The Faustian Clinic:Mastery, Suffering, and the Bargains Physicians Make

Figure 6

The Bargains We Make with the System

In the last act of Part II, Faust is a hundred years old and has become a developer. With Mephistopheles as his contractor, he is reclaiming land from the sea, building dikes and harbors, preparing living space for millions. One thing spoils his view. On a dune inside his estate stands a hut with a few linden trees and a small chapel, the home of an aged couple, Philemon and Baucis, who once sheltered a shipwrecked stranger and who decline to be moved. The sound of their chapel bell torments him; it reminds him that his possession of the world is incomplete. He has chosen a nice little property for them elsewhere. He tells Mephistopheles to see to the relocation. Mephistopheles and his three strongmen see to it: the hut burns, and the old people and their guest die in it. Faust is appalled. "I wanted an exchange, not a robbery," he protests (line 11371) [1]. Marshall Berman, in the finest modern reading of this act, called it the tragedy of development: the visionary who genuinely intends the good of millions, who delegates the dirty work, and who discovers that the project has its own logic, in which whatever obstructs the flow is simply cleared [19].

I know of no better parable of what it is to practise inside a contemporary health system. The hospital is an extraordinary moral achievement. It gathers diagnostic capacity, monitoring, specialist expertise, surgery, critical care, imaging, pharmacy and rehabilitation into one coordinated organism, and I have no nostalgia for the world before it. Yet its architecture and its clocks reveal its assumptions. Beds must turn over. Length of stay must fall. Documentation must close. Discharge must occur, metrics must be met, risk must be stratified, and patients must move. The institutional patient is therefore always in danger of becoming an object of throughput. The sick person experiences time as rupture; the hospital experiences time as flow. For the patient, Tuesday is the day someone said that the biopsy showed cancer. For the hospital, Tuesday is hospital day three. For the patient, the room is the place where a life divided into before and after. For the institution it isbed 412B. Neither perspective is false, but only one of them can become total without noticing what it has done.

The patients who do not move, the frail, the undiagnosable, the socially stranded, the dying who are dying too slowly, are our Philemon and Baucis. No one wishes them harm. They are in the way of something that is, in aggregate, a good.Relman saw the outline of this nearly half a century ago when he warned of a "new medical-industrial complex" whose commercial logic would not remain politely external to clinical judgement [20]. Pellegrino described the result as a paradigm shift from a professional to a market ethic, with moral consequences for everyone inside it [21]. My own critique of capitalism in health care argued that the physician's fiduciary relation to the patient is now routinely subordinated to obligations the patient never sees [22]. What I want to add here is the Faustian texture of how this happens to an individual doctor, because it does not arrive as a single dramatic signature in blood. It arrives as a series of reasonable exchanges. We accept employment and surrender the practice, and with it the power to decide how long a visit last. We accept the productivity formula and begin, without ever deciding to, to see the schedule in relative value units.

We accept the electronic record and its promise of legibility and find that for every hour with patients we now spend nearly two at the screen [23]. We accept the quality metric and start treating the number. We accept the word "provider." We accept the prior authorisation, the formulary, the script for the difficult conversation, the fifteen minutes. Each exchange buys something real: security, reach, data, protection from liability, a salary. Tausch wollt' ich, wollte keinen Raub. We wanted an exchange. We did not intend the robbery, and we were not in the room when it occurred. The clinical literature has lately found a name for what follows. Dean, Talbot and Dean argue that what is called physician burnout is better understood as moral injury: not a failure of individual resilience but the predictable wound of being repeatedly unable to give the care one knows the patient needs, because of constraints one serves and did not choose [24]. The epidemiology of burnout is by now dismally familiar, and its structural causes, workload, clerical burden, understaffing, lost autonomy, production pressure, are indispensable to any honest account; they must not be psychologized away [25].

But Faust suggests a further layer, and I have tried to describe it before as the physician's existential crisis [26]. A healer trained to derive identity from efficacy is uniquely vulnerable when efficacy is institutionally frustrated. The Faustian response to that humiliation of agency is more striving: work faster, learn more, stay later, answer every message, carry every outcome, become indispensable. The result is predictable. The self whose worth depends on inexhaustible efficacy eventually meets the body. Sleep is required. Attention is finite. Patients die, mistakes occur, families remain dissatisfied, and the institution absorbs the effort without gratitude. The physician discovers, often painfully, that finitude was never only the patient's problem.Goethe stages that discovery with terrible economy. After the fire, four grey women approach Faust's palaces at midnight: Want, Debt, Need and Care. Three cannot enter a rich man's house. The fourth, Sorge, slips in through the keyhole. Faust refuses to acknowledge her power, and she breathes on him and blinds him. Blind, he is more driven than ever. He hears the clatter of spades outside and rejoices that his laborers are digging the great drainage ditch that will complete the work.

