II. Degeneracy in Complex Systems

Social scientists often utilize concepts from complex systems theory. ^{Footnote 6} Complex systems are described in different ways, but four commonly recognized characteristics are salient for the following discussion. First, relationships between the elements of complex systems exhibit structure. Elements are not all directly connected to one another, but rather comprise a network in which each node only connects to some subset of other specific nodes. Second, elements of the system are specialized; each performs only some subset of the functions that all elements together can perform. Third, specialization entails interdependencies between elements, which often rely on one another in order to function. Fourth, the combination of these three traits means that the nexus of networked relationships that constitute the system has emergent properties that affect both the system as a whole and how the elements of the system function.

In complex systems, specialization and interdependence create the possibility of fragility. ^{Footnote 7} Each part of a complex system, in order to fulfill a certain function, requires the cooperation of other elements of the system. A system is fragile if some disruption or change in its environment causes a systemic breakdown of functionality. In the proverbial rhyme, a kingdom is lost for want of a horseshoe nail: a breakdown of one function (the nail securing a horseshoe to a horse) leads to the breakdown of another function (the horseshoe protecting the horse from injury), and so on.

Redundancy is the classic solution to the problem of fragility. If a single missing nail can render a horse lame, one might give riders access to multiple horses or send out multiple riders. Redundancy means that some elements of a system function identically. If one breaks down, the others pick up the slack. In economic theory, redundancy corresponds to roughly identical firms in a perfectly competitive market: each firm’s output is a perfect substitute for the output of other firms. Redundancy between elements can make a system robust, which is the simple negation of fragility. Robustness is the ability of a system to withstand different kinds of shocks or stressors.

Degeneracy differs in a subtle but important way from redundancy: degeneracy refers to the ability of structurally different elements of a system to perform similar functions. ^{Footnote 8} Degeneracy refers to the relationship between elements of a complex system. Elements exhibit degeneracy when they perform the same function under some circumstances but not other circumstances. Neural networks offer a prime example. Individual neurons are structurally distinct but densely connected, and offer many similar pathways for electrical impulses. Even if individual neurons are damaged, cognitive functions can continue as normal since there are substitute pathways for impulses to follow. ^{Footnote 9} Like redundancy, then, degeneracy facilitates robustness by multiplying the means of realizing any given function.

But degeneracy also allows for the possibility of transcending mere robustness and achieving “evolvability.” Evolvability is the emergent capacity of a system not only to withstand negative shocks but also to adapt to new circumstances or develop wholly new functions. ^{Footnote 10} Pure redundancy can generate robustness but inhibit evolvability as system resources are spent maintaining a population of identical elements. A degenerate system, by contrast, can respond to new environmental conditions by developing new capacities. Since elements function differently under different circumstances, the system as a whole is more adaptable. In economic theory, degeneracy characterizes dynamic competition between different firms developing distinct products that are imperfect substitutes for one another. This is a form of competition that allows for experimentation and innovation, not just lower prices on given goods.

Nassim Taleb offers a helpful taxonomy of qualities of complex systems that correspond to the above discussion of the relationships between their elements: fragility, robustness, and anti-fragility. ^{Footnote 11} He defines robustness and fragility in a manner equivalent with the above. Anti-fragility is not just the negation but the diametrical opposite of fragility: it is the ability of a system to take advantage of changes in its environment. Anti-fragile systems love volatility. Anti-fragility entails the ability to develop new capacities, and so corresponds to the conjunction of robustness and evolvability. Taleb's argument that anti-fragile systems respond to negative stressors through overcompensation, increasing the capacity of the system, corresponds to Edelman and Gally's association of degeneracy with ‘adventitious compensation for error.’ ^{Footnote 12} Redundancy creates robustness, while degeneracy creates anti-fragility.

Taleb raises several important points about the evaluation of complex systems. First, robustness has a built in expiration date. ^{Footnote 13} The best a robust system can do is resist adverse shocks. Without the capacity for evolution, a robust system cannot adapt to large changes or to exploit new possibilities. The conditions that allow a given system to thrive will eventually dissipate. Even if evolvability is not considered valuable for its own sake, it is necessary to adapt to adverse changes in conditions. A system is robust in the long run only to the extent that it is evolvable.

