Policy Is What Arrives: Lipsky, Ishiguro, Kurosawa, and the Discretion Compliance Cannot Name

By my second year teaching philosophy at Het Amsterdams Lyceum, I had stopped pretending I was delivering the curriculum.

I had twenty-seven students in 4V, forty-five minutes a period, and a syllabus written by people who had one main priority: ensuring that the full Western tradition would be taught, no essential texts excluded. The curriculum specified a careful trajectory through the so-called eindexamen canon or eindtermen — that year, this included Hegel and Heidegger on the essence of technology, with side excursions into Marx, Mill, and the philosophy of science. The trajectory assumed a reader who had completed the previous week’s reading, a class capable of sustained dialectical attention, and a teacher with the time to honour the complexity and subtlety of the text and argument.

None of those assumptions held in the room when the kids arrived at Monday morning 08:30. 50 to 75% of the class had not done the reading. Four boys at the back had decided in September that philosophy was beneath their attention; one of them was visibly struggling at home, openly defiant, and I had not yet found the time to find out why. The bell would ring in forty-three minutes and I had to choose what philosophy was going to be that morning, and which one or two interventions, pedagogical or didactical, I could land with any chance of success.

What I taught was not what the syllabus said. It was what the syllabus said, plus what the room could absorb, minus what would lose me the back row, plus the small accommodations that would let the strongest students leave with something they could think about on the tram home. Often the latter happened outside of the context of the classroom: in an email, or a question after the lesson ended. Minor, but important moments. The accommodations were unsanctioned. They were also the most important aspect of my work. The syllabus described the policy. I delivered it.

I did not, then, have a name for what I was doing. Years later, in a different vocation, I read Michael Lipsky’s Street-Level Bureaucracy and found the diagnosis. Lipsky writes about teachers, social workers, police, public defenders — the people who deliver the state to the citizen one encounter at a time. His central claim is that these workers do not implement policy. They make it, action-by-action, through the accumulated weight of their improvisations under pressure. The official policy describes an aspiration and objective. The actual policy is whatever the worker, with too little time and too many cases and/or clients, was able to assemble at the point of contact.

I was, although I would not have used the term, a street-level bureaucrat.

Lipsky’s Paradox

Lipsky’s argument, first published in 1980 and refined for thirty years afterward, is uncomfortable for anyone who builds governance frameworks. It is also, I am convinced, the most accurate description of how complex systems actually function under conditions of constraint on the street-level.

The structural diagnosis runs as follows. Public service work — teaching, policing, social work, frontline healthcare, and a dozen analogous roles in regulated industries — operates under five conditions that converge to make rule-following impossible:

1. Resources are chronically inadequate relative to the tasks workers are asked to perform.
2. The demand for services tends to increase to meet the supply.
3. Goal expectations for the agencies in which they work tend to be ambiguous, vague, or conflicting.
4. Performance oriented toward goal achievement tends to be difficult if not impossible to measure.
5. Clients are typically nonvoluntary; partly as a result, clients for the most part do not serve as primary bureaucratic reference groups.

These conditions are not, Lipsky insists, the result of poor management or insufficient funding, although poor management and insufficient funding are present in these sectors. These conditions are structural. The demand spiral alone — the finding that any expansion of service produces a corresponding expansion of demand — guarantees that no realistic budget will ever close the gap. He gives the example of the Long Island Expressway: every additional lane, intended to relieve congestion, induced enough new traffic to restore the original congestion at higher volume. Welfare offices, schools, emergency rooms, legal services all behave the same way. The capacity to serve creates the population to be served.

Inside this structural impossibility, the street-level bureaucrat has two choices. She can attempt to follow every rule, in which case she will accomplish almost nothing, fail her caseload, and be either fired or institutionally invisible. Or she can develop what Lipsky calls coping mechanisms: the routines, simplifications, triage heuristics, and informal categorisations that let her finish the day. These coping mechanisms are unsanctioned, indispensable, and the way the actual work gets done on the level where it matters most.

The decisions of street-level bureaucrats, the routines they establish, and the devices they invent to cope with uncertainties and work pressures, effectively become the public policies they carry out.