They are digging his grave; the German puns grimly on Graben and Grab. It is the most exact image I know of the late-career physician inside the productive machine: sightless, exhilarated by the sound of activity, unable to tell the project from the burial. Guggenbühl-Craig warned long ago that the shadow of the helping professions is power, and that the healer who cannot acknowledge his own wound will enact it on others and on himself [27]. A medicine of self-limitation therefore protects the clinician as well as the patient. To accept limits is not to become indifferent. It is to refuse a false divinity. The physician is responsible; the physician is not sovereign; and an institution that tells its doctors otherwise, in either direction, is lying to them.

Figure 7

Gretchen: The Person Left Behind by the Project

No clinical reading of Faust can avoid Gretchen. Her tragedy is not simply that Faust desires her. It is that she is incorporated into a project whose centre is elsewhere. She is utterly singular, a girl with a dead baby sister she once nursed, a particular bedroom, a particular conscience, and the striving consciousness treat that singularity as a medium of experience.The analogy to medicine must be drawn with care. Physicians do not ordinarily exploit patients in the manner of a seduction. Yet it is worth recalling how the catastrophe begins, because Goethe makes it pharmacological. To spend a night with Gretchen, Faust hands her a small bottle for her mother: three drops in her drink, he says, and she will sleep soundly. Gretchen asks whether it will do harm. Would I advise it otherwise, he answers?

The mother does not wake. Faust the former plague doctor, who already knows what it is to have dispensed a lethal remedy with good intentions, prescribes again, and reassures. The structural danger is recognisable. A patient becomes important primarily as the bearer of a diagnosis, a candidate for a procedure, a participant in a trial, a teaching case, a diagnostic challenge, a quality metric, a source of reimbursable complexity. The person becomes the occasion for the project.Goethe gives Gretchen one scene in which she turns the examination around. In the garden she asks Faust, quite directly, how things stand with him and religion (line 3415): the question Germans still call the Gretchenfrage, the plain question that demands a plain answer. Faust replies with a magnificent rhapsody. Who may name Him? Feeling is everything; the name is sound and smoke. Gretchen listens and observes, drily, that the priest says much the same in slightly different words. She has asked a second-person question and received a lecture. Every clinician has given that lecture. The patient asks, "Am I dying?" and receives a disquisition on response rates.

The patient asks, "Would you have this operation?" and receives the consent form. Eloquence deployed against a direct question is one of the quieter forms of abandonment, and Gretchen, who has no education at all, sees through it immediately. She also cannot bear Mephistopheles, whose presence, she says, closes her heart. The uneducated patient is frequently the best diagnostician of what is wrong with the team.One of the deepest ethical consequences of specialisation lies here. Specialisation has produced extraordinary expertise, but expertise divides the body by jurisdiction. Cardiology sees the heart, nephrology the kidneys, neurology the nervous system, oncology the malignancy, psychiatry the mind, surgery the operable lesion. The patient alone has all the diseases at once, and the patient alone has the life in which they occur. The moral work of medicine is therefore a continual reassembly of what expertise has necessarily taken apart, beginning with the elementary act of knowing the patient's name and something of her story; I have argued that medical education, which starts with an anonymous cadaver, trains us in the opposite direction [28].

Cassell's insight returns with force. Pain may be in a limb, but suffering involves biography, role, hope, dependency, shame, fear, family, memory and anticipated loss [2]. A physician can therefore treat the disease successfully while worsening the suffering and can fail to reverse the disease while genuinely reducing the suffering. The second possibility is hard for a Faustian imagination to see, because it lacks the dramatic visibility of rescue. There is no triumphal angiogram for having stayed beside a frightened dying person. No laboratory marker records that a patient was permitted to tell the truth, and no procedure code captures the moment at which the physician stopped talking and let grief occupy the room. Gretchen's last scene is the hardest. Faust arrives at the dungeon with stolen keys and magic horses to rescue her, and she will not go. She looks at the man who has come to save her and says, "Heinrich, I shudder at you" (line 4610) [1]. She chooses judgment over a rescue conducted on those terms, and a voice from above says that she is saved.

It is a competent refusal, and the rescuer cannot comprehend it. Gretchen reminds the clinician that the project and the person can diverge. The procedure can succeed and the patient still be lost; and sometimes the patient, seeing more clearly than we do what the rescue would cost, declines to be saved by us.