Second, one cannot judge whether a system is fragile, robust, or anti-fragile based only on its previous performance. Performance is the joint product of the qualities of the system and the circumstances it confronts. Notably, “black swans” — unexpected negative shocks — can reveal fragility in systems that previously appeared quite stable. ^{Footnote 14} The financial crisis of 2007 is one such example; the outbreak of World War I is another. Understanding whether a system is fragile, robust, or anti-fragile thus requires understanding how the elements that comprise the system relate to one another, not merely how the system has performed in the past.

Finally, it is worth making an obvious qualification: robustness and evolvability are both matters of degree and of kind. A system will be anti-fragile to the extent that relationships between its elements are degenerate and with regard to the functions which those elements perform. So it is possible for a system to be fragile, robust, and anti-fragile in different respects.

III. Degeneracy and Governance

Richard Wagner explores the implications of degeneracy for social phenomena. ^{Footnote 15} He models social phenomena as emerging from networks comprised of both private and public enterprises. Degeneracy requires that there be substitutes for the functions performed by individual enterprises. A system is degenerate to the extent that enterprises have multiple pathways open for achieving their ends. “If an enterprise finds one node blocked, there are other nodes that can serve the same function in the enterprise's plan.” ^{Footnote 16} If any enterprise has a monopoly on some function, the system is less degenerate to that extent. ^{Footnote 17} If all connections must pass through a single enterprise or all political enterprises are directly connected, there is zero degeneracy between nodes.

Mapping the concept of degeneracy onto governance more specifically requires exploring two key ideas from the work of Vincent and Elinor Ostrom and the broader Bloomington School of political economy: polycentricity and institutional diversity. ^{Footnote 18} Polycentricity means that there are multiple sources of authority in a governance regime, and diversity means these sources of authority offer different bundles of governance services. To the extent that a regime is polycentric and diverse, relationships between jurisdictions exhibit degeneracy, and thus the system as a whole is anti-fragile.

As noted above, in making this argument I do not aim to enshrine polycentricity or diversity as ideals that always trump other considerations. In the first instance, though, it suffices to point out that anyone designing or reforming a system of governance should aim to limit its fragility to some extent. Any governance regime worth having is probably worth making nonfragile.

Polycentricity means that decision-making units are independently constituted. ^{Footnote 19} Polycentric systems are comprised of “fragmented” and “concurrent” authorities. ^{Footnote 20} The various organizations that develop, interpret, and enforce rules adjust their actions to one another rather than to some central authority. In a polycentric system, there is no final authority to which all other authorities submit.

Polycentricity should be distinguished from several similar but distinct concepts. Many organizations subdivide functions into manageable components for strictly administrative purposes, but reserve the authority to direct, override, or disband those lower level units at will. For instance, the marketing department in a firm is not independently constituted and does not make the firm polycentric. ^{Footnote 21} It is not enough that a lower level authority implement or even determine policy if it is a mere appendage of a larger body.

Polycentic governance thus requires overlapping jurisdictions. Individuals fall under the jurisdiction of several authorities, with no one authority being reducible to the others. Vincent Ostrom identifies this condition with the distinction between a federal system and a “mere” confederation. ^{Footnote 22} In a confederation the individual's relationship to a supranational government is mediated through his relationship to the national government. To the extent that this is true, the larger governmental unit provides no genuinely alternative avenue for the individual to pursue his ends, and thus provides no check on the smaller unit. Overlapping jurisdictions also distinguish polycentricity from “balkanization,” in which the governing units are wholly exclusive entities and those subject to them have less scope for developing and acting on shared interests. ^{Footnote 23} Avoiding the twin perils of balkanization and centralization means that governing jurisdictions should be independently constituted but related to one another through formal rules and informal norms. ^{Footnote 24}

The other key ingredient for degeneracy is institutional diversity. Diversity connects to degeneracy in a straightforward and obvious manner. Independently constituted but identically organized jurisdictions would be redundant but not degenerate. Institutional diversity enables evolvability by allowing for experimentation, learning, and customization of rules to local conditions. ^{Footnote 25}

The difficulties and complexities of governance make error inevitable. Errors in governance come in a variety of forms, but essentially involve the enforcement of rules that create perverse incentives. These rules may themselves be the result of opportunism by those making or enforcing them, or they may come about more innocuously through bad design. ^{Footnote 26}

Complex systems defy simplistic predictive models. Especially in a cosmopolitan world of frequent international interactions, the idea that governing bodies or academics can design effective rules without substantial risks of failure is flatly implausible. Diversity enables (though of course it does not guarantee) error detection and correction, as individuals can observe how different sets of rules facilitate different types of social interaction. The most reliable way of learning about better rules is to try out different types of rules. Elinor Ostrom puts forward a strong case for experimentation:

