A Stall on Levee Street

Levee Street is an old street along the river in Apple City. On it stands a stall that a woman kept for twenty years. This year she died, and her daughter wanted to carry the stall on.

Under an ordinance of the City Council, the Permits Office had set clear rules for street stalls: permits go only to applicants who run the stall themselves; each block has at most three stalls; new applications are queued in the order received. What the rules did not write about was succession; there was only a note on the last page, which comes up later. Processed as a new application, the daughter would be fortieth in the queue; processed as a succession, she would take her mother’s spot directly. The thirty-nine people ahead of her were waiting too, and the residents across the street had to pass along this stretch of pavement to get in and out.

Three things first, as every example in this series begins. Whom it concerns: her, the others in the queue, the residents across the street. Which line: neither granting nor refusing crosses a line; livelihood, place in the queue and passage are all interests within the line. Baseline: the ordinary new application the rules have in view. No one here has a basic right that requires the spot to be his. Yet this does not make the matter indifferent: the daughter’s livelihood, the queuers’ place in line and the residents’ passage all rest on this one judgment. The line is not crossed, and the judgment must still be made; once it is made, someone must still answer for it.

It is judgments like this that the Theory of Government writes about. Its first paper asks four things: by whom the judgments within a government are made, in whose name they take effect, to whom they answer, and for what.

Who Judged

The permit examiner processed it as a new application. In the case file he wrote: “The rules do not provide for succession; processed as a new application, fortieth in the queue.” Three days later the daughter received the written decision, headed “Apple City Permits Office” and bearing its seal. Who made this judgment? The permit examiner did; the rules did not make it for him. To read the rules’ silence as “a new application” and not as “a succession” was his reading; to place her fortieth was his weighing; once it was handed down, carrying it was his as well. Action Theory says: “Any set of rules that tries to replace it still comes down, in the end, to a person judging whether it applies to the case in front of him.” Extending this point of Action Theory to a person sitting in a public position is the series’ own step; it borrows only the reason why rules cannot judge in anyone’s place.

Yet the decision was not issued in his name. What took effect was a decision of Apple City; the permit examiner signed it through the position that is the Permits Office. So two things hang on one judgment. It is made by the person in the position and can be attributed to his doing; it takes effect through the position, as an act of the construct, and the address of attribution is the construct. Both are there, and neither hides the other.

Two things must therefore be stated precisely. First, that the judgment is his does not mean that it issues from his own law. Perhaps in his heart he felt the daughter should have the spot, and judged by the rules all the same; perhaps he felt the rules really should be read this way. These are his, and the series does not avow them on his behalf. The Permits Office’s judgment, even if it coincides with what he holds, does not draw its effect from that, still less its legitimacy; and if his law disagrees with it, it does not lose its effect on that account. Second, his judgment lands on others: the daughter, the queue, the residents. Behind it stand the Permits Office and Apple City, and when necessary compulsion: whoever runs a stall without a permit can be fined, and cleared away. One person’s judgment can act on many others because a position stands behind it. That is why judgment within a government calls for a series of its own.

Aristotle said that political wisdom is concerned with action and deliberation, “for a decree is a thing to be carried out in the form of an individual act.” The permit examiner’s judgment is exactly such a decision. But Aristotle asked what kind of character makes a person deliberate well in the affairs of the city, and his answer lands on the one who judges; this series asks to whom, and for what, a judgment answers once it has been made, and its answer lands on the position and the construct. The paper drafts the strongest objection itself: give a person of poor character a structure of answering, and he only becomes better at evasion. Its reply: whether judgment is good does depend on the person, and the series does not deny it, only does not write it, since that would be writing an ought for a person. Evasion is precisely failing to answer the very matter one is asked about, and that is exactly what the burden of explanation is checked against. The structure of answering does not reward evasion; it brings evasion into view.

To the Very Matter Asked

The daughter went to her council member’s office to complain: “My mother kept a stall here for twenty years. Why should I have to start again at the back of the queue?” The office passed the complaint on to the Permits Office. Now someone has to answer. To whom: to those the judgment acts on who come to ask; here, her. Who answers, and what, is set out in three layers, the three layers of Institution Theory: the construct takes the ought, the position takes accountability, the person takes his own law.