This sentence is this essay’s hinge. The policy a citizen encounters is not the policy the legislature passed. It is the policy the worker assembled, in the moment, to discharge an impossible mandate within the available time with the resources at hand. The teacher who has given up on the back row is not failing to implement the curriculum. She is the curriculum. And the social worker who has informally divided her caseload into active and inactive piles has, in that division, become the agency’s caseload-management policy.

The deeper irony, which Lipsky names and the rest of his book unpacks, is that the agency depends on these coping mechanisms while officially denying them. Brutality is contrary to police policy, but, as Lipsky observes, “a certain degree of looking-the-other-way on the part of supervisors may be considered necessary to persuade officers to risk assault.” The Wire dramatises this dynamic at its hardest. When a young suspect strikes an officer in the field, the response — collective, immediate, well past anything the use-of-force policy would sanction — comes even from officers who, in any other setting, would condemn it themselves. The show does not endorse what it depicts; the moral cost is shown unsparingly. But the scene is structurally exact to Lipsky’s point: the unsanctioned group reaction is the policy that arrives at that street corner, and the institution that condemns it on paper depends, in practice, on the cohesion that produces it.

Examples in my industry may be less impactful, but they still follow the same internal logic. The validated environmental monitoring system depends on the cleanroom operator’s tacit knowledge that a particular particle counter spikes briefly at every shift change as gowning fibres settle, even though no field in the EM platform distinguishes that transient from a real contamination excursion. The compliance framework that is followed to the letter cannot function and the compliance framework that functions cannot be followed to the letter. The gap between these two dimensions is filled, every day, by workers whose unsanctioned discretion keeps the system operational and whose unsanctioned discretion the system, when it goes wrong, will often treat as a deviation.

Lipsky calls this the “essential paradox” of public service work. The work is “highly scripted to achieve policy objectives that have their origins in the political process,” and, simultaneously, “requires improvisation and responsiveness to the individual case.” Both are true. Neither can be fully honoured. The worker carries the tension in her body, every day, until she either burns out, leaves, or — most often — settles into what Lipsky calls a “corrupted world of service”: a stable equilibrium in which she does the best she can, accepts the gap between aspiration and reality, and processes the next case.

The two literary figures I want to set against this framework respond to it in opposite ways. Stevens, in Ishiguro’s The Remains of the Day, refuses the discretion that is structurally his. He insists, against all evidence, that he is merely executing the policy of his employer. Watanabe, in Kurosawa’s Ikiru, has spent thirty years doing the same — until cancer forces him to recognise, with months left, that the discretion he refused to exercise was the only thing that could have made his work mean anything. Both men work in street-level bureaucracies, in Lipsky’s structural sense. One mistakes the rule for the work. The other recognises, almost too late, that the rule was never going to deliver the work — and that the work, if it was going to happen at all, was going to be made by him, and him alone.

Stevens Will Not Choose

Stevens is the most carefully constructed unreliable narrator in late twentieth-century English fiction, and the reliability he is failing at is moral self-knowledge. Stevens is a traditional English butler, and the novel records a six-day motoring trip he takes in July 1956, alone, through the West Country. He undertakes the trip ostensibly to recruit a former housekeeper back to Darlington Hall, but, as we find out, actually to come to terms with what he has spent his life doing.

The core quality required for his vocation, as he understands it, is dignity. The word recurs throughout the novel as a load-bearing concept; Stevens has thought about it more than he has thought about anything else. It is through the concept that he constructs both his own identity and his relationship with his father, who was himself a butler. After a long discussion of his father’s career, he offers his definition:

‘dignity’ has to do crucially with a butler’s ability not to abandon the professional being he inhabits. Lesser butlers will abandon their professional being for the private one at the least provocation. For such persons, being a butler is like playing some pantomime role; a small push, a slight stumble, and the facade will drop off to reveal the actor underneath. The great butlers are great by virtue of their ability to inhabit their professional role and inhabit it to the utmost; they will not be shaken out by external events, however surprising, alarming or vexing.