The other key benefit of institutional diversity is the fact of local diversity. Different areas exhibit different ecological, geographical, and cultural traits. Similarly, different domains of governance — property law, security, environmental protection, and so on — also exhibit great heterogeneity in their scope for conflict and cooperation. If we mentally “multiply” domains of governance by the time- and place-specific characteristics of different geographical areas, we glimpse the manifold and daunting variation in the problems confronting institutional design. Even absent the possibility of error, well-designed rules would exhibit substantial variation. Elinor Ostrom again draws the relevant conclusion:

Polycentricity without diversity lacks evolvability. Diversity without polycentricity lacks robustness. Consider two — admittedly unrealistic — limit cases. If all jurisdictions were perfect substitutes, then they would exhibit redundancy and thus robustness. Such a regime would exhibit the qualities of standard economic models of static competition, and could be imagined as a world of cookie cutter municipalities or city-states competing for citizens who are free to move from one to another. Now consider the alternative case in which all jurisdictions are sui generis and total, exhibiting a wide range of diversity. They are interdependent, perhaps relying on one another’s exports. But movement between jurisdictions — whether for political or geographic reasons — is impossible. The system is perfectly balkanized. Such a regime would be highly evolvable, with different jurisdictions responding variously to any given change in circumstances. But it would not be robust. Any given jurisdiction, while capable of developing new capacities, will make some governance errors. If other jurisdictions are unable to compensate for these errors, the regime will gradually lose functionality as errors compound.

Polycentricity and diversity unleash the forces of dynamic competition, forces that are just as important to the provision of governance as for any other services. To take advantage of that competition, individuals require robust exit rights from jurisdictions. Access to alternative governance arrangements offers few benefits if jurisdictions do not have to compete with one another. The right of exit is a critical ingredient for robust and evolvable governance regimes, and that right can come in diverse forms. Migration or “foot voting” is the most intuitive example. This implication of degeneracy fits well with most brands of cosmopolitanism, which maintain a strong commitment to freedom of movement. Indeed, I argue in Section VI below that rights to migrate are the most important contemporary policy implication of degeneracy.

But many forms of exit allow individuals to geographically stay put. Jurisdictions can vary not only territorially, but also in the bundle of governance services they provide, creating more scope for them to overlap. A jurisdiction governing water usage need not also provide police services, and they may both operate far more effectively if they have different boundaries. Effective institutional design requires not only determining which governance services are needed, but also evaluating the bundling of governance services. ^{Footnote 29}

Chandran Kukathas makes the important point that all associations are partial associations. ^{Footnote 30} Governance providers are no different. If jurisdictions are to be beneficent and effective, their judicial and spatial boundaries should be sensitive to a host of factors: biological, geographical, technological, and cultural factors, and potential relationships to other jurisdictions. Individuals demand governance services in order to better pursue their ends or to secure or expand their rights. But because both the form and substance of projects and rights will vary even for a single individual, governance services — in order to fulfill their manifold functions — can and should take a wide variety of forms.

Degeneracy as a design principle thus tends to commend political fragmentation while casting centralized state authority in a more critical light. ^{Footnote 31} Vincent Ostrom consistently contrasts Hobbesian sovereigns with polycentric systems of fragmented authority that are genuinely rule-governed. ^{Footnote 32} Taleb likewise notes the tensions between nation-states as a system of governance and the goals of robustness and evolvability, arguing that nation-states are a fragile social system. Centralized power tends to concentrate rather than distribute the negative consequences of governance errors and unexpected shocks. ^{Footnote 33} As a result, although conflicts may be less frequent than with smaller units of governance, when conflagrations do occur they destabilize large areas and can lead to devastating consequences.

One weakness of centralized or final authority is that it tends to correlate and thus compound governance errors. Recall Elinor Ostrom's warning noted above: in complex social environments, errors of institutional design are inevitable. Both coping with and correcting such errors are vastly more likely in a polycentric system of diverse arrangements. A key insight of complex systems theory is that the correlated errors that affect every element of the system do the most harm. To the extent that errors are correlated, other elements of the system are not taking compensatory action. But even worse, such errors are likely to compound over time since they are not contrasted with alternatives. If different jurisdictions make different errors, it increases the probability that errors are detected and corrected. A lack of diversity and thus experimentation stifles social learning in precisely the social sphere in which it is most important to get the answers right: the rules by which we relate to one another. All of this is made yet worse by the fact that nation-states govern very large areas of diverse geographical and cultural characteristics, and modify rules relating to all sorts of social domains.