To the construct: give judgment a place to be made, and let it be open to questioning. If the construct wrote every place down so rigidly that the permit examiner could only stamp clause by clause, judgment would not vanish; it would move to the positions that write the rules, and the account would move with it. If it wrote nothing and left each examiner to his own lights, judgment would have a place, but those who ask about it would find nothing to hold it against. To give judgment a place to be made is to set it where it can both be made and be answered for: which matters the rules settle, which are left to the person in the position, and on what basis and with what authority he judges. And once a judgment is made, the question asked of it must have somewhere to go, someone to receive it, and an answer. Institution Theory draws only the floor, that the executor may not be sealed off; going on to the positive requirement that judgment have a place to be made is the series’ own step.

To the position: the burden of explanation and the burden of bearing. Institution Theory says that an institution “bears the burden of explanation for what it has itself produced, and must answer the very matter it is questioned about.” What the daughter asked was why she should start again at the back of the queue. To answer “That is what the rules provide” does not answer it: the rules are precisely silent on succession, and what she is asking is how that silence is to be read. As for bearing, Institution Theory says: “the construct’s name does not shield an individual responsibility that has independent grounds, and the construct’s failure is not automatically transferred onto any one person.” How well the examiner read the rules is his position’s account; how the section chief signed off is the section chief’s; that the rules left succession blank belongs to the place that set them; how the Permits Office fitted these together is the Permits Office’s. Each is checked by its own source, authority, registration and conduct, and none settles another.

To the person: no ought is written. Whether the permit examiner should sympathize with the daughter, whether he should be anxious on her behalf, is his own law. The construct can require his position to explain and to bear; it cannot require him as a person to think in any particular way.

Not on the Checklist

The same year, an inspector from the Health Department made a routine inspection of a restaurant on Levee Street. The restaurant had put in a new kind of charcoal warming cabinet, which was not on the inspection checklist; every other item passed. The kitchen was small, with only one small vent. The inspector judged that in this closed kitchen the cabinet might let carbon monoxide build up, and ordered the restaurant closed on the spot, pending testing by the state. Whom it concerns: the owner and staff, and the diners. Which line: if the risk is real, the life-substrate of staff and diners; the closure itself acts on the owner’s business, an action that may be restricted. Baseline: the ordinary kitchen equipment the checklist lists.

The checklist is a rule, and it did not list this item; so the inspector had to judge whether it was left off because it poses no risk, or because whoever wrote the checklist did not think of it. The checklist cannot answer that. One could add an item, “other equipment that may produce harmful gases,” but that item too would need someone standing in the kitchen to judge whether this cabinet counts. Kant said as much: “A physician therefore, a judge or a statesman, may have in his head many admirable pathological, juridical, or political rules … and yet in the application of these rules he may very possibly blunder.” Here the two agree. Kant goes on to how judgment is formed, through examples and practice; the series does not write that. What it writes is that, since rules cannot judge in anyone’s place, a judgment once made must have a place where it answers.

Aristotle had his own answer for where rules do not reach: equity, “a correction of law where it is defective owing to its universality,” whose measure is “to say what the legislator himself would have said had he been present, and would have put into his law if he had known.” The inspector, on that measure, should ask how whoever wrote the checklist would judge standing in this kitchen. The series does not take this measure as the standard of judgment, nor does it set up another; it does not teach judgment. It asks something else: by this measure or another, once he has judged, his judgment must answer to the owner and the diners, and answer the very matter they ask. What the legislator would say if present can be one reason in his explanation; it does not answer for the judgment in its stead.

James Scott wrote that states simplify complex local life into forms and schemes in order to see and govern it, and that such simplifications lose the practical knowledge held by practitioners and by those governed. A checklist is such a simplification. But the other half has to be written the other way round, without a quiet substitution. In this case the practical knowledge lies first on the cook’s side: how the cabinet is used in this kitchen, when it is turned on and for how long, he knows. The inspector’s judgment, made standing in the kitchen, is not thereby practical knowledge; it may be another simplification, pressing the cook’s knowledge down on the state’s behalf. An officeholder’s judgment does not become practical knowledge because it was made on the spot, nor is it thereby exempt from questioning.

“I Was Only Following the Rules”

Back to the stall. When the complaint reached the Permits Office, the permit examiner said: “I was only following the rules. I referred it up.” He had. While handling the case he had entered a note in the internal system: “The rules say nothing on succession; suggest referring this up.” The section chief minuted four words: “Proceed per the rules.” The referral went no further. Asked about it, the section chief said: “That is how the rules are written.” The Permits Office’s reply said: “The permit examiner handled it according to procedure.” Three sentences, one thing: I was only following the rules.

