Centrism as a Discipline

Governance under uncertainty and irreversibility

The last five chapters have been a pathology report. Utopian closure turns out to be a machine for degrading agency (Against Utopia). Authoritarian politics can exploit cheaper recurrent tendencies, while liberal safeguards demand maintenance (The Authoritarian Default). Liberalism carries structural messaging disadvantages even where its institutions perform well (Why Liberalism Loses). The collectives in whose name politics is conducted turn out not to exist as agents (Fictions of the Collective), and the vocabulary we would need to say any of this precisely turns out to have been inflated into noise (Words That Lost Their Teeth). The natural question is what survives the diagnosis. What does a political actor look like who has absorbed all of it — the pull toward submission, the fragility of liberal orders, the nonexistence of the collective chooser, the corruption of the words? What temperament is left standing?

My answer is centrism. Centrism is usually treated as a punchline. From the left it is framed as moral cowardice; from the right it is dismissed as incoherence or surrender. In both cases it is portrayed as a lack of conviction rather than as a considered response to political reality. That framing misses what centrism is when taken seriously. I am not defending moderation as aesthetic balance, compromise as a virtue in itself, or neutrality as an end. I am defending centrism as a governance discipline shaped by three facts about political systems: they are complex, our knowledge of them is radically incomplete, and many of their failure modes are irreversible. Systems with those properties reward variance tolerance, institutional friction, and humility about outcomes. Centrism is what falls out once those constraints are treated as real rather than as rhetorical concessions.

A Discipline, Not a Demographic

The word “centrism” is usually used sociologically, to name a cluster of voters, parties, or elites — the people in the middle of the distribution, the ones who split the difference. That is not what I am defending. The demographic middle has no special claim to wisdom; the midpoint between two errors is usually a third error. And many self-described centrists act from inertia, donor capture, or simple risk aversion — triangulating between whatever the loudest factions currently demand. Those failures are real, and they belong to the practitioners.

Centrism, as I mean it, is a discipline: a method for governing under uncertainty. Like any discipline, it can be practiced well or poorly, and failure to meet its standards indicts the agent rather than the framework — no one thinks a botched proof discredits mathematics. Treating centrism this way aligns it with the other load-bearing concepts of this philosophy: agency, coherence, survivability. What matters is not the label but adherence to the constraint. A politician who calls himself a radical while carefully bounding failure modes and preserving reversibility is practicing the discipline; a politician who calls himself a centrist while drifting wherever incumbency pulls him is not.

The Shared Error of the Extremes

Ideological extremes differ sharply in values, narratives, and policy goals, and they spend their energy advertising those differences. But they converge on a deeper structural assumption about what politics is. Politics is framed as a moral battlefield. Progress becomes victory. Opposition becomes enemy. Power becomes the mechanism through which righteousness is enforced.

Once politics is framed this way, the pathologies follow with the predictability of a theorem. Tradeoffs lose legitimacy — on a battlefield, weighing costs against the cause is treason. Variance becomes injustice by default — any difference in outcomes is evidence of the enemy’s work. Institutions appear as obstacles rather than safeguards, since anything that slows the righteous slows righteousness. And confidence in moral intent substitutes for analysis of consequences: the question “what will this do?” is displaced by the question “whose side are you on?”

Centrism rejects the battlefield frame at the root. It treats politics as an ongoing coordination problem among agents with partially conflicting goals, incomplete information, and asymmetric power. In that setting, moral clarity does not dissolve complexity. It often amplifies error, because clarity licenses action, and action whose downstream effects are poorly understood does damage in proportion to the confidence behind it. The extremist’s certainty is not a moral achievement layered on top of his analysis; it is a substitute for the analysis.

Society as an Evolving Ecology

Political metaphors are not decoration; they shape what actions seem permissible. The battlefield metaphor invites escalation. The mechanical metaphor — society as a machine with levers and dials — invites control fantasies, five-year plans, social engineering. A more accurate frame treats society as an evolving ecology of agents embedded in institutions.

Ecological systems exhibit properties that matter politically. They contain multiple actors with divergent incentives, so no intervention lands on a passive substrate; everything is answered. They display path dependence: early changes constrain later options, and you cannot rerun the founding. They produce nonlinear cascades, where small interventions trigger disproportionate effects through channels no one mapped in advance. They settle into local equilibria rather than global optima — which is one more reason the utopian project is ill-typed: there is no summit to march to, only the landscape and its basins. And they accumulate irreversible losses alongside slow recoveries: an institution that took three generations to grow can be destroyed in a legislative season.

Systems like this cannot be solved. They can be stewarded, stressed, or destroyed. Interventions do not simply produce outcomes; they alter selection pressures and reshape incentives, so the system that receives the tenth intervention is not the system that received the first. Success in such a system usually looks like resilience rather than perfection. Centrism aligns with this reality: governance becomes an exercise in damage control and coordination rather than moral purification. That description will strike the battlefield mind as a shameful lowering of ambition. It is actually the recognition of what kind of object a society is.