This essay advances no claim that these considerations are sufficient to call for a radical revision to existing forms of government. As noted above, degeneracy is not the sole sound principle of institutional design. Other considerations, such as the ability of institutions to serve as focal points of coordination, also bear on judging the appropriate level of centralization. Furthermore, centralization may have some corresponding advantages, such as the incentive of elites to quell much internal violence. ^{Footnote 34} Finally, historically existing nation-states are not idealized models of centralization. We must not confuse the abstract model of a sovereign state with institutional reality. Real world political regimes are already polycentric and thus degenerate to varying degrees. Anti-fragility, recall, is a matter of kind and of degree. Nonetheless, the next section argues that degeneracy can be particularly relevant for analyzing institutions of international governance.

The most obvious objection to placing a high value on polycentricity and diversity is that differing jurisdictions may come into conflict with one another. Other thinkers sympathetic to polycentric governance in international affairs have offered a number of convincing answers to this objection. Carmen Pavel offers a strong defense of the power of institutional pluralism to offer effective international governance. ^{Footnote 35} Among several key points she makes is that any nondespotic governance arrangement relies on norms that govern interaction that lack a final enforcer. Insisting on a final authority confronts an infinite regress problem: How is the authority to be held to account? The relevant question is not whether there is an ultimate guarantee against conflict, but which sort of governance regime makes conflict least likely. Allen Buchanan offers an important point in this regard, noting that independent but related governing bodies can offer reciprocal legitimation: the Supreme Court of the United States enhances the legitimacy of the presidency by checking its abuses, and vice versa. This legitimation serves an important sociological function: it lowers the cost to authorities of securing compliance with their rulings. ^{Footnote 36} I would add that, in so doing, it gives authorities an incentive to generate and adhere to norms of peaceful conflict resolution with other (potentially legitimating) jurisdictions.

Moreover, a sanguine analysis of human affairs recognizes that some amount of conflict between jurisdictions is inevitable. Given the inevitability of some conflict, it bears asking how different governance regimes respond to and recover from such conflict. Politically fragmented orders have several advantages in this regard. First, exit options for escaping from conflicts are greater. ^{Footnote 37} And second, since rulers need subjects to prevail in such conflicts, they are more likely to offer increased economic and political liberties to subjects. Several scholars have argued that Europe’s preindustrial political fragmentation made it more resilient to negative shocks than were other societies and paved the way for its modern success. ^{Footnote 38}

Moving beyond general objections to this approach, several more-substantive objections may be raised to the specific application of these ideas to the realm of international governance. One objection is that, insofar as the above analysis is true, it is irrelevant because the international order is already largely characterized by degeneracy between jurisdictions. A second is that, even if degeneracy is important, the view of governance outlined thus far is incompatible with cosmopolitan ideals. How is degeneracy not a recipe for isolationism or rampant injustice? A third objection is that attempts to substantially implement degenerate institutions are infeasible. The remainder of this essay addresses each of these objections in turn.

IV. Do We Already Have Degeneracy?

One potential criticism of degenerate cosmopolitanism is that it is irrelevant. We already live in a world with polycentric and diverse jurisdictions governing international affairs. Allen Buchanan, for instance, defends a great deal of contemporary practice in international law precisely on the grounds that it arises from multiple sources of authority. ^{Footnote 39} One might also point to the very real progress that has characterized the past few decades, in which liberalization and globalization have lifted billions out of poverty. These gains should not be undersold.

But current international governance arrangements nonetheless remain fragile in important respects. Recall Taleb’s point that the fragility of a system cannot be ascertained simply by examining its past performance. Decades of low economic volatility preceded the “Great Recession” of 2007, and Europe had been at peace for many years prior to World War I. Detecting fragilities requires examining how the jurisdictions that provide international governance are constituted.

Organizations typically associated with international governance — the United Nations, World Trade Organization (WTO), International Monetary Fund, and so on — conform to Vincent Ostrom’s definition of a confederation. ^{Footnote 40} Their constituent members are nation-states. Citizens of those states who might appeal to those organizations do so through the medium of their respective national regimes. This is where the concept of degeneracy becomes especially helpful: it makes clear that organizations like the UN fail to constitute concurrent authorities. To the extent that individuals and organizations appeal to them through their national governments, nation-states act as a funnel or choke point through which international governance arrangements are made. So while states do not exercise monopolistic control over every aspect of citizens’ lives, they come closest to a pure monopoly when it comes to questions international governance. This introduces a great deal of fragility into international governance.