Institution Theory has said that this sentence fails on both sides: “‘I was only following the rules’ fails on both sides: it cannot make the executor disappear into the construct, and it cannot make the construct disappear behind the executor.” In a government the shape is the same and the weight is greater: a government’s rules often have compulsion behind them, many people carry them out, and still more people are reached by them.

On one side, the permit examiner does not disappear into the construct. His referral is an entry of his own; but referring it up does not read the rules for him. On the last page of the rules there is a note: “Family operation may be recognized at discretion.” He did not see it. Missing the note is his position’s account, and it is not settled because the section chief minuted “Proceed per the rules.” Institution Theory says: “That a demand bears the name of procedure does not prove that complying with it is within the line.” On the applicant’s side nothing here touches the line, and the paper borrows the same reasoning, marking the step as its own: having followed procedure does not prove the judgment right. On the other side, the Permits Office does not disappear behind the examiner. The procedure was its configuration: its rules leave succession blank; the note is printed on their last page but missing from the handling guidelines it issued to examiners; referral stops at the section chief, who signed off without looking at the note. That is the section chief’s entry; that the configuration is as it is, is the Permits Office’s.

There is one more thing, on the construct’s side. Institution Theory says that “a construct’s configuration may not leave the executor both unable to refuse a demand and without any path for passing it into a reviewable procedure.” This is checked by the threshold of sealing off: refusal may still have consequences. If the examiner’s not complying would cost him an entry against his name, that is a consequence, not sealing off. If not complying meant losing the position, while the question “how should the rules be read?” had nowhere to be sent, the configuration would have reached that point. The paper does not rule on which side this case falls; it says what is to be checked. Not “why didn’t he refuse?” but “could he refuse, and when he could not, was there a path for this question to be sent to a place able to review it?” Where that path is missing, the line has already been crossed, on the examiner, and the account lies with Apple City. This line lies on the examiner, not on the applicant; when the opening said that neither granting nor refusing crosses a line, it meant the applicant’s side.

The Front Line

In the same week, two other people in Apple City each made a judgment. One was a patrol officer. At night, outside a bar on Levee Street, two people were arguing, and the argument rose to shoving. He had to judge: step in now, or watch a little longer? Too early, and a quarrel is treated as a fight; too late, and someone may already have been hit. The other was a social worker. A neighbor reported that a nine-year-old was being harmed at home. She had to judge: take the child away tonight, or come back tomorrow? The protocols of the Department of Children and Families required a decision within twenty-four hours of a report; each social worker was responsible for sixty families.

Together with the permit examiner, all three stand on the outermost front line. They sign no decrees and write no rules, yet the judgments they make every day land on the people of Apple City, often more directly than the documents the mayor signs. The series counts them among officeholders.

Front-line judgments invite two errors of bookkeeping. One is to charge the whole account to the front line. Suppose that she had not gone that night, because reports on three other families came in at once and she could go to only one first; the child was hurt again, and the department said: “It was her error of judgment.” (In fact she went that night.) The twenty-four hours and the sixty families were not set by her; the construct’s failure is not automatically transferred onto any one person. The other error is to push the whole account back onto the configuration. Still within that supposition, suppose she said: “Those are the protocols, that is the staffing. What could I do?” But which family to go to first, and on seeing what to decide on removal, were hers to judge; the configuration’s faults do not settle the account of her judgment. So the configuration must keep the two accounts apart: one falls on the front-line position, the other on the positions that set its rules and allocate its resources. The position that set the number sixty must explain on what grounds that number, what is to be done when it cannot be met, and where the account falls when something goes wrong.

Michael Lipsky called teachers, police officers and social workers street-level bureaucrats: they wield considerable discretion and in effect make policy; heavy workloads, ambiguous goals and limited resources push them to ration, to screen and to turn their dealings with people into routine. The paper agrees, and takes it as a premise; it parts in scope. Lipsky describes how the front line judges and why; the series writes how the accounts fall once those judgments are made. His routines are precisely why the configuration must answer: they are mostly not the front line’s own preference, but what workload and resources press out of it. To the objection that the two accounts are tangled and cannot really be cut apart, the reply is that what is required is not that they can always be cut, but that neither swallows the other. Where they cannot be told apart, it must be written where, and each side must explain; “they cannot be told apart” may not settle the account of either.