Read against Lipsky, this is a definition of anti-discretion. The professional self, in Stevens’s account, is the part that does not respond to the individual case. The dignity of the role lies in refusing the responsiveness that the role’s actual conditions demand of the moral character of the worker. Stevens’s whole life is an argument that he is not a street-level bureaucrat. He is an executor of his lordship’s policy. He is not making policy; he has become the lordship’s policy made flesh.

This is best demonstrated in one of the novel’s central scenes. In the mid-thirties, Lord Darlington — under the influence of an antisemitic acquaintance — instructs Stevens to dismiss two Jewish housemaids, Ruth and Sarah, who have been on the staff for six years. Stevens carries out the instruction. The housekeeper, Miss Kenton, who supervises the maids and who knows them, refuses to accept it:

“Does it not occur to you, Mr Stevens, that to dismiss Ruth and Sarah on these grounds would be simply — wrong? I will not stand for such things. I will not work in a house in which such things can occur.”

“Miss Kenton, I will ask you not to excite yourself and to conduct yourself in a manner befitting your position. This is a very straightforward matter. If his lordship wishes these particular contracts to be discontinued, then there is little more to be said.”

Stevens speaks to her, throughout the scene, in the voice of someone reading aloud from a manual. We must not allow sentiment to creep into our judgement. His lordship is somewhat better placed to judge what is for the best. I leave it entirely to yourself whether or not you inform them beforehand. He is, he insists, simply doing his job. The girls are dismissed. They leave the house in tears. Miss Kenton, who has threatened to resign, does not — and Stevens, in retrospect, teases her about this for months, as if her failure to act on her conscience confirms his own discretion-free position. For him, it’s a token of his professionalism, his ‘dignity’ as a butler.

A year later, Lord Darlington — sobered, having “severed all links with the blackshirts” — asks Stevens whether the girls can be traced and recompensed. They cannot. Stevens reports the conversation to Miss Kenton in a misguided attempt at consolation, and she breaks: she had wanted to leave, she had thought of it daily, she had felt cowardly, and she had stayed because she had nowhere to go. She felt an accomplice and had felt the moral objection, but had not acted upon it by leaving. The dismissals — and Stevens’s part in them — have been, for her, a moral wound that did not heal.

The structural reading is precise. Stevens did exercise discretion, although he himself prides himself on his ‘dignity’ and detached professionalism. There is no version of “carrying out his lordship’s instructions” that does not involve a hundred small choices about how to carry them out — the tone with which the news is delivered, whether the maids are given any warning, whether their references are warm or perfunctory, whether Stevens stands quietly with Miss Kenton against an instruction both know to be wrong, or polices her into compliance. He chose throughout. He cannot say he chose, however, because the entire architecture of his self-respect depends on the claim that he did not. I trusted in his lordship’s wisdom. All those years I served him, I trusted I was doing something worthwhile. I can’t even say I made my own mistakes.

That sentence, delivered to a stranger on a pier on the last evening of the trip, is Stevens’s confession. Stevens is mourning the discretion he refused to acknowledge. He is mourning, in Lipsky’s terms, the policy he made and would not own. His legitimacy as a moral agent depended on his recognising what his role required of him. His legibility as a butler — his perfect dignity, his refusal to abandon the professional being — required him to deny it. He achieved the second at the cost of the first.

The cost is paid foremost by Ruth and Sarah, who do not return. It is paid by Miss Kenton, who carries the moment and moral injury that resulted from it for thirty years. It is paid by Stevens last, and most quietly: a man who has dedicated his life to a profession whose constitutive principle, not abandoning the professional being he inhabits, turns out to have been a refusal to be a moral being at all. He has refused to ‘act’ in the Arendtian sense. The novel ends with him on a bench in the dusk in Weymouth, watching the pier lights come on, deciding that he should learn a quality that he perceives other humans to enjoy: to banter. Not because it might connect him and lend him agency and connection to the world outside his vocation he rejected, but because his new American employer expects it. The décalage between the magnitude of the realisation and the smallness of the resolve is the novel’s last cruelty. The professional being closes again. The actor underneath is not permitted to speak. Stevens will not act.