Confederation, to the extent that it creates an organization with any power at all, tends to exacerbate the problems of centralization. First, it further centralizes and correlates governance errors. Central organizations place pressure on member states to adopt more uniform standards and thus eliminate institutional diversity. Second, confederation offers only blunt instruments for the enforcement of its rulings. Since members of a confederation are collectives rather than individuals, sanctions — economic, political, or military — cannot help but cause indiscriminate harm. ^{Footnote 41} Consider the example of the Agreement on Trade-Related Aspects of Intellectual Property Rights (or TRIPS), adopted by the WTO in 1994.

Prior to TRIPS, international law typically held that local intellectual property (IP) law be extended to foreigners doing business in a country. This arrangement facilitated trade while allowing for wide variation in IP law from one country to another. In fact, IP was rarely seen as a trade issue until a concerted push by various lobbies in the United States in the 1980s. The International Intellectual Property Association and related groups eventually turned the Office of the U.S. Trade Representative into an IP enforcement agency, and the United States began wielding trade policy as a means of convincing other nations to enforce the IP claims of U.S. firms. Following this success, a smaller portion of the same group — the Intellectual Property Committee, comprised of just twelve firms — essentially wrote the TRIPS rules. ^{Footnote 42} These rules make trade sanctions a generally acceptable response against countries that fail to enforce a certain set of IP laws, eliminating a great deal of institutional diversity. This is exactly the sort of bundling of governance functions that a degenerate approach is suspicious of, and reflects the dual problems of confederation noted above.

As a result of its origins, TRIPS reflects the ideas and interests shaping U.S. IP law in the years leading to its adoption. Not coincidentally, this period saw a cluster of governance errors regarding intellectual property. Not only was IP made into a trade issue, much legislation during this period was motivated by an economically illiterate fixation on “competitiveness” that foresaw the “deindustrialization” of the U.S. economy as a looming threat. ^{Footnote 43} Simultaneously, the strengthening of both patent law and its enforcement unleashed a tidal wave of rent seeking by nonproductive IP holders. ^{Footnote 44} In the United States, patent acquisitions are often used to initiate or ward off legal challenges rather than to alter production processes. ^{Footnote 45}

With TRIPS, the WTO centralized these governance errors on a global scale, and promotes their enforcement with the indiscriminate bludgeon of trade sanctions. The results are well known and should be unsurprising. Despite the rapid expansion of IP law and enforcement into less developed countries over the past twenty years, there is no evidence of increased research into the diseases that tend to afflict these countries (including after the TRIPS reforms in the early 2000s). ^{Footnote 46} This outcome is often lamented, and it is easy to recognize that good IP policies for poor countries are probably not the same as those for wealthier countries.

But there is a deeper point here: no one has any solid evidence about what a beneficial IP regime looks like under any circumstances, much less how it should vary across the many domains it might govern. There is no evidence that existing IP law increases innovation; studies that claim to support IP only show that it increases the number of patents or copyrights. ^{Footnote 47} Proponents of IP and TRIPS rightly recognize that the incentives for innovation matter. Indeed, it would be odd if there were no governance arrangements that could increase innovation beyond that of a pure intellectual commons. But those arrangements will not be discovered as long as institutional diversity is stifled by the current international trade regime. This is especially true when it comes to fields in which technological conditions are rapidly changing. Any IP law is shooting at a moving target, while TRIPS insists that everyone aim along the same trajectory. The current system has proven woefully vulnerable to political capture, and as a result represents a potentially disastrous centralization of error.

Confederation engenders fragility by promoting the suppression of institutional diversity while failing to offer individuals an alternative avenue for seeking governance services. Even if states serve useful functions, it does not follow that they should be either the alpha or the omega of international governance. Increasing diversity in international governance, by contrast, would require creating wide scope for circumventing national governments rather than merely adding layers to them.

