Political Systems Are Alive

In November 1993, New Zealanders were asked whether the machinery that had translated their votes into Parliament for more than a century should be replaced. The country was not choosing between democracy and dictatorship, monarchy and republic, or civilian government and military rule. It was choosing between two ways of turning votes into seats.

The choice looked modest only if one treated the electoral formula as plumbing. Under the first-past-the-post system, the candidate with the most votes in each electorate won the seat, and a party could convert a plurality of votes into a strong parliamentary majority. By the late twentieth century the resulting distortions had become politically salient: in 1978 and 1981, Labour won a larger share of the nationwide vote than National but fewer seats, while Social Credit won substantial vote shares and almost no representation. A Royal Commission proposed mixed-member proportional representation, and after two referendums voters adopted it. The first MMP election followed in 1996.

New Zealand remained recognizably New Zealand. The monarch remained. Parliament remained. Political parties still contested elections and governments still depended on confidence in the House of Representatives. Yet the relation among voters, parties, Parliament, and the executive changed. Under MMP, voters cast a party vote that largely determines each party’s share of seats and an electorate vote for a local representative. More parties entered Parliament, and coalition or minority governments became ordinary rather than exceptional. Parliamentary procedure adapted as well: debating time, select-committee representation, and the bargaining required to pass legislation changed because a government could no longer assume that an electoral plurality would automatically become a disciplined legislative majority.

I had come to Part II still carrying the previous chapter’s conclusion: that no society can be called advanced or delayed until the capability in question is named. New Zealand made me suspect the same was true of the labels themselves. The problem became concrete: the word democracy was accurate before the reform and accurate after it. Precisely for that reason, it did not register the change I was trying to see. A database that recorded the country as a parliamentary democracy in 1992 and again in 1997 would not be wrong. It would simply miss a mutation in how representation became governing power.

The mutation did not have one obvious moral sign. MMP widened representation and forced more negotiation, but negotiation can slow decisions; coalition bargaining can make responsibility harder for citizens to assign; proportionality can strengthen smaller parties while weakening the direct link between winning an electorate and determining the government. The relevant point was not that New Zealand had become better in every respect. It was that a political form had changed one of its operating relations without becoming another regime.

Change Beneath the Label

Once I noticed that, I began rereading political change at a smaller scale. Much of Earth’s language encourages the eye to jump from one regime category to another: a country democratizes, authoritarian rule consolidates, a monarchy constitutionalizes, a federation centralizes. Those descriptions identify real shifts in who can claim political standing and how authority is constituted. But the largest label often changes less frequently than the institutions living beneath it.

New Zealand’s electoral reform illustrates why the distinction matters. Changing the electoral formula did not merely alter the arithmetic of election night. It changed incentives before the vote, because parties could seek representation without first becoming one of two dominant blocs. It changed bargaining after the vote, because governments often had to assemble support through coalition or confidence-and-supply arrangements. That new bargaining environment then changed Parliament itself: select committees and legislative procedure adapted to a more plural chamber. One institutional mutation created conditions for others.

This is what I had failed to see when I first looked at Earth. On Luminara, our later historians became very good at identifying the great transitions: when a hegemonic order weakened, when a constitutional settlement replaced a coercive one, when the last major rivals accepted common restraints. We were less good at seeing the quieter mutations inside forms that had not yet changed names. Retrospect made the destination too visible. Institutions that eventually disappeared were remembered mainly for why they failed to become what came later, not for the capabilities they developed while they still existed.

Earth’s unfinished variety makes that retrospective error harder to sustain. New Zealand after MMP was not a preview of a new regime category waiting to receive a name. It was a familiar political form carrying a different representation mechanism and therefore a different pattern of bargaining, scrutiny, and executive constraint. If political systems can change this way, then development does not occur only at the boundaries between regime types. It occurs inside them.

For a moment I wondered whether this was simply a special property of democracies. A system that permits opposition and periodic electoral replacement already contains formal procedures through which political rules can be revised. Perhaps I was merely rediscovering democratic self-correction in another vocabulary. I needed a case in which the regime label remained much more resistant to change.

