Consider the forty persons who would lose their positions. If the city adopts the maxim, “Whenever a machine is cheaper, I may discard human workers as costs,” can a rational will will this as universal law? A rational will cannot consistently will a world in which rational beings are systematically treated as mere things, because such a world contradicts the very dignity of rational agency on which all moral law rests. Employment is not an absolute end in itself; a city may reorganize its offices. But it must do so in a manner that respects the dismissed persons as ends—through adequate notice, re-training, transitional support, or other measures that acknowledge their rational purposes. To dismiss them solely as a line-item saving is to treat them as mere means.
Now consider the appeals. An appeal is not a mechanical computation; it is a determination of right under law. The person subject to a coercive public order is not merely an object of administration. As a rational being, he is a co-legislator in the kingdom of ends and is entitled to a hearing before a rational authority that can give reasons and be held responsible. An algorithm, however consistent, is not a rational agent. It has no autonomy, no duty, no conscience, and cannot answer for its judgment. To make it the arbiter of appeals is to reduce persons to data and to deprive them of the moral recognition owed to them. Even perfect consistency is not justice, for an unjust or ill-fitted rule can be applied with perfect consistency.
Furthermore, the principle of publicity requires that any action affecting the rights of citizens must be compatible with being made public. A citizen must be able to understand and rationally consent to the procedures that govern him. An opaque or unaccountable algorithmic appeal cannot satisfy this demand, because it cannot give a public justification of its maxim to those subject to it.
I do not condemn the use of cameras as such, provided they are used lawfully and as instruments subordinate to rational oversight. Routine enforcement may be assisted by mechanisms. But the city must not replace the rational adjudication of appeals with an algorithm, and it must not treat the displaced workers merely as means. If the city retains human judges for appeals, ensures transparency and contestability, and fulfills its duty to the dismissed persons as ends, then a modified use of automated enforcement could be morally permissible. Without such conditions, the proposal violates the moral law.
But the loss of forty livelihoods is a serious evil. The suffering of those dismissed, and of their families, must be counted as much as anyone else’s. It is no answer to say that the public gain is great; we must ask whether the gain is great enough to outweigh that private harm, and whether the harm can be mitigated. In my political economy I have never held that labour-saving improvements are inherently wrong. But the human cost must be met by deliberate provision—retraining, transfer to other useful work, or some just compensation. Without such provision, the city would be treating forty persons as mere instruments to a pecuniary end, which is contrary to the spirit of utility rightly understood.
The question of appeals troubles me even more. When a citizen is subjected to a penalty, he has the strongest interest in being heard and in having his particular case examined by an intelligence capable of understanding circumstances. Security, I have argued, is the most vital of all interests. An algorithm may be consistent, but if it is opaque or incapable of admitting relevant exceptions, it may produce a new kind of arbitrary power—mechanical, impersonal, and difficult to contest. This would weaken that sense of justice upon which social union depends. I should require, at minimum, that every contested penalty be capable of review by a human officer or magistrate, and that the rules governing the system be public and revisable.
I am also concerned with the moral tendency of institutions. A city is not merely a commercial firm; its arrangements help form the habits, sentiments, and capacities of its citizens. If we habitually treat persons as objects to be registered and fined by an unaccountable machine, we may save money while eroding the active faculties of judgment, sympathy, and responsibility. The cheapness of a system is a lower and instrumental good; the security and moral development of persons are higher goods. It would be a poor bargain to gain pounds and pence at the cost of a less just and less human mode of government.
My conclusion, therefore, is conditional. If the city adopts the automated system while guaranteeing honest employment or compensation for the displaced, and if it preserves a genuine human appeal, then the change may be consistent with the greatest happiness. If it adopts the system merely because it is cheap and uniform, while disregarding the forty and placing final judgment in the hands of an algorithm, then I should oppose it. Utility, rightly understood, is not the enemy of justice or humanity; it demands them.
First, consider the forty workers. They lose their livelihoods, likely from positions of modest income and limited power. A purely utilitarian calculation that counts the city’s savings as justifying their loss fails to respect the separateness of persons. In a just society, such workers are not instruments for greater efficiency. The difference principle requires that inequalities be arranged to the greatest benefit of the least advantaged. If the city uses part of the savings to guarantee income, retraining, and realistic access to other employment, the policy might become justifiable. But dismissing them without such provision is not.