A maximally degenerate system of governance would be open to both the formation of new governance associations and the dissolution of existing ones, as well as allowing free movement between them. The degenerate ideal, then, is quite far from a modified Westphalian world with smaller states. This raises a host of hoary questions about secession and dissolution that are far beyond the scope of this essay; the framework presented here is too thin to settle the difficult questions of territorial claims and obligations generated by existing jurisdictional arrangements. ^{Footnote 48} As noted above, my purpose is not to claim that degeneracy trumps all other principles of institutional design. But it does merit weighty consideration, and there are intermediate forms of degeneracy that are less radical. As noted above, political fragmentation need not take the form of breaking up existing nations. Indeed, the degenerate approach denies the idea that smaller units of governance are always better. What matters is that jurisdictions are overlapping and competitive.

Degeneracy requires that individuals have alternative pathways open to them for seeking governance services. Degenerate cosmopolitanism does not reject the need for rules to govern interactions between members of different societies, but rather rejects that such interactions should always occur in the shadow of the nation-state. Vincent Ostrom alludes to this idea in several brief passages applying polycentricity to problems of international governance, ^{Footnote 49} arguing that a self-governing international order requires “the opening of societies to diverse interpersonal and organizational relationships.” ^{Footnote 50} Ostrom imagines a world in which individuals and communities can enter into governance arrangements with one another unmediated by existing nation-states. It would be more accurate to refer to degenerate cosmopolitanism as a theory of transnational than international governance, a world of covenanting without borders.

This is all rather abstract, one might object. How might a degenerate system of international IP law look? Part of the point is that we cannot know what it would look like through armchair theorizing, but we can identify margins on which a regime could be more or less degenerate. Imagine a world of competing IP standards. Some apply more broadly, some to only specific types of goods like pharmaceuticals. Some standards are bundled with trade rules, others with public health, and some with no other laws at all. They offer varying degrees of protection and enforcement. And rather than whole nations falling under these rules, their members might be individual firms, trade associations, industrial parks, and distribution networks. On the margin, each of the above characteristics increases the ability of individuals to seek alternative governance arrangements. This sort of system would allow for a great deal of experimentation with IP law, with competition between competing standards helping to illuminate and weed out less effective forms of governance.

This vision may seem unrealistic, but it is exactly the sort of process that created the lex mercatoria that governed international trade for centuries. ^{Footnote 51} It is precisely because they had to solve the problems of competing local laws, competing standards, and the opportunism that plagues all systems of governance that merchants were able to develop effective rules — but not centrally imposed ones — that facilitated long distance exchange. Even today, private arbitration dominates the governance of international trade. ^{Footnote 52} Other governance problems involving common pool resources or public goods may be more difficult, but — as I argue below — the benefits of degeneracy extend to those problems as well.

Finally, it is worth noting that, even to the extent that degeneracy obtains in the global order, it is still a useful concept for evaluating proposals for new forms of governance. Few cosmopolitans today advocate for a world-state, but the framework elucidated here provides a compelling account of why such a move would be extremely dangerous by correlating governance errors and inducing fragility. Centralized states are not the only source of fragility, however. A monopoly over some governance function eliminates degeneracy in that policy domain. This is the problem with proposals for worldwide governance bodies that fall short of a world state, such as a body that would govern the environment or trade. While such bodies would not induce fragility on the level that a world state would, they would nonetheless centralize and compound governance errors within their particular domain. An institutional regime will be evolvable and adaptable over a range of various domains according to the level of degeneracy between jurisdictions within each domain. Similarly, degenerate cosmopolitanism offers an important critique of calls to synchronize laws across countries. Even without a central body initiating such changes, attempts to homogenize rules can stifle institutional diversity. The drive to eliminate bad laws should not be confused with the attempt eliminate differences between jurisdictions.

V. Can Degeneracy Promote Cosmopolitan Ideals?

Degeneracy resonates well with the original sense of cosmopolitanism from the Cynic Diogenes, for whom the word “cosmopolitan” was primarily a negation of any special obligation to any particular polis. ^{Footnote 53} Individuals and communities should enter and exit from a variety of governance arrangements — including those that reach across traditional borders — according to their own values and interests. Polycentricity fits comfortably with the idea of not owing one’s allegiance solely to any one government.