Watanabe Stamps One Last Time

Akira Kurosawa’s Ikiru (1952) opens with an X-ray of Kanji Watanabe’s stomach and a narrator who tells us, in the second sentence, that the man we are about to meet is already dead. He just does not know it yet. He is the Section Chief for Public Affairs at Tokyo’s City Hall. He has stamped documents at the same desk for thirty years. He has never taken a day off. The narrator informs us — over a shot of Watanabe surrounded by towers of paper, his face as still as the documents — that this zombie-like life is what passes for living.

The film’s first major sequence is a Lipsky thought experiment performed as cinema. We arrive at the city hall where we see a big sign that says ‘Here is where city hall meets its citizens’. A delegation of women from a poor neighbourhood arrives at the city hall to petition for a fetid mosquito-breeding cesspool to be drained and turned into a children’s playground. They are sent to Public Affairs. Watanabe’s clerks send them to Public Works. Public Works sends them to Parks. Parks sends them to Sewerage, which sends them to Roads, which sends them to Education, which sends them to Health, which sends them to Fire Prevention, which sends them to City Planning, which sends them to the Ward Office, which sends them, eventually, back to Public Affairs. They have spent the day climbing and descending stairs. The film cuts between their faces, growing tired, and the perfect bureaucratic courtesy of each official explaining that this is, regrettably, not his department. By the time they arrive back at Watanabe’s desk, the system has metabolised them. Their request has been processed without anything happening. This, Kurosawa is showing us, is policy as actually delivered. The legibility is total. The legitimacy is absent.

What changes Watanabe is the diagnosis, and his encroaching death. He has stomach cancer; he has, perhaps, six months. The official record will show no event corresponding to this fact. He continues to be Section Chief for Public Affairs. But the discretion he has refused to exercise for thirty years suddenly becomes the only resource he has to establish his agency. He cannot get back the years. He can use the days remaining to do something. After a brief, despairing detour through the Tokyo nightlife and a poignant friendship with a young woman from his office whose vitality he cannot inherit, he returns to the office, retrieves the women’s petition from the bottom of the stack, and starts to push.

Kurosawa films the pushing as bureaucratic combat. Watanabe walks the petition from department to department himself. He is rebuffed, redirected, ignored. He bows lower than is required. He waits in corridors. He goes over heads when he must, accepts public humiliation when he must not. A deputy mayor with political ambitions tries first to bury the project, but then, when it nears completion, he tries to take credit. Watanabe says nothing. His silence — a silence Stevens would have recognised as professional dignity — is now in service of something his profession had no language for. He is using the capacities of the role to deliver a service the role was not designed to deliver. He is, for the first time in thirty years, behaving like the policy-maker Lipsky says he was and should have been the entire time.

The playground is built. The film shows it briefly: small swings, a sandpit, snow falling, the women standing at the edge in worn coats, looking at the thing that has, against the structural logic of city hall, come into being. Watanabe sits on one of the swings and sings, very quietly, “Gondola no uta” — Life is brief, fall in love, maidens, before the crimson bloom fades from your lips. He dies there. The snow continues.

The film does not end with this scene. It ends with the wake. Watanabe’s colleagues — ranked at the funeral by seniority, drinking sake, talking — slowly piece together what happened. They had not understood, and some still do not. They reconstruct the months: the inexplicable persistence, the willingness to be humiliated, the refusal to defer the petition any further. The deputy mayor explains, with great dignity, that the playground was the result of his department’s vision and that Watanabe’s role was secondary. The colleagues, drunk, mostly accept this. One younger clerk does not. He stands and shouts that they all know what Watanabe did, that they should remember it, that they should each, at their own desks, do the same. The room goes quiet. They drink. They go home.

The film’s final sequence is the next morning. The young clerk who shouted at the wake is back at his desk. A petitioner approaches. He hesitates — and then sends the petitioner to another department. The bureaucracy reabsorbs him. The system has digested its anomaly.

Kurosawa’s bleakness is precisely calibrated. It is not that Watanabe failed. The playground exists, and children are using it. It is that Watanabe’s discretion was individual and could only be individual. He used the structural latitude that Lipsky describes to its maximum capacity — the latitude every street-level bureaucrat possesses, which is the latitude that constitutes the actual policy of any complex agency — to make one good thing happen before he died. The system is unchanged. The next morning, the petitions are routed in the usual way. The next playground will not be built.