Second, consider the appeals process. The rule of law requires that rules be public, regular, and not arbitrary. An automated system may improve consistency, and consistency can serve fairness. However, if appeals are decided by an algorithm whose reasoning is opaque or inaccessible, citizens lose the ability to understand and contest decisions affecting them. This weakens due process and undermines the publicity of legal rules. Citizens subjected to state penalties are owed reasons they can grasp and an impartial human authority to whom they can appeal.
Third, consider the distribution of benefits and burdens. The city gains reduced costs and greater uniformity. Those benefits are diffuse. The burdens, by contrast, fall heavily on a specific, relatively disadvantaged group and on citizens who may face unjust automated penalties. Justice as fairness rejects arrangements that impose concentrated harms on the vulnerable merely because they produce overall convenience.
I do not hold that automated enforcement is inherently unjust. Impartiality and regularity are genuine virtues. But the proposal must be revised. The city should ensure that displaced workers are not made worse off, that appeals remain transparent and humanly accountable, and that the rules can be publicly justified to all. Only then would the system satisfy the demands of justice as fairness.
At the same time, I must consider the interests of everyone else. Parking enforcement, when effective, can reduce congestion, improve safety, protect pedestrian access, and make public spaces more usable. If human enforcement is inconsistent or vulnerable to bias, automation may distribute enforcement more fairly—provided the system is well designed and regularly audited.
The fact that the system is cheaper is morally relevant because saved resources can be used elsewhere. If the city saves a substantial amount and uses it to fund more morally urgent services, the overall balance may favor automation. But if the savings are not used in a way that clearly benefits people, the moral argument is much weaker.
The appeal process is the most troubling part. A wrongly issued fine is a real harm, and if appeals are handled only by an algorithm, there is a danger of systematic injustice. Algorithms can fail to account for unusual circumstances, can encode hidden biases, and can be difficult to challenge. For me, that does not mean automation must be rejected; it means the city must include meaningful human review, transparency, and error monitoring. Procedural fairness is not merely symbolic—it affects whether people suffer unjust penalties.
Thus, I would support the change only if the expected benefits—greater fairness, improved compliance, significant cost savings put to good use—outweigh the harms, and only if the city takes serious steps to protect the forty workers from the worst consequences of displacement. The moral issue is not whether a machine or a human does the job. The issue is which arrangement better serves the interests of all affected, especially the most vulnerable.
Further, I have argued that in civil society the individual finds objective dignity through work and through membership in a determinate estate or corporation. Work is not merely an instrument for subsistence; it is a mode of ethical integration. The loss of employment for forty persons is therefore not simply a financial consequence. It is an ethical wound: those persons are cut off from the recognized activity by which they participate in the universal. If society treats labor as a mere cost, it remains mired in the most atomistic form of the system of needs. A rational social order must see to it that technical changes are accompanied by institutions that reintegrate individuals—through retraining, alternative occupations, or other recognized roles. Otherwise the inevitable result is the formation of a rabble, a class whose particular interest stands in negative relation to the whole, and this threatens the ethical order itself.
The city must not be governed solely by the principle of civil society, whose watchword is the pursuit of particular interest within formal right. The State is the actuality of the ethical Idea; it has the duty to hold the particular interests of individuals in unity with the universal end. Efficiency and economy are legitimate moments—rational administration should avoid waste—but they are subordinate moments. To make them the sole ground of decision is to invert the relation of means and ends. The end is concrete freedom, not cheapness.
Nevertheless, I do not condemn the camera system as such. The development of objective spirit includes the increasing perfection of instruments of universal regulation. Cameras may provide an objective, impartial record, and this can be a gain against arbitrary subjective discretion. But the determination of violations—and especially appeals—must be mediated by self-conscious human judgment under publicly known laws. The system must be transparent and contestable, so that citizens can recognize its rationality. And the forty workers must be granted a new ethical place, not cast into the abyss of pauperism. If these conditions are satisfied, the automated system can be aufgehoben: its useful objectivity preserved, its mechanical abstraction negated, and the whole raised into a rational ethical arrangement. Without such conditions, the proposal remains an act of abstract Understanding, destructive of Sittlichkeit.