Heard, Not Moved

The daughter did not let it drop. The Permits Office has an objection procedure: whoever disputes an approval decision may object in writing within thirty days. The objection bypasses the permit examiner and the section chief and goes to a review officer in the Permits Office’s Regulations Unit, who does not belong to the Approvals Section. Whether to look into it, and whether it has been fully answered, are not for the Permits Office to decide; the reply goes out under the review officer’s signature, directly to the applicant, and the Permits Office cannot change it. She objected. The review officer replied in writing within the time limit and answered the very matter she had asked: why it had not been processed as a succession. The reply admitted that the examiner had not seen the note, and recommended that the Permits Office reconsider. The Permits Office did not. The procedure gives the review officer only the power to recommend; the Office gave its reasons in writing: “The note says only that it may be recognized at discretion, not that it ought to be; the original decision stands.”

The shape of this is worth a pause. The line is not crossed. The channel is there: she could object, someone received it, it was answered on time and to the point, and no one made things hard for her for objecting. Yet her question, though heard, moves nothing: that judgment remains that judgment. Over the past five years every objection this procedure received was answered on time, and not one changed a decision. Institution Theory, in its interface paper, had come upon this place and drafted an objection on Pettit’s behalf: “A mill opens a complaint channel, but the channel has no teeth, and the mill still dominates people.” It did not claim to have answered it; it registered a research candidate: “The line is not crossed and the channel is there: do those affected have actual control over the relevant power? Wide discretion within authority is one instance.” The Theory of Government takes this slot on, and the reason is on Levee Street: many of a government’s judgments are exactly like this. A series that checked only for line-crossing and only for channels would have nothing more to say about them; yet those they act on live their days inside them.

Two antecedents keep this slot from sliding elsewhere. The line is not crossed: a hidden line-crossing belongs to the first ought. The channel is there: independence, reach and no retaliation all hold, and the burden of explanation was met. Suppose the review officer’s reply had been a stock phrase that missed her question; what would be missing is the burden of explanation, under the fifth ought. Suppose the entrance to objections were held by the Approvals Section; what would be missing is independence, also under the fifth ought. This slot speaks of something else: the question reaches, it is answered, and still nothing moves. What is missing is not review; it is efficacy on the decision after review.

Two readings must be blocked. That no decision was reversed does not by itself show that control has failed: had the procedure the power of re-examination and simply met no material sufficient for reversal in five years, no reversal would prove nothing. No reversal is a signal to check, not by itself a diagnosis; what is checked is the structure. Nor is control a veto: suppose the residents across the street objected and demanded that every stall on the block be removed, and the Office did not comply; that would not show that they had lost control. So the paper asks three things here: who can activate which procedure on what matter; what actual efficacy the procedure has on reasons, actions, rules, or the holding of office; and whether the party questioned can cancel it on its own. On Levee Street the daughter can activate it without the Office’s consent; it has efficacy on reasons, since someone must answer and answer to the point, but none on actions, since whether to reconsider the Office decides alone; it offers no path for sending “how succession is to be read” for revision, and it touches no office. The Office cannot cancel it at the layer of reasons, since the reply goes out signed and unchangeable; at the layer of actions it can. The functions of control are answered in the sixth essay.

Pettit understands freedom as non-domination, and he himself requires that control be real: the people control the state through an influence that is individualized, unconditioned and efficacious. So the move from contestability to actual efficacy is not a step beyond him. The series’ own step is to join control to the three conditions of the channel. And it does not define legitimacy by control, as he does; a government’s legitimacy has its own necessary conditions, and control is a question that comes after them, within the line.

“I Never Agreed to It”

Back to the parking rule of the Prequel. The resident who got the ticket asked: “I never agreed to it. By what right does this rule govern me?” Whom it concerns: residents, shopkeepers and visitors in the old quarter. Which line: within it; parking is an action that may be restricted, and the rule touches no position that cannot be overwritten. Baseline: the old rule. The Conflict Treatise has answered this question between two people: the binding force of a rule they settle on after a quarrel “must pass separately through each party’s own law, or come from a procedure both have acknowledged”; “It is not a third law above the two.” The answer is clean, but in Apple City it is not enough. The parking rule did not pass through this resident’s law: he disagrees. Nor did it come from a procedure he acknowledged; he never told the Council, “Whatever you decide, I’ll accept.” A government’s decisions are often neither, and still they act on him: the ticket was issued, and an unpaid ticket can mean a towed car.