Figure 8

Medicine as Secular Theodicy: Faust, Job and Moses

Goethe opens the drama proper with a Prologue in Heaven that is openly modelled on the first chapters of Job. The Lord points to his servant Faust; Mephistopheles, like the Satan of the biblical court, proposes a test; the Lord permits it, remarking that man errs if he strives (line 317) [1]. From the outset, then, Faust announces itself as a rewriting of the canonical book about innocent suffering, with one inversion. Job suffers and demands an explanation. Faust acts and demands an experience. They look like opposites, and they meet at the same boundary: both ask whether existence can be possessed by the explanatory structures available to them. Job's counsellors say yes, in the register of moral causality. Mephistopheles says yes, in the register of instrumental causality. Both threaten to dissolve the singular event into a system.The theological structure of our problem comes into focus if medicine is understood as a form of secular theodicy. Traditional theodicy tries to reconcile suffering with an intelligible moral or divine order. Medicine does not ask why God permits disease.

It asks what caused the disease, how it behaves, and what can be done, and that shift is one of civilization’s great achievements. The child with bacterial meningitis needs antibiotics and not metaphysical speculation; the patient in status epilepticus needs the seizure stopped. A world in which suffering is investigated causally rather than justified morally is in most circumstances a more humane world. But medicine acquires a theodical temptation of its own when causal explanation begins to perform the existential function that theology once performed. If the suffering has a mechanism, it seems partly domesticated. If the mechanism can be named the event appears less chaotic. If a treatment exists the event appears less absurd, and if the treatment fails another must be found. When no treatment remains, the system undergoes a crisis not only of efficacy but of meaning, and the patient is said to be someone for whom there is "nothing more to offer." The phrase should trouble us. Nothing more to offer, or nothing more to administer? The two are not the same.

Levinas argued that the justification of the other's suffering, by whatever system, is the root of all immorality, and that suffering is intrinsically useless, acquiring meaning only in my response to the other's pain and never as an account of it [29]. Braiterman gave the refusal a name, anti-theodicy, and showed it to be a native possibility within post-Holocaust Jewish thought rather than a departure from it [30]. Job remains its canonical text because Job's friends possess a theory before they possess the sufferer. Their theology is coherent: the world is morally ordered, and so suffering must be intelligible within the order. Job's protest threatens not only their conclusion but their system. The modern clinician can reproduce that structure without any theology at all. Your pain must fit the scan. Your decline must fit the expected trajectory. Your response must fit the trial. Your grief must fit the stages, and your acceptance must arrive on schedule [31]. Your refusal must be explained by impaired capacity, denial, poor health literacy, depression, or that useful word "non-compliance." Arthur Frank observed that modern medicine knows only one plot, the restitution narrative, yesterday I was healthy, today I am sick, tomorrow I shall be healthy again, and that patients whose illnesses will not follow that plot find that no one can hear them [32].

The patient who resists the expected script threatens the clinical system much as Job threatened his counsellors.The rabbinic traditions about the death of Moses carry the argument further than Job does, because there the sufferer is dying and knows it. In Devarim Rabbah and the related Midrash Petirat Moshe, the greatest of the prophets refuses the serene developmental script that we like to impose on the dying [33]. He bargains, pleads, proposes alternatives, offers to enter the land as a beast or a bird, draws a circle and refuses to leave it, and storms heaven with prayer until the gates of prayer are ordered shut. He is never given a reason. Rav lakh, he is told enough; do not speak to Me further of this matter. In my reading of these texts, I argued that they both fit and refute the familiar stage theory: the bargains are never withdrawn, only overruled; acceptance is caused by the collapse of role and function rather than achieved through insight; and the absence of any explanation is not a gap in the story but its point [6].

Theologically this is anti-theodicy, a revelation that consists in concealment [34]. Clinically it is permission. The patient need not make suffering meaningful to deserve accompaniment and need not die well to be allowed to die.None of this is an argument against medical explanation. Job himself wanted an answer, and patients want diagnoses. The issue is whether explanation retains humility before what it has not explained. A molecular mechanism may perfectly explain how a tumour proliferates without explaining why the diagnosis has shattered this family. A mutation may explain why a neurological disease occurred without saying what the patient should do with the remaining years. A lesion may explain a pain pathway without explaining the isolation produced by years of being disbelieved. Medicine answers one kind of why. Human beings inevitably ask another. A good deal of clinical wisdom consists in recognising which of the two has just been asked.