Some cosmopolitan theories should thus find degeneracy intrinsically appealing. The ideal of degenerate cosmopolitanism most closely maps onto Chandran Kukathas's Liberal Archipelago. ^{Footnote 54} Kukathas argues that the proper metaphor for a liberal society is an archipelago of smaller societies. He recognizes the importance of the independent constitution of associations, and argues that his vision breaks down the distinction between domestic and international society. ^{Footnote 55} For Vincent Ostrom, polycentricity is coterminous with the concepts of the rule of law and self-government: if, in the last resort, there is a final authority issuing, interpreting, and enforcing rules, then governance becomes an exercise in mere command. ^{Footnote 56} He argues that polycentricity is the true “meaning of democracy,” without which governance is left to the mere accident of force and history. ^{Footnote 57} For individuals attracted to these and similar ways of thinking, degenerate cosmopolitanism commends itself.

To thinkers who do not share similar moral or political ideals, degenerate cosmopolitanism may seem like a betrayal of the concept of cosmopolitanism since it advocates strong institutional diversity that can conflict with moral universalism. Cosmopolitan ideals are often rooted in a liberal conception of human rights, which aim to protect the welfare and liberty of individuals. ^{Footnote 58} But moral norms such as human rights are usually formulated in an abstract fashion, and can be realized in different ways. There is no prima facia reason to believe that universal adherence to moral norms is incompatible with institutional diversity. Degeneracy enables experimentation with different institutional means of realizing deeper normative aims, incentivizes jurisdictions to realize those aims, and provides an exit option for those who live in unacceptably bad regimes.

Degeneracy, recall, is valuable in complex systems precisely because it both avoids fragility on the downside and enables evolutionary learning on the upside. In what follows I argue that degenerate jurisdictions promote particular forms of robustness and evolvability: they tend to protect (on the downside) liberty and welfare in bad conditions and tend to promote (on the upside) increases in liberty and welfare.

By “liberty” I simply mean the sorts of procedural rights that liberals tend to value, such as property rights, freedom to trade, political and civil liberties, equal protection under the law, and so on. “Welfare” I interpret in the broadest possible sense, but focus on economic well-being and efficiency, simply because economic well-being is robustly correlated with other measures of welfare. I avoid making any claims at all regarding the moral foundations, metaethical status, or proper weighting of liberty and welfare. I simply assume that, in general, more of each is preferred to less.

Moreover, I am out to establish only comparative and functional relationships. Many jurisdictions will get it wrong. But an errorless system is not on the table. Polycentric and diverse institutional arrangements generally do a better job promoting liberty and welfare than do monopolized arrangements. Degenerate jurisdictions have historically been fertile soil for the advance of liberty and welfare, exhibiting evolvability. And they do a better job protecting individuals against gross violations of liberty and welfare, exhibiting robustness.

VI. Conclusion: Is Implementing Degeneracy Feasible?

Degenerate cosmopolitanism takes seriously the hard problems of aligning incentives and coping with human ignorance. Degeneracy commends a system of polycentric and diverse governance arrangements, held together by robust exit rights to promote competition and experimentation. Such systems are less vulnerable to negative shocks and capable of generating new and better forms of governance than more centralized systems, and there is ample evidence that degenerate governance promotes both more liberty and greater welfare.

But all of these arguments are for naught if implementing degenerate governance is implausible. I consider two versions of this objection here. The first is that degenerate cosmopolitanism assumes an unrealistic moral psychology. A polycentric order requires that rules and norms guide interactions between independent centers of authority without particular jurisdictions exercising dominance. Inattentiveness to these rules tends to result in either balkanization or centralization, both of which sacrifice the benefits of degeneracy.

Maintaining such rules, Vincent Ostrom properly notes, requires citizens who are willing to take up the burdens of self-governance and practice the Tocquevillian art of association. ^{Footnote 76} Institutional craftsmanship is necessary in forging particular rules and in negotiating relationships between jurisdictions. This may seem like a heavy burden, but the Bloomington School did not begin with idealized citizens. Rather, the analysis of polycentric and diverse institutions began with the empirical, comparative investigation of real world governance systems. Such practices have existed all throughout history in all corners of the world.

The relevant challenge is not whether this model can function, but whether it can function across borders and whether it will be allowed to grow. Degenerate cosmopolitanism views universal citizenship not as an abstract moral ideal but as a willingness to search out better rules on which to ground social cooperation among diverse individuals and groups. Calling for degeneracy does not involve abstaining from judgments of institutions, but views improving them as an ongoing process of cooperative artisanship that can only be effected from the inside out. We should be sanguine that the threat of such conflict is always present; polycentric orders are different not in that the potential for conflict exists but in that it is diffuse rather than concentrated. The alternative to polycentric governance, Ostrom makes clear, is not enlightened despotism but accident and force, driven by either populist or elitist whims.