What the Framework Depends On

Here is where the reading turns toward my actual work, and where some of the discomfort with Lipsky sharpens.

I build compliance and quality systems for regulated industries — biotechnology, vital-sector cyber, public administration. The frameworks I implement are detailed, rigorous, and largely well-designed. They are also, in Lipsky’s sense, specifications for impossible work. The validation lifecycle assumes a laboratory whose analysts have time to follow each SOP precisely. The incident reporting framework assumes a security operations team whose pattern-recognition can be exhaustively articulated in advance. The continuity framework assumes a recovery team whose decisions during a real incident can be reduced to a runbook. None of these assumptions fully hold. They cannot hold. The work, when it is done well, is done by people exercising structural discretion that the framework does not formally acknowledge.

The frameworks depend on this discretion. They cannot say so. The auditor cannot find “the operator’s tacit feel for which particle-counter spikes are IPA on the sensors and which are real excursions” in a controlled document and tick a box for it. The regulator cannot inscribe the latitude that an operator needs to make a real-time triage call between incident classifications. So the framework is written as if the work were rule-following, and the work is done as if the framework were aspirational, and the gap between the two is filled, every day, by workers whose unsanctioned coping is the only reason the system functions.

This is dangerous in two directions.

It is dangerous when the frameworks are too tight — when the validation protocol, the SOP catalogue, and the audit programme leave so little space for discretion that the work cannot be done at all without procedural deviation, and the workers begin to live in a state of permanent low-grade non-compliance that everyone understands and no one acknowledges. The deviation log fills with entries that the team treats as routine, or even worse: the procedural deviation is not captured and remains an ‘open secret’ to the organisation, but is hidden from the auditor. The regulator, with diligence, scales the response based on what he finds. The framework gets even tighter. The deviation rate rises, but by itself fails to improve the street-level reality.

It is also dangerous when the frameworks are too loose — when the discretion is unbounded and the only safeguard is the worker’s professionalism and conscience. This is Stevens’s failure mode at scale. The vital-sector operator who reports an incident in the minimum terms required, because no one has clarified what fuller reporting would serve. The validation engineer who signs off because the document says she can, time pressure mounts, and colleagues say she must. The discretion is exercised, the policy is made — but it is made in service of the framework’s optics rather than its purpose, and no one in the room is positioned to insist on the difference.

The corrective is neither. Lipsky’s argument, read carefully, is not that frameworks should be abandoned. He spent his career working to improve them. The corrective is to design frameworks that acknowledge their dependence on the discretion they cannot fully specify — and that build into themselves the legitimacy that the workers’ discretion needs in order to be exercised properly. A validation protocol that asks what operational knowledge exists that the document does not capture — and that authorises the team to depart from it when necessary. An incident response framework that includes a sanctioned mechanism for the operator’s mētis to be heard before the report is filed. An audit programme that tests whether the controls are alive and verifiable in the room, not only whether the documents describe them. An organisational culture that protects the kind of judgment Watanabe finally exercised on behalf of the petitioning women, before the man exercising it has to be dying to feel free to do so.

This is the work I am supposed to be doing, and it is harder than what I usually do, which is to write the documents in the validation framework as cleanly as possible and trust the auditors to trust my words and the intentions of my actions. The harder work involves admitting, in the design of the framework itself, that the framework is a partial description of work the framework cannot and should not fully specify. That admission is uncomfortable. It is also, structurally, the only thing that distinguishes a system whose legibility is in service of its legitimacy from a system whose legibility has begun to substitute for it — a system, in the language of an earlier essay in this series, that has slid into self-referential compliance.

Stevens is what self-referential compliance looks like in a single individual. The professional being is intact, the dignity preserved, the documentation — his own internal monologue — in order. But the dismissed maids, who do not return, are part of the record his framework cannot read.