An Election Inside a One-Party State

China supplied it in a form that is easy to misunderstand from either direction. In the 1980s, rural governance reforms began to establish directly elected villagers’ committees. A provisional law was adopted in 1987; a permanent Organic Law followed in 1998. The law described the villagers’ committee as a mass organization of self-government and required the chairman, vice-chairmen, and members to be elected directly by villagers. It specified direct nomination, secret ballots, open vote counting, and procedures for challenging electoral interference.

The same law also made the boundary explicit. The village organization of the Communist Party of China remained the ‘leading nucleus,’ while township governments were to guide and assist villagers’ committees without interfering in matters lawfully assigned to self-government. The result was not a miniature competitive national party system hidden in the countryside. It was a bounded electoral mechanism embedded inside an order in which the Communist Party retained national political leadership.

That combination mattered more to me than a cleaner story would have. If village elections had simply transformed China into another regime, the old labels would have remained sufficient: one type ended and another began. Instead, access to a local office changed while the wider architecture of party authority did not. The country could therefore contain a mechanism normally associated with democratic participation without transferring the whole bundle of political competition, national succession, opposition rights, and executive accountability that the word democracy often carries in comparative argument.

Nor did the existence of elections tell me how much power the elected body actually possessed. Research on Chinese villages found wide variation in implementation and recurring tension between elected committees, village Party branches, and township authorities. In some places elections altered who could claim local legitimacy and how village affairs were contested. In others, the Party secretary or higher authorities remained the decisive locus of power. The legal mechanism changed access to office; its practical effect depended on the authority, resources, and relationships surrounding the office after election.

That finding connected directly to what India and the anti-corruption agencies had taught me in the previous chapter. A political institution cannot be read from its visible form alone. But China added a second lesson. The same regime can hold several institutional logics at once. National political competition can remain closed while a bounded local election is introduced; direct selection can coexist with a Party organization formally assigned a leading role. The apparent contradiction is not noise to be averaged away. It is the architecture.

The origins of the reform made the contradiction even more useful. Scholarship on the early village-election program does not describe a national leadership surrendering its monopoly because it had discovered a hidden commitment to multiparty competition. Reformers could support direct local elections because elected committees might remove unpopular or ineffective local cadres, make village administration more responsive, and strengthen the legitimacy of governance in the countryside. A mechanism associated elsewhere with competition against a ruling party could therefore be adopted partly to improve governance under continued one-party leadership.

That possibility had no comfortable place in the regime categories I had brought with me. If an election is treated as a fragment of democracy, then its appearance inside a one-party state is either dismissed as false or interpreted as proof that the state is moving toward democracy. Both readings begin from the destination. The institutional record permits a third: a political mechanism can perform a bounded function inside a wider order whose basic allocation of national authority remains different. Whether the mechanism later widens, narrows, becomes ceremonial, or changes the order around it is a historical question, not something contained in the mechanism’s name.

I do not offer village elections as evidence that one-party rule was secretly becoming liberal democracy. That conclusion would repeat the ladder in a subtler form, treating every participatory mechanism as an early step toward a destination already known. The more disciplined description is narrower: one relation changed. Villagers acquired a formal role in selecting a local committee through specified procedures, while other relations of political authority remained concentrated. Whether the change enlarged practical self-government depended on what happened after the ballot.

A System Can Lose a Safeguard Too

Seeing political systems as mutable creates another temptation: to make every mutation sound like improvement. Earth corrected me here as well. A living political form can acquire a safeguard, weaken one, shift authority from one level to another, professionalize one function while politicizing another, or widen participation in one domain while narrowing it elsewhere. Movement is not the same thing as progress.

China again provides a clean institutional example because the relevant text can be compared directly. The 1982 Constitution stated that the president and vice-president would serve no more than two consecutive terms. In 2018, a constitutional amendment removed that sentence. The current Article 79 still links the terms of those offices to the term of the National People’s Congress but no longer contains the two-term ceiling. Other offices did not change in the same way: the current Constitution still limits the premier, vice-premiers, and state councillors to two consecutive terms.

Even this change has to be described carefully. The state presidency is not identical to the office of general secretary of the Communist Party or chair of the Central Military Commission, and the constitutional amendment cannot by itself describe the whole Chinese succession system. That qualification is part of the lesson. One formal constraint on one pair of state offices changed while the wider party-state remained recognizable. The regime label stayed broad and stable enough to conceal a consequential movement inside it.