The second, more salient objection to the foregoing analysis is that degeneracy is itself the outcome of bewilderingly complex interactions among a multitude of governance institutions and organizations. It is not a blueprint for reform. Degeneracy may be a blessing when it happens to coalesce, but itself provides no guidance since it is essentially an historical accident. This is especially true to the extent that degenerate cosmopolitanism itself eschews top-down reforms.

This argument is compelling, but it goes too far. Anti-fragility as a systemic property cannot be designed, but degeneracy between jurisdictions can be intentionally cultivated. There are policies and reforms that individual jurisdictions can adopt that can increase the degeneracy of international governance arrangements even if other jurisdictions do not adopt them.

The general policy stance that a degenerate cosmopolitan can advocate is one of “opening doors”: seeking marginal additions to exit rights from one jurisdiction to another. Degeneracy and polycentricity are matters of degree. More robust exit rights allow for more experiments in governing and more competition between jurisdictions. Exit rights do not have to come all at once to produce good consequences. Increased jurisdictional competition puts pressure on existing jurisdictions that can lead to reform; but even if it does not, at least those who are allowed to exit are likely made better off. Marginal additions of exit rights allow for gradual reform and experimentation without revolutionary changes in existing governance arrangements.

Concretely, the most important single policy issue from a degenerate cosmopolitan perspective must be international migration, especially migration from less to more developed nations. Freedom of movement and freedom of association are vital components of both economic and political liberty. They allow immigrants to directly improve their own lives, as well as to aid family and friends back home through remittances. Moreover, they do this by generating wealth. Economists' estimates of the gains from completely free migration cluster around 100 percent of global GDP. ^{Footnote 77} Worldwide production could probably double if there were no restrictions on free exit. And this probably underestimates the gains, because these models take the existing quality of governance as a given. As noted above, the ability to exit a regime puts pressure on rulers that tends to result in more-liberal governance.

Outright open immigration strikes people as radical. This is precisely why marginal additions to exit rights are an attractive approach. Each incremental step to liberalize immigration controls at least allows for some gains, and helps address legitimate countervailing concerns. This is especially true when we recognize that exit rights have different dimensions. Consider two possible reforms. First, a jurisdiction might require immigrants to pay an immigration tax roughly equal to the expected cost of providing him or her with public services (including possibly funding retraining programs for displaced workers). ^{Footnote 78} This policy addresses the concerns of those who worry about immigrants being a fiscal drag. It also allows those who wish to improve the lives of potential immigrants to do so, by paying the tax for them. And it secures at least some incremental exit rights, enhancing degeneracy.

Second, priority may be given to some incremental exit rights over others according to other normative principles. A jurisdiction might create special immigration allowances for individuals in particularly poor countries. Such a move should appear to be an attractive first step for those who are concerned with distributive justice, who may worry that freer migration would be used by the rich to avoid paying their fair share in taxes. But allowing individuals to migrate from desperately poor countries would have far more effect than any welfare program. By far the largest windfall that the world's truly poor can reasonably expect is the opportunity to participate in an advanced economy. This would have the added benefit of putting pressure on the rulers of precisely those nations most in need of more liberal governance.

Two related proposals address more fundamental problems of governance: charter cities and startup cities. ^{Footnote 79} The idea of both reform efforts is to create semi-autonomous municipalities that offer bundles of public services and — much more importantly — governance arrangements to residents. Modeled on special economic zones or districts, both proposals aim at creating a space in which the institutions that primarily drive economic and social development can be changed all at once in a small area without asking existing governments to surrender their control. Again, even if the creation of such jurisdictions does not set off a chain reaction of greater political competition and innovation, at least those who are able to opt into such arrangements might find better lives.

But perhaps more importantly, degenerate cosmopolitans should cast a suspicious eye on moves towards greater centralization of international governance. Trying to regularize practices or create uniform policies across nations eliminates the crucial forms of diversity that are the engine of social learning. Enforcement of such policies also tends to create entrenched interests with little patience for reform. Standardization can sometimes be beneficial but should always be approached with caution, with avenues for robust challenges to any such standards carved out in advance. Governance is a complicated and messy affair even at the most local of levels. Anyone serious about promoting beneficent interactions between individuals of radically different cultures and histories should adopt an attitude of humility about the best way to govern those interactions. Degenerate cosmopolitanism idealizes the discovery, not the imposition, of new ways for individuals with diverse backgrounds and cultures to live better together.