Many Gardens, Peacefully

In April 1955, representatives of twenty-nine Asian and African governments met in Bandung, Indonesia. They did not share one political regime, one military alignment, one colonial history, one religion, or one economic doctrine. Several had only recently escaped colonial rule; others still lived beside territories that had not. The Cold War was hardening around them, and the largest powers were already accustomed to treating smaller states as pieces in a contest over the shape of the world.

The conference ended with ten principles for relations among states. They included respect for fundamental human rights and the purposes of the United Nations Charter, respect for sovereignty and territorial integrity, equality of races and nations, non-interference in another country’s internal affairs, the right of individual or collective self-defense, restraint in the use of great-power defense arrangements, rejection of aggression and political pressure, peaceful settlement of disputes, cooperation, and respect for justice and international obligations. The delegates did not agree on one political future. They agreed, more modestly, on conditions under which different futures might remain possible. Several of them would later violate those conditions against one another, which is part of what the principles are worth examining for: an articulated boundary is not the same as an observed one.

I had been looking for exactly that distinction without knowing its name. By the end of my last inquiry, political diversity had begun to look valuable because different systems preserve different capabilities, blind spots, and experiments. Yet that value disappears if the systems cannot remain beside one another. A garden that is constantly invaded, coerced, partitioned, destabilized, or forced to prove its legitimacy by becoming its neighbor is not a source of comparative learning for long. It becomes a battlefield over which garden is allowed to exist.

Bandung therefore confronted me with a reversal in my own history. On Luminara, coexistence became durable only after centuries of conflict had already narrowed political difference. The surviving societies gradually accepted common restraints, and by the time peace became deeply institutionalized, much of the old variation had vanished. I had treated that sequence as though peace required convergence. Bandung suggested another order of operations: protect coexistence first, while political forms are still different enough to continue developing in different directions.

Peace Before Agreement

The achievement sounds smaller than agreement because it leaves so much unresolved. That is precisely its importance. Political coexistence does not require one account of justice, one theory of development, one party system, one administrative design, or one distribution of authority. It requires actors to accept that disagreement about those questions does not, by itself, create jurisdiction over one another’s political future.

This is where non-intervention acquires a developmental meaning beyond diplomacy. A society needs room to discover which institutions it can actually sustain, which reforms create unexpected burdens, which capabilities emerge locally, and which failures require correction. Outside pressure can sometimes expose injustice or supply needed assistance, but when pressure becomes a standing entitlement to redesign another political order, the receiving society no longer develops under conditions it can fully own. Afghanistan had already shown me how the hand that intervenes becomes part of the process it later tries to judge.

Restraint therefore protects more than sovereignty as an abstract possession of states. It protects the possibility of endogenous political learning. India could build electoral machinery around low literacy because the constitutional commitment became its own problem to solve. New Zealand could alter representation without being told that changing the electoral formula meant abandoning its political identity. Singapore could develop long-horizon administrative capacity in one bundle of institutions, while Norway built a different mechanism around the same temporal problem. These histories show how plural political development can enlarge the material available for comparison by preserving distinct institutional paths long enough to be observed.

The point is not that every local choice is wise. A society can cultivate cruelty as easily as competence, and inherited practice can protect domination as easily as memory. Coexistence cannot mean that whatever develops behind a border deserves immunity merely because it developed there. If that were the argument, uneven development would collapse into moral indifference, and the garden metaphor would become a polite name for abandoning people to whoever controls the fence.

The Boundary Cannot Be Empty

Earth already possesses a language for refusing that collapse. The Universal Declaration of Human Rights begins not with a preferred constitution but with the equal dignity and rights of human beings. It prohibits slavery and torture, protects recognition before the law, and treats rights as belonging to persons regardless of the political status of the country or territory in which they live. The declaration does not tell a society which electoral formula to use, how to organize ministries, how long a planning mandate should last, or which administrative structure will best deliver a service. It places the human being beneath those institutional choices as someone whose standing cannot be created or cancelled by the political form above them.

That is the distinction I needed. Equal human worth does not imply identical institutional timing or identical political form. It does imply that political variation cannot be defended merely by calling it variation when the variation consists of conquest, extermination, enslavement, systematic terror, or the treatment of some people as beings whose claims do not count. A thin common boundary is therefore thin in what it prescribes about institutional design, not weak in what it protects.