This acting is answered in three layers. The first, its factual effect: the ticket, the tow; this only says that it acted, not that it may. The second, under what authority and procedure it is valid: whether the Council has authority over street parking in the old quarter, whether the rule went through the procedure it had to, whether the fine was issued under it. This is validity, not legitimacy; a valid decision can be illegitimate. The third, on what grounds it may normatively be so. Here there are three necessary conditions; the step from the oughts to them is the series’ own. Not crossing the line: where there is violation, there is no ground that could answer “on what grounds.” Guaranteeing the recognition-based rights embedded: a government that holds the power side of a right it has embedded but does not let the right operate holds, at that point, power that is not legitimate, an undertaking the third essay sets out. Openness to questioning: a government can never be perfect, so a lapse is always somewhere; a decision with no channel through which it can be questioned leaves those it acts on with no institutional position from which to call for an answer and a correction, however reasonable its content. The three are necessary, not sufficient.

With the layers apart, one sentence can be said precisely: an individual’s disagreement does not by itself cancel the second-layer validity of a decision made under valid authority and procedure. The resident disagrees, and the rule takes effect for him all the same. The sentence has a range, and it must be stated at once: it speaks only of actions that may be restricted. Suppose the rule also had a clause doubling the fine for anyone who publicly opposes it. That clause penalizes not parking but his saying “I disagree”; it touches a position that cannot be overwritten, crosses the line, and is judged by the first ought. Within actions that may be restricted, his disagreeing does not leave him with nothing. He keeps his own law: he may disagree, may say so, may question. He did question, and the Council replied in writing with its reasons for the fees. And the rule still cannot legislate for him: it can require him not to park at night; it cannot require him to identify with the view that not parking at night is right. Whether anyone has the standing to require his assent to a judgment is a question for the critique of judgment; the parking rule does not take effect by requiring it.

Judging Wrongly

Two weeks after the restaurant was closed, the state’s test results came out: in that kitchen, with that ventilation, the charcoal warming cabinet did not exceed the carbon monoxide limit. Whether the inspector judged wrongly still has to be checked: what exactly he judged, whether his grounds stood at the time, whether the scope of the closure and its two weeks had grounds. Action Theory put one sentence up front: “Judging wrongly is possible; this series does not guarantee that anyone judges right.” So it is with a government’s judgments. Since rules cannot judge in anyone’s place, judgment can be wrong; this is not one inspector’s carelessness but the shape of judging.

Fallibility does not exempt it from answering. That does not mean the inspector must be punished. What he saw was a kitchen with one small vent and a cabinet not on the checklist; to close first and wait for the test may have had its reasons. How his entry is recorded is checked by his authority and the facts he saw then. What not being exempt says is this: a judgment can be wrong, and it lands on others, so someone must explain, and the loss must be booked. The Health Department must explain on what grounds it judged there was a risk, why it chose closure rather than some other measure, and whether, now that the test does not support it, the original judgment still stands. The owner’s two weeks of losses must have a place where they are reckoned: whether they are compensated, by whom, how much, and on what grounds. Answering is not waived because there were reasons at the time, nor does it wait until the error is established; whether the judgment was wrong is exactly what the answering is there to check.

Even a right judgment must be able to say on what grounds; fallibility is a further reason. Because judgment can be wrong, and a wrong judgment lands on others, there must be a path by which those it acts on can ask and by which errors can be found and corrected: the owner can ask, the Department must answer, and once the results are in, the closure order must be lifted. Going from fallibility to the requirement that public judgment be open to questioning is the series’ own step. Said through to the end: no department of Apple City will always judge right, and neither will its government. The series undertakes the sentence that a government can never be perfect. It is not a discouraged sentence; it says that being questioned is the normal state, not the exception, and that a claim to perfection is itself the place where questioning is due. Institution Theory says that an institution “may not close off the conditions of questioning and correction on the grounds that it claims to cultivate, that it has already reformed, or that its aims are right.” Were the Health Department to say, “Our inspections are never wrong, so there is no need for review,” that would be the first sentence to question.

“The System Gave the Score”

An Apple City ordinance provides a winter heating subsidy for low-income households, applied for at the Housing Department, with a decision within thirty days. In its second year the Housing Department contracted a scoring company to make the initial assessment by algorithm: each application got a score, low scores went automatically into a “further documents” queue, and assessors looked only at the high ones. In the third winter, one applicant had waited fifty days when she found her application in that queue, without knowing what she was to supply. She went to ask, and the person at the counter said: “The system gave the score.” Whom it concerns: the applicants given low scores. Which line: none is crossed; what is at stake is whether a recognition-based position the government has embedded can actually operate. Baseline: assessors reviewing each application themselves.