Placed beside the village-election history, the direction of political change becomes impossible to compress into one arrow. A bounded participatory mechanism appeared at one level; a formal succession constraint disappeared at another. The same political formation could therefore be changing along several dimensions at once. To say that the system had ‘liberalized’ or ‘authoritarianized’ might be useful for a particular argument, but either word would become misleading if it pretended to describe every relation that moved.

This was the point at which I stopped thinking of regime categories as containers. They are better understood as high-level descriptions of an arrangement whose internal relations keep moving. Some of those descriptions matter enormously. A system in which organized opposition can lawfully compete for national office is not equivalent to one in which it cannot. A system in which courts can bind executive action is not equivalent to one in which review remains advisory. The categories tell us something real about political standing. They simply do not tell us everything the political form can do, how it is changing, or which safeguard is strengthening or decaying.

What “Alive” Does and Does Not Mean

I use alive here as a political metaphor, not a biological claim. Political systems do not reproduce like organisms, and institutions do not possess genes that determine their destiny. They are arrangements maintained by people, organizations, laws, habits, resources, technologies, and expectations. They change because those relations are revised, contested, copied, evaded, strengthened, or allowed to decay.

Once the metaphor is kept that modest, it changes the unit of comparison. Instead of asking first whether a country is a democracy, dictatorship, monarchy, or one-party state, I can ask what happens when leadership changes, how administrative competence survives that change, whether inconvenient information can reach decision-makers, where an affected person can challenge a public action, and which exercises of power remain unavailable even when they would be convenient. Those questions do not erase regime differences. They open the label so the differences inside and across regimes can be seen.

The change in viewpoint also resolves a problem left by uneven development. If political capacities mature on different schedules, then a regime is not one maturity level. It is a configuration. A society can have highly professional public administration and brittle succession; broad electoral participation and weak long-horizon execution; strong local problem solving and poor national correction; impressive fiscal capacity and dangerous information bottlenecks. The combinations matter because one capability can compensate for, conceal, or worsen the weakness of another.

This is why a country can remain in the same regime category while becoming politically different enough that some comparisons based only on the old configuration stop working. New Zealand after MMP required different coalition practices and parliamentary procedures from the New Zealand that preceded it. Chinese village committees could introduce direct local selection without changing national party competition, and one constitutional succession rule could later be removed without rewriting every other office. The political form at time two is descended from the political form at time one, but it is not identical to it.

On Luminara, that seems obvious to me now because I know the end of the history. Our present garden contains mechanisms that emerged in very different ancestors and only later came to coexist. Yet that knowledge is precisely what made me careless when I first arrived here. I treated the combination we eventually reached as if the combination itself were the unit of learning. Earth is teaching me to separate the pieces again.

That separation does not imply that the pieces are detachable. The previous inquiry had already taught me the opposite: a mechanism works through enabling conditions and relations. The point is to distinguish before comparing. One can ask whether a succession rule increased predictability without assuming the rest of the political system should travel with it; whether proportional representation changed executive bargaining without treating proportionality as a complete constitution; whether local elections altered accountability without pretending they supplied national political competition. Mechanism-level description is the beginning of comparison, not permission to transplant.

What the Machine Should Notice

At the scale of two or three reforms, a patient scholar can do this without artificial intelligence. The difficulty appears when the unit of interest is no longer the country label but the movement beneath it. A political system changes through constitutional amendments, electoral statutes, administrative rules, court decisions, party procedures, budget practices, local experiments, and informal conventions. Some changes are announced as constitutional transformation; others arrive as one sentence deleted from Article 79 or as a revised rule for nominating a village committee.

The records are also written for different purposes. Electoral law describes ballots and thresholds; parliamentary records reveal how committee practice adapted; local-governance law describes an elected body while field research shows where authority actually settled; constitutional texts reveal formal succession rules while party rules and political practice may organize leadership differently. A regime classification compresses these sources into one label because classification needs a manageable unit. Developmental comparison needs the compression to be reversible.

This is where machine-scale comparison becomes more than convenience. A system can align constitutional and statutory versions across time, retrieve changes whose wording differs across languages, and surface candidate cases in which a similar function moved under different institutional names. It can search for places where electoral reform changed representation without regime conversion, where a review mechanism gained authority while executive structure remained intact, or where a succession safeguard appeared, weakened, or disappeared while the broad political category stayed the same. The output is not a verdict. It is a map of mutations worth examining.