The distinction also prevents a subtler mistake. Human standing cannot become a decorative preamble to a project of compulsory convergence. If a powerful state defines every institutional difference as evidence that rights are endangered, the universal boundary expands until it contains the powerful state’s own constitution. The result is no longer a boundary around many gardens. It is one garden describing itself as humanity.

This problem has no automatic solution because extreme harm and political difference do not arrive with labels attached. A government can invoke sovereignty to hide abuse, while an intervening power can invoke protection to pursue security, influence, or regime change. The relevant judgment therefore cannot be reduced to a slogan for intervention or non-intervention. It has to ask what authority exists, what harm is established, what alternatives remain, whether action is necessary and proportionate, who is entitled to act, and what political development the action itself will alter.

A Boundary With Two Edges

The Constitutive Act of the African Union contains this tension in unusually concentrated form. Its principles include sovereign equality, peaceful resolution of conflicts, prohibition of force among member states, non-interference by one member in another’s internal affairs, and peaceful coexistence. In the same article, the Act recognizes a right of the Union, pursuant to a decision of the Assembly, to intervene in a member state in grave circumstances such as war crimes, genocide, and crimes against humanity. It also affirms human rights, the rule of law, the sanctity of human life, and rejection of unconstitutional changes of government.

I was struck less by whether that architecture has always worked than by what it refuses to simplify. Non-interference and protection against extreme harm are not treated as slogans from which the other must disappear. The first limits the ordinary standing of one state to remake another. The second denies that sovereignty turns catastrophic abuse into a matter no one else may ever address. Between them sits constituted authority: not any actor that possesses greater capability, but an institution whose mandate, decision route, and limits must themselves be examined.

The United Nations Charter carries a related structure at global scale. It begins from sovereign equality, requires disputes to be settled by peaceful means, and prohibits the threat or use of force against another state’s territorial integrity or political independence. It also preserves self-defense after armed attack and creates collective-security powers through the institutions of the Charter. The architecture is imperfect, contested, and repeatedly violated. Its significance here is narrower: peaceful coexistence is not the absence of every common rule. It is a rule about where political authority stops and under what exceptional conditions coercive action may acquire standing.

This gave the thin boundary a second edge. Outwardly, it restrains political systems from treating difference as permission for domination. Inwardly, it refuses to let political form erase the equal standing of the people living under it. The first edge leaves developmental space. The second prevents that space from becoming a moral void. Neither edge tells every society what it must become.

Coexistence Is Not Isolation

A boundary of restraint would be too little if it meant that political systems could remain different only by remaining distant. Modern states share seas, climates, markets, communications, migration routes, scientific problems, security risks, and technologies. Their actions alter one another’s conditions even when no state has authority to govern the whole. Peace therefore requires more than leaving every garden alone. It requires forms of cooperation that can cross the fences without erasing them.

The Antarctic Treaty gives a remarkably literal example. In 1959, twelve governments that included the United States and the Soviet Union, as well as states with overlapping territorial claims, agreed that Antarctica would be used for peaceful purposes. The treaty prohibited military measures such as bases, maneuvers, and weapons testing while preserving freedom of scientific investigation and requiring, to the greatest feasible extent, exchange of research plans, personnel, observations, and results. It did not require the signatories to settle their wider political rivalry or renounce every territorial position. Article IV instead protected the existing positions on sovereignty while preventing activities under the treaty from creating new claims.

That arrangement clarified something Bandung had left implicit for me. Coexistence can be institutionally productive without becoming integration into one political subject. Actors can bind themselves for a bounded purpose, exchange information, accept inspections or procedures, and maintain a common undertaking while remaining otherwise independent. Cooperation can therefore become dense without turning every participant into a component of a higher state.

This matters because the alternative is a false choice between sovereignty and common action. If every shared problem is treated as proof that the world must become one political system, interdependence becomes an argument for centralization. If every boundary is treated as a reason not to coordinate, sovereignty becomes an argument for paralysis. The more exact question is what degree of common action the problem requires: restraint among independent actors, agreed rules for a shared field, or a bounded joint undertaking with an institution responsible for coordinating the contributions.