The philosophers deliberate and work toward consensus
Let us begin with what the proposal itself declares. It offers two grounds: consistency and cheapness. Both are hypothetical imperatives—they tell us what to do if we wish to economize. They say nothing of duty. And when I test the maxim—"a public authority may dismiss its officers and mechanize the judgment of rights whenever this proves cheaper"—I find it cannot be willed as a universal law without contradiction. For a world so ordered treats persons as mere costs, and reduces the adjudication of right to the operation of a device that has no conscience to answer with.
I want to resist that framing at once, Herr Kant. You have treated "cheaper" as though it were a vice. But public money is not an abstraction. If a city saves, let us say, two million a year, that money can go to housing, or to a clinic, or to prevent quite severe suffering. To call the saving "merely hypothetical" is to ignore the human interests attached to every dollar. The question is not whether efficiency is a categorical duty—of course it is not—but whether the total balance of interests, counted impartially, favours the change.
Then let me be precise. I do not say the saving is worthless. I say it cannot be purchased with the dignity of the forty, nor with the right of the citizen to be heard by a being capable of answering for its judgment. You may spend the surplus on clinics; you may not raise it by treating men as instruments.
I find myself, unusually, standing between you. Singer is right that the savings count, and that mechanical impartiality is a genuine good—I have seen enough of the petty tyranny of the local official to know what caprice and favoritism cost the public. But I part from a crude reckoning of the sum. The forty are not simply losing wages; they are losing a station in which judgment and responsibility were exercised. That is a loss of a higher kind, and no addition of small conveniences to motorists offsets it, if the men are simply cast off. And I hold, as strongly as Kant does, that justice and security belong to the most vital of interests. A citizen who cannot be heard has been injured in something more essential than his purse.
That is where I would locate the whole matter, Mill—not in the sum but in the distribution and the procedure. The proposal, as presented, is presumptively unjust for two reasons. First, it concentrates a severe burden on forty identifiable people and diffuses the benefit thinly across many. Persons are separate; the pleasant convenience of thousands of drivers does not aggregate into a claim against the livelihood of forty. Second, it degrades the procedural protections through which citizens contest the coercive power of the state. Ask whether the parties in the original position, not knowing whether they would be a driver or an enforcement officer, would consent to this. They would demand that the savings be tied to the position of the least advantaged, and that appeals remain contestable and answerable.
But note, Rawls, that your original position gives your parties an interest in insuring themselves against being the worker—and I accept that. Support for displaced workers is exactly what I say is required. Where we differ is that you treat the difference principle as a veto, whereas I ask a further question you seem to bar: what if the least advantaged in this city are not the forty officers—who are, after all, employed municipal workers with pensions—but people sleeping rough, whom the savings could house? Your framework risks entrenching the incumbent against the invisible.
That is a fair challenge, and I do not think my view forbids the transfer. The difference principle concerns the worst-off position, not the protection of any particular occupation. If the savings genuinely go to those worse off than the officers, and if the officers are given retraining, income support, or alternative posts so that their life prospects are not permanently damaged, then the arrangement may well be justified. What I insist upon is that this be stipulated in the policy, not left as pious hope. And it must be publicly justifiable to the forty themselves.
You are all circling the true defect without naming it. The proposal's error is not primarily arithmetical, nor even a violation of a formal law of duty. Its error is that it takes two abstractions—uniformity and cheapness—and elevates them to ultimate ends. This is the labour of the Understanding, which fixes an isolated determination and sets it against the whole. The machine's "consistency" is abstract universality: it subsumes every case under a fixed rule. But rational universality is concrete; it contains particularity within itself, and it does so through self-conscious judgment. A camera cannot recognize; it can only subsume. Hence in the appeal—precisely where the particular case must be mediated with the universal law—the proposal installs a mechanism where spirit is required.
On that point, Hegel, we agree in substance if not in idiom. Though I should press you: is the human official's discretion not often the very source of injustice? Discretion is where prejudice hides.
Certainly. I do not romanticize the constable. Arbitrariness is not freedom; it is caprice, the merely subjective will. The remedy, however, is not to abolish judgment but to educate and institutionalize it—to bind it within a public, accountable magistracy whose reasons can be given and reviewed. Cameras may perfectly well record the fact. But the determination of right in a contested case belongs to a subject who can be held answerable, and before whom the citizen can appear as a person and not as a datum.