“The system gave the score” is a new form of “I was only following the rules.” It pushes the judgment onto a machine, as if the machine had judged and people only passed the word along. Power Theory says: “The algorithm does not wield power; designers, deployers, invokers, and maintainers wield power through the algorithmic apparatus.” In this case several things were set by people: the Housing Department decided to hand the initial assessment out; it agreed with the scoring company on what to score and how; it laid down that low scores go to further documents and that assessors look only at high ones; and on those scores it kept a family waiting fifty days. Each step is a judgment made at a position. The bearing falls on these positions, not on the algorithm.

And the scoring company? It holds its own accountability toward the Housing Department under their contract, as one construct to another; it does not answer to applicants in the Department’s place. When an applicant asks “Why this score?”, it is the Housing Department that must answer. If it cannot, because it does not itself know why the algorithm gave this score, then it has configured itself into a construct that cannot answer. When a construct hands its judgment to a place it cannot itself explain, its burden of explanation is not handed over with it; it simply goes unmet, and where it goes unmet, the account is the construct’s own. Whether an institution run by machines is still an institution in the full sense, the series leaves open; until that is answered, a machine holds no bearing.

With Weber

The paper ends with Weber, who in Politics as a Vocation divided responsibility in a political order between two kinds of person. “The honor of the official lies in the ability, when, despite his dissent, the authority above him insists on an order that seems to him wrong, to carry it out on the responsibility of the one who gives the order, conscientiously, exactly as if it corresponded to his own conviction.” The political leader’s honor is the opposite: an exclusive personal responsibility for what he does, one “which he cannot and may not refuse or shift onto others.” The two accounts agree on three things: neither lets those above push the account onto the one who carries it out; both leave the one who carries it out a place to speak; both know that nothing gets done without execution.

They part at three places. First, how responsibility is divided. Weber divides by kind of person; once the official has stated his dissent and complied, responsibility for the content of the order lies with the one who gave it. The series divides by position and source. Suppose the section chief had minuted not “Proceed per the rules” but “Reject all successions; do not apply the note,” and the permit examiner, having read the note, stated his dissent; the chief insisted, and he complied. For Weber the responsibility for the rejection lies with the chief. In this series the chief’s entry is not reduced in the least, and the examiner’s is not moved away by his compliance: he had read the note, knew the judgment could have gone otherwise, and compliance does not prove it right. The one who gives the order bears his share; the one who complies bears his.

Second, where dissent is sent. Weber’s dissent goes upward, to the very superior who insists on the order, and that superior is exactly the party the question is about. What independence guards against, Institution Theory says, is “the party under review sealing off review of itself on its own.” The permit examiner’s referral had this shape: it ended at the section chief, who was both the one questioned and the one who alone decided whether the matter was over. It may be useful; it is not independent; checking this path for independence is the series’ own step. The review officer was different. Third, “as if it corresponded to his own conviction.” Weber’s word is “as if”: carrying it out as though from conviction, without discount, not exchanging his conviction. He made this discipline the official’s personal honor, an ethic; the series sets up no honor for persons. Apple City can require the examiner’s position to proceed per the rules; it cannot require him to take the rules as his conviction: “A construct may require a position to do something; it may not require a person to think something.” When he complies, it can be mere compliance; what he holds in his heart is his.

The paper drafts Weber’s strongest objection: without the official’s discipline the apparatus falls apart; give every position its own answering and the one who carries out a path to review, and administration bogs down, anyone can stall an order with “I want it reviewed,” anyone can say “this entry is not wholly mine,” and once responsibility is divided no one is truly responsible; Weber concentrates responsibility in the leader precisely so that people know whom to go to. The reply: the path is needed only where the one who carries out cannot refuse, and refusal may still have consequences; an apparatus without it is not more stable, since wrong orders are complied with all the way down and the accounts are all the harder to sort out afterward. Dividing the accounts is not dispersing responsibility; the leader’s entry is not reduced at all, and what is blocked is the transfer in the other direction. Knowing whom to go to matters, and in this series that is the business of revoking authorization, in the sixth essay. Discipline and answering do not collide: the examiner can keep discipline and proceed per the rules, and the judgment he complies with must still be able to answer.

The stall on Levee Street, the restaurant and the heating subsidy will come back in the essays that follow. What this one has shown is limited: a government’s judgment is made by a person in a position and takes effect in the government’s name; it answers in three layers; rules cannot do it in anyone’s place, and “I was only following the rules” cannot shield it; the front line is included; where the line is not crossed and the channel is open, whether those affected have actual control is still a question; a decision acts without consent and must be answered in three layers; judgment is fallible; and the bearing does not go with it to a machine. What good judgment looks like, the series does not write.