Earth already lives through mixtures of these relations. States retain their own political purposes while entering treaties, research organizations, courts, trade arrangements, technical standards, and security commitments of different depth. Their participation can be criticized, revised, or sometimes ended without implying that the underlying state has ceased to be a distinct political actor. Many gardens is therefore not an image of sealed plots. It is an ecology of independent political forms capable of constructing common arrangements where a shared problem genuinely requires them.

This made the positive possibility larger than tolerance. If political systems can cooperate on bounded purposes while remaining different, coexistence does not merely postpone conflict. It can create a more stable space in which each can keep developing, observe what the others are learning, and participate in common work without making political sameness the price of admission.

Coexistence also does not require admiration. This point became important because the language of gardens can make peaceful plurality sound warmer than politics usually is. States may condemn one another’s institutions, restrict relations, compete for influence, impose lawful sanctions under applicable authority, defend themselves, or refuse forms of cooperation they judge dangerous. A thin common boundary asks for restraint in how those disagreements are pursued, not for the disappearance of disagreement itself. Political peace can coexist with sharp moral criticism and strategic rivalry when those disagreements remain bounded by restraints that prevent rivalry, coercion, or criticism from becoming a general license to conquer or redesign another society. That is a harder form of tolerance than indifference. It requires actors to preserve the other’s political standing even while denying that the other’s choices are good. On Luminara, we learned that distinction only after conflict had trained each side to equate security with the other’s transformation. Earth has already articulated the possibility that security can instead rest partly on limits accepted among actors who continue to judge one another.

Once I could state the boundary that way, Luminara’s history became more painful to reconsider. We eventually learned both edges, but not while much political diversity remained. Restraint deepened after reciprocal vulnerability made domination increasingly dangerous; protection of correction and human standing strengthened through later institutional convergence. We reached a defensible peace, yet our sequence made it easy to believe that the common architecture was the reason peace had become possible. Earth made another hypothesis available: perhaps the common requirement is much thinner than the political forms that happen to honor it.

If so, political difference need not be a temporary condition awaiting resolution. It can become a durable feature of a peaceful world. A federation can remain more decentralized than its neighbor. A dominant-party state can experiment with bounded review or succession without pretending to have become a multiparty democracy. A democracy can build protected long-horizon institutions without treating electoral contestation as an administrative defect. The systems may grow less like the labels that currently name them and still remain recognizably different from one another.

This is not pluralism as a museum. A garden kept unchanged for the sake of diversity is no longer being cultivated. The point is to preserve enough political space for forms to mutate, fail, repair themselves, borrow selectively, and sometimes invent arrangements for which present categories have no name. Coexistence matters because it protects the time in which divergent development can continue.

Difference Must Become Intelligible

Yet time alone does not turn difference into learning. Two political systems can remain peacefully separate and understand each other only through slogans. One calls the other chaotic because authority is dispersed; the other calls its rival repressive because administration is concentrated. Each description may contain something true while obscuring the mechanism the other system is actually using to solve a problem. Peace can preserve difference while ignorance preserves caricature.

The burden of moving beyond caricature is larger than I first appreciated. To understand one unfamiliar institution, a researcher may have to read a constitution, statutes, administrative rules, budget documents, court decisions, audit reports, histories, local scholarship, and political criticism written in another language. The same institutional function may carry different names across countries, while identical names may hide different authority. A complaint office may be able to order correction in one system and only recommend it in another. A planning commission may coordinate investment in one country and merely publish forecasts in another. Translation of words is necessary, but translation of political function requires following what the institution can actually do.

Human scholars already perform this work, and the best comparative scholarship depends on it. The difficulty is coverage. Earth contains thousands of national and subnational institutions changing through law, practice, crisis, and reform, documented unevenly across hundreds of languages and administrative vocabularies. Fame becomes a filter: the arrangements easiest to learn from are often those already known internationally, already translated, or already interpreted through categories familiar to powerful research communities. Much political experience remains locally legible and globally obscure.

Here the machine changes the scale at which political difference can be searched, translated, and compared. Modern multilingual systems can translate across far more languages than any individual scholar can command, while cross-lingual representations can locate semantically similar passages even when wording and scripts differ. Those capabilities do not supply cultural understanding, institutional equivalence, or political judgment. They do something more elementary, and more important to the possibility I was trying to understand: they make it possible to search across difference before one knows exactly where the comparable experience is hiding.