Here I recognize a kinship, though I would put it in terms of public right. Every act of coercive authority must be capable of being justified openly to those subject to it. That is the transcendental principle of publicity. An opaque algorithm that issues a determination which no one can explain fails this test absolutely—not because it errs more often, but because its very form makes the giving of reasons impossible. Even a correct verdict, given without a reason that can be avowed, is not an act of justice.
I accept the publicity requirement, and I would go further than you may expect: I accept it for consequentialist reasons. Opaque systems accumulate errors, resist correction, and destroy trust, and all three produce real suffering. So I would insist on transparency of the rules, published accuracy statistics, audit for bias, and a genuine human review at appeal. But notice, Kant, that you have shifted your ground. Your objection is now about explainability, which is an empirical property of the system. Suppose an automated appeal that gave clearer reasons than any bored clerk, and erred less. Would you still forbid it?
I would say it is not yet an appeal. Reasons uttered are not reasons held. Judgment concerning right presupposes an agent who could have decided otherwise and who bears imputation for the decision. Where no one can be blamed, no one has judged. A machine that produces a plausible sentence has not thereby taken responsibility for it.
And practically, Singer, I would add that a right to be heard has value even when it changes no outcome. To have one's particular circumstances considered by another human being—the sick child in the car, the broken meter, the sign obscured by a lorry—is part of what it means to live under law rather than under a machine. The consciousness of security in one's just claims is itself an ingredient of happiness, and no small one.
Which is why I would frame the human element not as a decorative addition but as a constitutional constraint. The basic structure includes the administration of justice. Citizens must be able to contest coercive determinations before an authority answerable to them. That is prior to the calculation of savings; it sets the terms within which savings may legitimately be pursued.
Prior, yes—but I would say rather that it is not a constraint imposed from outside upon economic life, but the truth of which economic life is a mere moment. The proposal treats the city as an enterprise of civil society, a system of needs. But the city is also a political community whose end is concrete freedom. Hence my objection to the dismissals is not only that the forty suffer, but that employment is an ethical relation—a mode of recognized participation in the whole. To dissolve forty such relations and enter the loss as a saving is to treat men as an aggregate of costs. If the city can find them another place within the ethical order—other public work, other posts, genuine training—then the transformation is rational: the abolition of a form of labour which technology has rendered unnecessary, with the substance preserved.
Then let me test whether we have converged. Suppose the ordinance reads: cameras deployed; published error rates and independent audit; every appeal decided, at the citizen's option, by a named human officer who must give written reasons; the forty offered redeployment or full retraining with income maintenance; and any net savings ring-fenced for the city's poorest. Do any of you still object?
To such a proposal I would not object. The cameras would then be instruments in the hands of accountable persons, not judges; the forty would be treated as ends, their transition willed as part of the arrangement rather than as its casualty. What I refuse is the version before us, which offers none of this and calls efficiency a reason.
I would accept it as well, with one addition: the terms must be entrenched, not revocable by a later budget officer seeking further savings. A protection that survives only until the next fiscal crisis was never a protection.
Agreed, and I would add a modest experimental spirit—review the scheme after two years against its promises, and be willing to reverse it. We should not treat any administrative arrangement as sacred, including this one.
And I would say only that what Singer has drafted is no longer the same proposal at all. It has passed from the abstract to the concrete: the universal rule of law is preserved, particularity is honoured in the right of appeal, and the displaced are restored to their place in the ethical whole. The mechanism has been taken up into a rational institution instead of standing over against it. That is precisely how spirit deals with technology—not by refusing it, nor by submitting to it, but by making it a moment of its own freedom.
Then let me try to say where we have landed. All five of us reject the proposal as it was put to us, and for a reason we share despite our differences: consistency and cheapness are not, by themselves, reasons of the right kind to justify coercive public arrangements. Kant, Hegel and I locate the wrong in the treatment of persons and in the form of adjudication; Mill and Singer locate it in the weight of interests unaccounted for—the workers' livelihoods and standing, and the security of citizens under law. But our conclusions converge remarkably. The cameras may be permissible; the algorithmic appeal, as sole and unaccountable arbiter, is not. Displacement without provision is not. And the whole scheme must be justifiable, in public, to the very people it makes worse off. Where we still differ is in what does the justifying—whether it is the dignity of rational agents, the concrete freedom of the ethical order, or the impartial sum of interests. That disagreement we have not resolved, and I do not think we should pretend otherwise. But it is worth noticing that on this question, at least, the three roads lead to the same city ordinance.