The Variance Distinction

The immediate objection is that this counsels acquiescence — that treating society as an ecology means shrugging at whatever the ecology produces. The answer to that objection is the moral compass of the whole discipline, and it turns on a distinction about variance.

Variance appears wherever systems are allowed to operate. Differences in outcomes, capacities, preferences, and performance arise even under scrupulously fair rules, because agents differ and choices compound. Treating all variance as injustice therefore leads somewhere perverse: since variance cannot be eliminated, the political strategy becomes abolishing its measurement, punishing its visibility, or attributing all of it to malice — anything but understanding its causes.

But it does not follow that all variance is acceptable. The crucial distinction separates variance produced by voluntary interaction from variance produced by coercive constraint. The former reflects decentralized choice under conditions of agency: people wanted different things, tried different things, and got different results. The latter reflects the systematic suppression of agency through credible threats of harm: people got different results because some of them were not permitted to choose.

Centrism tolerates the first kind and treats the second as a structural defect requiring intervention. The question is never whether outcomes differ — they always will — but whether the differences arise from preserved agency or enforced constraint. Once that distinction is explicit, the standard accusation collapses. Tolerance is not complicity. It is selective intolerance of coercion: indifferent to divergence as such, implacable toward the boot.

The Epistemic Core

Centrism is an epistemic stance before it is a political one. Its premises are about knowledge, not values: knowledge is partial, incentives are complex, feedback is delayed, and error scales with power. A wrong idea held by a pamphleteer is a wrong idea; the same idea held by a ministry is a catastrophe.

Several commitments follow. Intentions do not guarantee outcomes, so a policy’s moral advertising is evidence of nothing. Second-order effects often dominate first-order gains, so the visible beneficiary is not the ledger. Policies should be judged by how they behave under stress — under bad administrators, adversarial gaming, and conditions their authors never imagined — not by how they sound in principle. And authority should be distributed and constrained, because confidence magnifies harm precisely when it is wrong, and it cannot announce in advance which time that will be.

This posture is Conditionalism applied to politics. Political claims, like all claims, make sense only relative to background conditions: a policy that works in one institutional ecology fails in another, and an unconditional policy prescription is not strong but incoherent, because its effects depend on a context it refuses to specify. Centrism internalizes this and builds governance structures that expect revision rather than finality. It is the political face of error correction — and it explains why the fictions of the previous chapter are not innocent. A regime that claims to act for “the national interest” has not merely made a category error; it has disabled the machinery by which its errors could be named, because dissent from a fiction can always be moralized into disloyalty.

Institutions as Scaffolding

Institutions attract constant criticism for being slow, imperfect, and unjust, and the criticisms are frequently accurate. The mistake is in the implied standard. The function of institutions is not to instantiate moral truth. It is to constrain error and limit damage when judgment fails — and judgment always eventually fails.

This is Hayek’s insight about knowledge.¹ No central authority can gather the dispersed, tacit, local information that a society runs on, so centralized control is not merely tyrannical but ignorant, and its confidence is exactly as dangerous as its reach. Institutions, on this view, encode lessons learned through failure — each procedural annoyance is usually the scar of some past disaster — and they create friction precisely where confidence would otherwise run unchecked. The independent court that blocks the popular measure, the supermajority requirement that stalls the urgent reform: these are not defects in the machinery of righteousness. They are the machinery working.

Centrism therefore treats institutions as ecological scaffolding, and reforms them the way one reforms scaffolding: to improve resilience and feedback, not to enforce ideological purity. The objective is survivability under stress rather than optimization under ideal assumptions. This is also why the centrist can never share the revolutionary’s cheerful talk of demolition. The revolutionary sees an obstacle between the people and the good; the centrist sees a load-bearing wall between the people and the recurring disasters that built it.

Metastable Evil

Ecological stability, however, is not sufficient for legitimacy — and here the discipline diverges sharply from the conservatism it is often mistaken for. Some systems remain stable precisely by suppressing agency. Feudalism persisted for centuries. Caste systems persisted for millennia. Apartheid was internally coherent, administratively sophisticated, and brutally enforced. These were not equilibria in any honorable sense; they were metastable failure states — arrangements held in place not by the consent of their participants but by the systematic, credible threat of harm against anyone who moved.

Under the centrist discipline, such systems qualify as defects, not as ecosystems to steward. The variance distinction does the work: what these systems stabilized was coercively produced variance, the exact thing the compass condemns. When a social order’s stability depends on coercion and irreversible agency suppression, disruption becomes justified even at significant risk. Abolition, the civil rights movement, and decolonization required decisive rupture, because incrementalism, in those cases, was not prudence — it was the preservation of injustice on the installment plan.

So centrism does not oppose disruption as such. It evaluates disruption the way it evaluates everything else: by its failure modes, and by whether it expands or destroys agency over time. The discipline that counsels patience with a flawed but correctable institution is the same discipline that licenses rupture against an incorrigible one. The two verdicts are not in tension; they are the same test applied to different facts.

