The Price of Agency

Why preventing evil can endanger agency

For most of its history, the problem of evil belonged to theology. The free-will defense answers that genuine agents can choose harm and that eliminating every possibility of evil would eliminate the freedom that makes authorship valuable. Remove the deity and a real design tension remains: institutions can reduce opportunities for intentional harm, yet guarantees require controls strong enough to threaten the authorship they aim to protect. That is an argument about this framework’s chosen conception of agency, not a structural theorem proving that every preventive institution destroys freedom.

Here is the secular form of the problem. Against any political philosophy that restricts force to defense — libertarianism is the usual target, and the ethics this volume has built inherits the charge with interest — there stands a familiar objection: the framework may work as law, but it has no way to deal with evil. Harm exists. Some people reliably and deliberately cause it. A philosophy that only draws boundaries around coercion, the objection runs, is evasive about the one thing an ethics is for.

The objection is not wrong about evil. It is wrong about systems. The issue is not whether evil exists — it does, and this volume did its definitional work partly so that the word would mean something. The issue is what it would even mean for a legal or political framework to eliminate it.

Upstream of Law

Evil is intentional harm caused by an agent — an agent knowingly acting to reduce another agent’s viable futures. The definition was earned earlier, and it matters here for what it locates. Evil does not live in statutes or their violation. Its intention arises upstream of law, in preference structures, identities, beliefs, incentives, and commitments. By the time harm issues in an act, that formation and intention are already in place; the act supplies the material setback that completes the definition.

Law operates downstream. It can respond to harmful action once it occurs; it can contain an ongoing threat. What it cannot do is reach backward into intention without extending coercion into the domain of thought. A framework concerned with the boundaries of force is answering a deliberately constrained question — under what conditions may one agent override another’s agency? — and it stays silent about motives, values, and character not because they don’t matter but because it refuses to claim jurisdiction over them. That silence is routinely read as inadequacy. It is a refusal to conflate governance with moral authorship.

The Eradication Trap

Suppose we refuse the refusal. Suppose we demand a system that deals with evil at the source — that suppresses harmful intent before it becomes harmful action. Such a system needs preemptive force, and preemptive force restricts agency in advance of any demonstrated harm. By this volume’s own standard, that restriction can itself impose harm when it materially reduces an agent’s viable futures relative to an appropriate baseline. Its wrongfulness still depends on evidence, authority, necessity, proportionality, alternatives, and review; blanket preemption cannot escape those burdens by calling itself prevention. The instrument of eradication risks entering the category of the thing it claims to eradicate.

From this follows the trap, and it is structural, not merely historical. A system that claims the power to eradicate evil faces two recurrent paths. It can redefine evil as disobedience — punish the violation of rules rather than the intention of harm, so that “dealing with evil” quietly becomes dealing with dissent. Or it can authorize action on suspicion and prediction — police the interior, treat beliefs as risks and profiles as verdicts. Both paths transfer responsibility from agents to institutions. Harm becomes procedural; what began as prevention can settle into administration. The risk does not depend on the wickedness of a particular regime. It follows from the logic of moral preemption itself, and the historical systems that promised to abolish evil repeatedly administered it instead.

The Containment Layer

When law withdraws from moral preemption, the remaining work does not disappear. It moves into culture — and this is the part of the answer the objection never sees, because culture does its work without uniforms. Reputational memory shapes incentives across time: the cheat finds his options quietly narrowing for years. Selective association — in employment, housing, membership — limits exposure to harmful actors without touching them. Shunning and exclusion impose real cost without violence. Covenant communities let people live under thick shared norms without demanding that everyone else live under them too. And exit makes avoidance possible where reform would require domination.

These mechanisms guarantee nothing; nothing does. Where they remain plural and exit is realistic, their failures can stay comparatively local. Where reputation, employment, housing, or membership controls scale into coordinated exclusion, cultural sanctions can become system-wide without acquiring a legal monopoly. The dependence runs the other way too, and it should be stated plainly: where culture is weak and law is restrained, predation can fill the gap. Where culture is strong and law is restrained, agency can remain viable without collapsing into naïveté. Legal minimalism is not a stand-alone machine. It presupposes cultural robustness — not as a moral aspiration but as a functional requirement, the way a bridge presupposes bedrock.

Defense, Not Punishment

None of this leaves the framework passive once harm arrives. When harm or credible threat exists, defensive coercion can be justified where evidence, authority, necessity, proportionality, minimally harmful means, and review support force, containment, or isolation. Incarceration can fit here, and the fit reorders what a prison is for. It is defensive isolation of an ongoing source of harm: not punishment, not deterrence theater, not moral correction. Its justification persists only while those conditions do, and its moral cost accrues continuously to the institution imposing it — a running liability, never a settled account. The question that governs release is not whether the prisoner has repented, which no institution can observe, but whether the conditions that produced the harm have materially changed in ways an institution can observe. Containment is an engineering problem. Treating it as an evaluation of souls is how defensive isolation slides back into the eradication trap.

The Supplement

The critic’s move, at this point, is to concede the analysis and demand a supplement: very well, the boundaries of force are what they are, but the framework needs something more to deal with evil. Everything turns on the nature of the something. Historically it has exactly one form: a moral authority empowered to act ahead of harm. Values become enforceable. Beliefs become risks. Control presents itself as care. And authority justified by moral necessity expands, because the harms it produces arrive pre-legitimized as protection. What the promise of moral safety conceals is a transfer — of authorship, from individuals to systems. The supplement does not deal with evil. It deals with agency, and calls the result safety.

The Price

So what does sufficiency look like? This chapter cannot establish that no alternative arrangement reduces intentional harm while preserving agency more effectively; that would require comparing institutional designs under shared evidence and metrics. It establishes a narrower claim: a guarantee eliminating every intentional harm would require preemptive restrictions incompatible with the strong authorship this framework values. Partial prevention, due process, environmental design, and reversible safeguards can improve safety without abolishing agency, and remain empirical institutional questions. A world that preserves robust agency permits some misuse; a world that guarantees no misuse would replace authorship with control. Between those poles lie many imperfect designs.

This is theodicy without God, narrowed into a design constraint. The possibility of evil persists where agents retain meaningful authorship. Eliminating that possibility would require eliminating or exhaustively controlling relevant choice; reducing its incidence does not.

This volume opened by tearing down a comfort: value without a valuer is a category error, and unconditional morality died with it. What that death purchased was not license but load — the birth of moral agency, responsibility landing on us because there was never anywhere else for it to be. Everything since has been the work of carrying that weight: exposing the standards and models inside moral claims, treating sacrifice as defeasible evidence of value, defining coercion, consent, and harm precisely enough to argue with, refusing the moral arithmetic that would spend one agent’s agency for another’s aggregate, and choosing reciprocal protection of authorship as this framework’s governing premise. The premise can classify acts beyond the circle of those who endorse it; it does not become agent-independent by doing so. And the price of the chosen framework — of agents who retain robust authorship — is that some of those authors will choose harm. Evil is not the refutation of this ethics; it is the receipt. A world in which evil is impossible is a world in which relevant choices have been eliminated or exhaustively controlled. Every world this framework regards as worth living in keeps some possibility of evil as the cost of keeping its agents authors — and we accept that price with open eyes because the value judgment is ours.