which everyone should abide.

This directly leads to the concluding sections of the book, where Gaus reflects on the nature of liberal democracy and the role of scientific expertise in it. This is where Gaus’s views espouse those of Hayek the most. Gaus is wary of self-governance by (self-proclaimed?) experts. Because of our natural history which makes us creatures of conditional cooperation, skilled at learning but also argumentation, governance, and policymaking cannot ignore the requirement of public justification. And, according to Gaus, there cannot be public justification outside an authentic democratic regime where the “wisdom of crowds” is allowed to operate fully. Sidestepping democracy and public justification mean failures of governance, i.e., unsolved strategic dilemmas and unfair (and thus contested) outcomes and rules. On the other hand, democracy cannot be but liberal. The system of natural liberty and the rights of persons and property are part of the Open Society Moral Package sustained by the demand for public justification. Ignoring them means, again, that some will impose their goals and values on others. Ultimately, as Gaus puts it, “Hayek was deeply mistaken: without justification, the Open Society could not self-organize” (p. 246).

***********************

My discussion has been mostly descriptive. It is obvious however that such a rich book calls for many comments, including critical ones. As this post is already long, I will content myself with pointing out what I regard as the most contentious aspects, without discussing them in detail. I shall have the opportunity to return to some of them in future posts – once I will be back from holidays.

I think that most of the original and controversial content is to be found in the second and third parts of the book, and so concerns Hayek’s thesis #2 and #3. As I said above, the first part echoes a now large and relatively well-known literature about our natural history. Though there are still many controversies and unsolved issues in this literature, Gaus is mostly following the mainstream views. The most interesting aspect of this part is the way he makes use of these views to basically refute Hayek’s first thesis in a very convincing way. That Hayek was wrong is of course not a surprise: the literature has exploded only relatively recently, after Hayek’s death, and so he mostly had to rely on speculations.

The second part bears the most significant connection with Gaus’s previous work, especially with ideas advanced in his 2012 book The Order of Public Reason. Gaus indeed builds on the “deliberative model” he develops in this book. This model characterizes justified moral rules as rules that agents, idealized as “members of the public” who are willing to justify the use of a moral rule to everyone else, would unanimously prefer to use in a given context rather than “demoralizing” this context (i.e., relying on no rule at all). Among these rules, only those that are undominated (not deemed as inferior to another rule by everyone) are viewed as belonging to the “optimal eligible set” of morality. Gaus identifies three shortcomings with this model: (i) the optimal eligible set may be empty or (ii) quite large, and (iii) it ignores the fact that moral preferences are, at least for some persons, dependent on the number of individuals who adhere to the different available rules. The simulations based on the agent-based model that Gaus is running are thought to overcome these shortcomings. I must say here that I’ve not been fully convinced by the results. There are several concerns that can be listed. Some are related to the deliberative model but not to the shortcomings Gaus identifies. In general – but I can’t develop this point here – I find it inconclusive as a model of moral rules; the model is great at identifying a necessary condition for a rule to be moral (that it can be justified to everyone) but misses other necessary conditions. In particular, the fact that individuals agree on a rule for different reasons does not guarantee that it has a moral content; it might be argued, along Rawlsian lines, that this agreement must rely on some shared public reasons.[3] Other concerns are related to the methodology of agent-based models which typically generate that it is hard to see whether they can be generalized. In this specific case, Gaus’s simulations only consider convergence results when two rules are in competition. There is no a priori reason to suppose that this convergence result generalizes when more rules are in competition (no indication is given about this). Furthermore, when applying the model to the issue of polarization, Gaus essentially considers populations with widespread and polarized disagreement over the inherent moral value of the two available rules, while it can be thought that polarization is also the result of the fact that a growing fraction of agents are not willing to revise their moral preferences even when they are in minority (i.e., they do not give important weight to the fraction of agents that follow a given rule to make their comparative assessment). In this perspective, a useful extension of the model would be to endogenize agents’ types. Finally, even if we accept the convergence result of the simulations, one of the limits of the deliberative model still remains: there is no guarantee that we are converging to a rule that is genuinely moral, in spite of the fact that it publicly justified.