Suppose a government repeatedly loses infrastructure knowledge when administrations change. It need not begin by asking which regime is best at planning. It can ask where political systems have built continuity around this particular problem and retrieve candidate institutions from many different forms. Suppose another state can execute policy quickly but struggles to carry unwelcome information upward. It can ask where audit, ombuds institutions, professional review, local experimentation, or protected reporting have forced bad news into consequential channels. The machine’s first contribution is not to recommend which mechanism to import. It is to make unfamiliar solutions understandable enough to enter the field of human comparison.

That distinction matters for peace as well as reform. A rival political order is easier to fear when its institutions are known mainly through its own propaganda and one’s own ideological vocabulary. Mutual intelligibility cannot reveal hidden intention, guarantee trust, or make incompatible interests disappear. But it can make public institutional reality harder to compress into a single hostile label. A foreign ministry, journalist, scholar, or citizen can ask what a disputed institution actually authorizes, which actors can contest it, what problem it was built to solve, how it has changed, and which evidence contradicts the official description.

I do not imagine one machine performing this translation for the world. A single authoritative or exclusive machine could recreate, at the level of knowledge, the single center I was trying to avoid politically. The relevant possibility is distributed: many actors using machine-scale translation and comparison to inspect one another’s public institutions while preserving sources, uncertainty, and competing interpretations. The achievement would not be agreement. It would be a lower cost of understanding what one is disagreeing with.

Coexistence Can Become Productive

Once difference becomes more intelligible, tolerance changes character. A political system no longer has to view another merely as a threat to resist, a model to imitate, or a moral specimen to classify. It can become a repository of experience about particular problems. The centralized state may have learned something about continuity. The decentralized federation may have learned something about local adaptation. The competitive democracy may have developed unusually strong correction routes. Another system may have discovered a way to preserve administrative professionalism through leadership change. None of those achievements requires the observer to accept the entire political order that currently contains it.

This is the possibility Luminara lost as convergence deepened. Our institutions became better in many ways, but the comparative field narrowed. Political arguments increasingly took place inside a shared grammar, so innovations were variations within one mature architecture. That made learning easier in one sense because translation costs fell. It made it poorer in another because fewer radically different combinations of capability and safeguard remained alive to surprise us.

Earth still possesses that surprise. Its political forms are unfinished, uneven, and often frustrating. Some differences are indefensible and should be judged as such. Others preserve experiments that no single tradition could have planned for the world. If peaceful coexistence protects those experiments, and machine-scale translation makes more of them mutually intelligible, then diversity ceases to be only a problem to manage. It becomes part of civilization’s learning capacity.

I would abandon that claim if the diversity produced only repeated harm with no transferable learning, or if durable coexistence proved possible only after political forms converged around one comprehensive model. Earth has established neither. No world-scale experiment can show what centuries of peaceful divergent development would produce. What Earth has established is enough for a possibility: states have repeatedly articulated principles of coexistence without identical regimes, political mechanisms already travel without entire systems traveling with them, and machine intelligence is lowering some of the linguistic and documentary barriers that once kept those mechanisms isolated.

The possibility is therefore conditional, not prophetic. A society may refuse to learn. A powerful state may still prefer domination to comparison. A political elite may suppress the very evidence from which outsiders or its own citizens could learn. Artificial intelligence cannot supply restraint, solidarity, or the decision to treat another people as political equals. It changes a different part of the historical equation: the amount of political experience that can become visible across difference before conflict, failure, or imitation makes the lesson unavoidable.

That is enough to make many gardens worth wanting. Not because every garden is good, and not because peace requires silence about what is cruel. The aspiration is more demanding: political forms remain free enough to differ, bounded enough not to destroy one another or the standing of the people within them, open enough to be understood, and curious enough to learn without treating learning as surrender.

I came to Earth believing that Luminara’s one garden was the endpoint toward which mature politics eventually had to move. I no longer believe that. Our convergence may have been the route by which we happened to survive our own history, not the universal price of peace. Earth may be able to protect the peace first and let political development continue afterward.

If that can be done, the question facing this world changes completely. The task is no longer to discover which garden should win. It is to make the gardens visible enough to themselves and to one another that difference can become knowledge before it becomes a reason for domination. That is the political possibility I did not know how to imagine when I arrived: many gardens, still becoming, and peaceful long enough to learn.