The Last Bell

There is a memory I keep returning to from the Lyceum. A boy in the back row, in February, who had not spoken in class for two months, who had done nothing other than sabotage my classroom monologues, raised his hand during a discussion of Mill’s harm principle. He had a question that was better than the question I had been hoping someone would ask. I had thirty seconds to acknowledge it, eight minutes left in the period, and another class arriving. I asked him to stay after, and he did not. He was not the type. He left with the bell. I never got the question back.

I thought of this example when writing this essay, because he is the part of the policy I made that month that does not appear in any record. The official curriculum was delivered. The exam results, in due course, were respectable, good even. The school’s accountability documents — the kind of documents that, in another life, I now write — would have shown nothing irregular. The boy in the back row, who had a real question and one window in which to ask it, is the mētis of the room: the knowledge that had no field, the moment that escaped the framework, the policy I made at that specific moment, by failing to make a different one.

Lipsky’s diagnosis is structural. It is also, when held against one’s own work for long enough, personal. The teachers, social workers, validation engineers, and section chiefs of the world are not deviating from policy when they triage, simplify, or improvise. They need to make these policy decisions at the street-level, otherwise they could not get their work done. The question for the framework architect — the role I now occupy — is not how to eliminate this, but whether the framework I build is honest about what it depends on.

Stevens never asks the question. He goes back to Darlington Hall and will try to teach himself to banter. Watanabe asks the question in the last six months of his life, exercises the discretion that was always his, and dies on a swing in the snow with the playground built behind him. Neither path is enough. The colleagues at the wake are right that what Watanabe did was admirable. They are wrong to think his individual heroism is what changes anything. What changes things is whether the next morning’s clerk can do the same without dying — whether the framework, when the petitioner walks in, has been built to let him.

The bell will ring. The classrooms will be full. The policy will continue to be made, regardless of the documents written by decision-makers. The only question is whether we are going to acknowledge that the actual practitioners are the ones making it — and whether the systems we build know enough to leave room for the work they could not, on their own, perform.

Extended Reading & Watching

• Lipsky, Michael. Street-Level Bureaucracy: Dilemmas of the Individual in Public Services. Russell Sage Foundation, 1980 (30th Anniversary Edition, 2010).
• Ishiguro, Kazuo. The Remains of the Day. Faber & Faber, 1989.
• The Remains of the Day. Dir. James Ivory, 1993. Merchant Ivory adaptation with Anthony Hopkins and Emma Thompson.
• Kurosawa, Akira, dir. Ikiru. Toho, 1952.
• Living. Dir. Oliver Hermanus, 2022. Kazuo Ishiguro’s screenplay relocates Ikiru to 1950s Whitehall, with Bill Nighy as the dying section chief.
• Brodkin, Evelyn Z., and Marston, Greg, eds. Work and the Welfare State: Street-Level Organizations and Workfare Politics. Georgetown University Press, 2013. Lipsky’s framework applied to contemporary welfare reform.
• Dubois, Vincent. The Bureaucrat and the Poor: Encounters in French Welfare Offices. Ashgate, 2010. Ethnography of French welfare offices in the Lipsky tradition.
• Entre les Murs (The Class). Dir. Laurent Cantet, 2008. A year in a Parisian middle-school classroom that won the Palme d’Or.
• Short Term 12. Dir. Destin Daniel Cretton, 2013. A short-term residential facility for at-risk youth, seen through the staff, who themselves were/are at risk.
• I, Daniel Blake. Dir. Ken Loach, 2016. The British welfare bureaucracy from the claimant’s side.
• The Wire, Season 1. Created by David Simon, HBO, 2002. Baltimore patrol policing as Lipsky case study at its hardest: the formal use-of-force policy, the unsanctioned group response when an officer is struck, and the institution’s quiet dependence on the cohesion that no formal document can describe.
• The Wire, Season 4. Created by David Simon, HBO, 2006. Baltimore public schools as a Lipsky case study: teachers as the policy, the curriculum as the fiction, the test results as the framework that is both indispensable and incapable of seeing what is in the room.
• Graeber, David. The Utopia of Rules: On Technology, Stupidity, and the Secret Joys of Bureaucracy. Melville House, 2015. The chapter “Dead Zones of the Imagination” is the closest companion in spirit to Kurosawa’s first act.