I have also some qualms with Gaus’s attack on “progressivism” and Millian type of liberalism. Of course, I think we can agree with him that, as a matter-of-fact, excessive secularism and perfectionism jeopardize the very possibility of public justification on which a liberal and open society rests. But, even if Gaus explicitly defends himself against this charge (p. 102), his views leave open the justification of customs and traditional norms which can be reasonably regarded as oppressive. Gaus is right that philosophers do not have any special authority in judging what is the right or the good. On the other hand, there must be some principles that enable us (the “competent” judges to use a Rawlsian expression) to discriminate between genuinely moral and non-genuinely moral rules, even when all the members of the population regard them as justified. In contractualist accounts, this is usually done by idealizing the moral deliberators. As he already argued at length in The Order of Public Reason, Gaus’s deliberative model settles for a very weak form of idealization so that something akin to the stability problem does not arise. The downside however is an account that says relatively few about the content of moral rules, and maybe too few in light of the contemporary crisis of liberal democracy. Again, I think that part of the difficulty stems from Gaus’s unwillingness to grant that the reasons why people accept moral rules are at least as important as the fact that they de facto accept them. In the public domain, it is just not true that it does not matter why someone accepts a rule or a proposition. This is not true because, given one’s beliefs, there is no guarantee that he will not come to accept authentically immoral rules when circumstances change, rules that they will be willing to impose on others.

I must also say that I have been at times irritated by the way Gaus presents some arguments. This is particularly flagrant in the third part when Gaus discusses the Legislative Theory of Self-Governance, taking the delicate case of abortion as an example (p. 172-7). Here, Gaus makes use of some basic time-series data on the evolution of views about abortion in the U.S. (proportion of pro-life/pro-choice persons, views about the moral acceptability and the legality of abortion) to argue that a law can reinforce norms that oppose the law. As a general fact, there is no doubt that this exists. But the evidence that Gaus presents in the case of abortion does not support this hypothesis at all. Indeed, while the law regarding abortion changed following a decision of the Supreme Court in the 1970s, Gaus’s data start in 1995 and so they cannot establish that in the case of abortion “the law sometimes strengthens the very norms and attitudes it sought to displace” (p. 177).[4]     

The third part is otherwise very interesting in the way it builds on the articulation of philosophical insights with social sciences. There is nonetheless a tension that goes through it all along. On the one hand, as we have seen, Gaus is wary of the experts’ pretense for governing an open society. On this, he is deeply Hayekian. On the other hand, he is obviously keen on relying on results that are produced by methods that are representative of the increasing engineering nature of social sciences, especially randomized controlled trials. Proponents of RCT like Esther Duflo and others are not shy at expressing their beliefs that it is the role of social scientists to help to improve the living conditions of people, especially the poorest. We may of course agree with this claim and the underlying scientific and political project. But even though Gaus takes care of insisting that this only applies at the micro level where, presumably where individuals’ interests are largely and not controversially converging, this looks like more Popperian piecemeal engineering than a Hayekian approach to self-governance. This is not an objection in itself of course.[5] More problematic maybe is the fact that these methods can also be seen as participating in the growing technocratic nature of liberal democracy, something which Gaus must have obviously seen as worrying. Indeed, the book ends in a way that suggests that Gaus was planning to pursue his thinking on the nature of liberal democracy and on the role of experts in it. Others will have to carry on the task for him in the future.

---

[1] Unsurprisingly, in light of the fact that his work has not been translated into French, Gaus is virtually unknown by French-speaking philosophers (at least those working in a French institution).

[2] Gaus makes the interesting observation that experiments suggest that harm seems to be an exception in the system of natural liberty that characterizes the morality of open societies. That is, individuals tend to regard harm (including self-harm) to be prohibited, even if it is not explicitly so by moral rules.

[3] There is a lively debate undergoing in the post-Rawlsian literature between so-called “convergence” and “consensus” accounts of public reason. Gaus and his followers are the main protagonists of the convergence side.

[4] Indeed, the data do not display any significant change regarding views about the morality and legality of abortion in the U.S. population over the last three decades. There is a significant change in the proportions of persons who describe themselves as pro-life or pro-choice. But this change didn’t start before 1995, more than 20 years after the legal change. One wonders how it is possible to make the inference quoted in the main text on this basis!

[5] Even though in a 5 years-old essay, Gaus was clearly expressing his despise for the “sectarian” Popperian account of the open society and his preference for the Hayekian version. Nonetheless, the tension I’m outlining was already palpable in this essay.