Selective Decisiveness

The standing charge against centrism is passivity — that it is a philosophy of the fence, constitutionally incapable of action at the speed events demand. The record of the previous section already answers half of this. The other half is answered by how the discipline handles irreversibility.

Centrism is selectively decisive. It supports strong action where failure modes are bounded and inaction produces irreversible loss. Existential risks — climate instability, uncontrolled AI deployment, demographic collapse, institutional failure — are precisely the cases where the discipline demands nonlinear response, because “wait and see” is itself a bet, and against an irreversible downside it is a bad one. What centrism opposes is not decisive action but a particular justification for it: moral certainty in place of analysis of failure dynamics. The activist who demands sweeping power because the cause is righteous and the activist who demands it because he has mapped the failure modes and bounded them are asking for the same thing on entirely different warrants, and only one of them should get it. How badly this goes when the battlefield frame captures a genuine risk domain — when “safety” becomes a permission layer for installing one faction’s values — is a story told elsewhere in this book.

The discipline is the same across every domain. Act where delay compounds harm. Constrain power where certainty outruns knowledge. Preserve reversibility where possible. Accept irreversibility only where every alternative collapses the future entirely.

Why Centrism Provokes Hostility

If centrism is this — a compass, a set of tests, a licensed capacity for rupture — why is it despised from both directions? The polemical answer is that it is despised for what it denies.

Extremes depend on narrative clarity. They require villains, heroes, and historical destiny; the story is not an ornament on the politics but its fuel. Centrism denies every faction its moral monopoly. It treats opponents as constrained agents rather than embodiments of evil, which spoils the villain role. It treats institutions as safety devices rather than as sacred obstacles or disposable tools, which spoils the plot. Above all, it refuses to convert conviction into entitlement — to accept that the intensity of a belief confers any right to enforce it. For movements built on identity and certainty, this refusal is experienced as an attack, and the resulting hostility reflects psychological loss as much as political disagreement. Centrism takes away the story people were living inside.

There is an aesthetic version of the same complaint: centrism is boring. It offers no catharsis and no final victory, and this is taken as evidence of weakness. A different interpretation fits the evidence better. Systems that endure optimize for continuity and error correction rather than drama. The politics that keeps societies functioning rarely trends, because it does not satisfy narrative hunger; the sewage system does not inspire anthems either. Extremism compresses complexity into moral slogans — that is its product, and the compression is why it travels. Centrism refuses compression, and pays for the refusal in reach. Boredom, here, is not the absence of a position. It is what a position sounds like when it declines to lie about how complicated things are.

Principled Actors

A discipline is only as real as its practitioners, and this returns us to where Part VI began. The authoritarian cycle described at the opening of this part runs on instrumental principle: factions champion free speech, tolerance, and limited government while out of power, because universal principles build coalitions, and abandon them upon gaining power, because the same principles then obstruct. Both left and right run the cycle on schedule.

What breaks the cycle is not a better constitution but a rarer kind of agent. Genuinely principled behavior — upholding values regardless of short-term advantage — is inherently costly. It means sacrificing expediency, enduring discomfort, and tolerating dissent one has the power to silence. That is why principled actors are rare, and why they are indispensable: without them, political landscapes devolve into competing authoritarianisms, each justifying its abuses by citing the other’s excesses, forever. Principled actors provide stability, reduce transactional friction, and uphold the norms of accountability that no rule can enforce on its own. The requirement is that principles be held as terminal values, not instrumental tactics — valued for what they are, not for what they currently win.

Centrism as a discipline is what that terminal commitment looks like in office. Its practitioner defends the speech rights of people whose speech he detests, preserves the institutional frictions that currently frustrate his own side, and constrains powers he could use well because his successors might not. None of this is moderation of belief. It is the refusal to let belief, however intense, exempt itself from the constraints that make correction possible.

What the Discipline Promises

The positive ethos, assembled: pluralism grounded in constraint rather than in relativism. Agency over purity. Coercion treated as a costly instrument, never as a moral reward for being right. Reversibility, feedback, and institutional memory as design criteria. And the standing acknowledgment that no political arrangement resolves all tensions — politics remains an ongoing coordination problem, not a final reckoning, and progress appears as incremental improvement punctuated by necessary rupture where agency is systematically suppressed.

Centrism, understood this way, does not promise moral satisfaction. It promises survivability under uncertainty. It treats power as hazardous, certainty as brittle, and agency as the condition under which correction remains possible. In a world shaped by irreversible choices and incomplete knowledge, this posture is not timid; it is exacting. It asks political systems to remain governable even when they are wrong, and it asks political actors to design for that inevitability. That requirement does not inspire chants. It does something harder: it keeps open the space in which correction, pluralism, and future choice remain possible.

But a discipline for wielding hazardous power is still a discipline for managing a hazard, and it invites a further question: whether the hazard must be wielded at all — whether politics, in the coercive sense, is a permanent feature of human coordination or an exploit that can be engineered away. That question opens the final part of this volume, beginning with The Conversion of Coercion.