The difference from a regime score is fundamental. A score asks the machine to compress several political relations into an ordered result. The map I want does the opposite: it preserves the relations long enough for a reader to ask what changed and what did not. It can show that New Zealand altered the conversion of votes into seats; it cannot infer from that fact whether another society should adopt MMP. It can show that a constitutional sentence disappeared from one version of Article 79; it cannot decide what succession arrangement China, or any other society, ought to choose.

The practical gain needs a more precise distinction because Earth already possesses a formidable comparative apparatus. Projects such as V-Dem and Polity were built precisely because regime labels compress too much: they disaggregate political systems into separately coded dimensions and make those dimensions comparable across countries and years. Their strength is disciplined comparison over questions specified in advance. The lineage I am after begins one step earlier and often lower down. What changed in the primary record? Which political function moved, what stayed fixed around it, and under what conditions did the change persist? Where else did a similar function mutate under another institutional name? A machine working across constitutional versions, statutes, local rules, administrative records, and scholarship could help surface candidate mutations that no codebook had anticipated as an indicator. That does not make the machine a comparative theorist. It cannot decide that two arrangements are functionally equivalent, determine which change deserves to become a variable, or tell a society what the comparison means. Its contribution is to lower the retrieval and alignment cost of building an inspectable lineage before human judgment decides what has actually been learned.

The difference is temporal as much as comparative. Regime classifications are often revised after a large transition has already become obvious. A mechanism-level map can notice quieter movement earlier: an appeal body gains power to reverse decisions; a legislature loses control of one delegated authority; a local election becomes more or less binding; a succession rule changes while every familiar institution name remains in place. None of those observations tells citizens whether the movement is desirable. It does let them argue about the actual change before a later label turns many small decisions into one apparently inevitable transformation.

That would have mattered on Luminara. Our great transitions were easy to remember because wars, settlements, and constitutional refoundings left large monuments in the record. Smaller institutional adaptations were easier to forget, especially when the political form that produced them later disappeared. A capability could be rediscovered generations later under another name because no one had preserved the relation between the old problem and the old mechanism. Earth may be able to make more of those quiet experiments mutually visible before history erases their lineage.

Again, the machine cannot supply the desire to learn. A government can see that another system has developed a useful safeguard and still reject it. An institution can know that its succession rule has become brittle and prefer the advantage of brittleness. A party can understand that a representation mechanism excludes certain voices and decide that exclusion helps it. Comparative visibility changes what can be known and compared; it does not convert knowledge into reform or give an external analyst standing to demand it.

From Types to Capabilities

I had ended Part I needing coexistence to come before comparison if the aim is to preserve living, independently developing political variation: societies need enough political space for comparison to reveal their own trajectories rather than mainly the effects of coercion and transplantation. Now I could see why that space need not mean stagnation. A political system is not a finished specimen waiting to be ranked. It can continue changing while remaining recognizably itself.

New Zealand taught me that an established democracy could alter the machinery of representation deeply enough to change governing practice without ceasing to be the same broad political order. China taught me that a one-party state could contain a bounded direct-election mechanism whose practical force depended on surrounding authority, and that the same broad system could later remove a formal succession constraint at another level. Neither case supplies a model for anyone else. Together they make static comparison increasingly difficult to defend.

I am therefore less interested now in the question that dominated so much of Luminara’s old comparative politics: Which form is the mature one? That question assumes political types are destinations and makes every internal difference secondary. Earth suggests another possibility. The more useful object may be the configuration of capabilities and safeguards a political form currently possesses, the ones it is gaining or losing, and the relations that make them durable.

But this creates a harder problem rather than solving comparison. If democracy and dictatorship are too coarse to tell us what a system can actually do, abandoning the labels cannot mean abandoning judgment. Some arrangements execute competently while correcting badly; some correct publicly while failing to sustain action; some preserve order by making criticism dangerous; some preserve contestation while allowing implementation to fragment. The next inquiry must compare those capacities without smuggling the regime ranking back in through the measurements.

A regime label tells me what kind of political order I am looking at. It no longer tells me enough, by itself, about what that order can do, what it cannot correct, or what it may yet become.