Figure 9

The Third Grammar at the Bedside

In earlier work I described two seductive languages of suffering. The first makes suffering meaningful by placing it within a cosmic sequence; the second makes it meaningful by placing it within a  psychological or archetypal one. Both can offer real insight, andboth pay for it in the same coin, the singularity of the sufferer. The third grammar arose from the need for a clinical language that could remain meaningful without making the wound necessary [4]. Faust sharpens that argument by showing that Faustian medicine is itself a grammar, and a complete one. Its nouns are disease entities. Its verbs are diagnosed, predict, intervene, correct, prevent, optimise. Its tense is the future, its temporal structure is progress, its moral mood is the imperative, its most feared word is "nothing," and its eschatology is cure. None of this is wrong. The problem arises only when the grammar claims to be the only one.

There are encounters in which the right verb is not fix but hear; not prolong but accompany; not explain but acknowledge; not reassure but remain. The third grammar does not abolish the first. It begins where the first ceases to suffice, and this is what distinguishes it from therapeutic nihilism. The anti-Faustian physician master’s anatomy, pharmacology, evidence and technique with full rigour. What she refuses is the inference that, having mastered a technique, she has mastered the encounter. The patient remains beyond her, and that "beyond" is not mystical decoration. It is the ethical condition of care.Buber offers the clearest vocabulary for the required movement [7]. Clinical medicine alternates between I-It and I-Thou: to diagnose we objectify, and to care we must also encounter. The mistake is to imagine that the two can be blended into a pleasant middle. They cannot; the physician must cross the boundary again.

Consider the examination of a patient with advanced motor neuron disease. I observe fasciculations, grade weakness, test reflexes, record the vital capacity, note the weight loss, document the bulbar signs. Every act is necessary. And then at some point the examiner looks up, and the person whose diaphragm has just been a respiratory variable is looking back. That moment changes the ontology of the room. The object has become a subject without ceasing to have a disease, and clinical ethics begins in that interruption.Levinas radicalizes the point. The other does not become ethically significant because I have understood him; the face commands before it is comprehended, and responsibility precedes knowledge [35]. Clifton-Soderstrom has drawn the consequence for our profession: medicine is ethical before it is scientific, and the patient as other is the foundation rather than the application of its ethics [36].

The inversion matters. In Faustian medicine, knowledge authorizes responsibility: I understand;therefore, I act. In a Levinasian medicine, responsibility arrives first: you address me;therefore, I am already implicated. The difference is felt most sharply in chronic unexplained symptoms, refractory pain, functional disorders, psychiatric distress and terminal illness, where clinicians unconsciously ration recognition according to intelligibility. The patient whose disease makes sense receives one quality of attention; the patient whose story resists our categories receives another. But the face does not become binding only after the MRI agrees.Jewish mystical language supplies a further figure for the same movement: tzimtzum, the divine self-contraction by which, in Lurianic teaching, the Infinite withdraws to leave a space in which something other than itself can exist. I do not invoke tzimtzum as a cosmological explanation of disease; to do so would violate the very discipline I am defending. I use it, as I have before, as a model of clinical self-limitation in which the question shifts from what I know about the patient to what kind of space my knowing leaves her [37].

The clinician enters the room with enormous interpretive power. We name, we classify, we decide which statements are clinically relevant, we control access to therapies, and we write the record from which later clinicians will inherit their understanding of the patient. The medical note is never a neutral transcript. It is an act of jurisdiction; a patient may spend an hour explaining a life and emerge as four paragraphs written by someone else. Clinical tzimtzum begins when the physician recognises that expertise creates an obligation to contract as well as to act. To contract is not to disappear. Self-effacement does not care; patients come to us precisely because we know things they do not and can-do things they cannot, and the physician who renounces expertise in the name of equality has abandoned the asymmetry on which care depends. Tzimtzum is harder than that. It means keeping the expertise while refusing to occupy the whole space: asking a question to which one does not already hold the acceptable answer; allowing a silence not to be filled at once with reassurance; letting the patient's vocabulary stand rather than translating every word into ours; knowing when an interpretation, however elegant, has become an invasion.

I described this elsewhere as the interval between the patient's cry and the clinician's incapacity, an interval that neither party manufactures and neither is entitled to fill prematurely [4]. That interval is the precise opposite of the Faustian pact. Mephistopheles promises that no void need stay empty. Clinical tzimtzum protects the void.Two cautions follow, both directed at my own project. The first concerns hermeneutic medicine, which rightly insists that patients must be read and heard rather than merely measured [38,39]. Even interpretation can turn Faustian. To call the patient a text risks implying that sufficient hermeneutic sophistication will finally disclose the hidden meaning. But some narratives stay contradictory. Some patients change their minds or want at once to live and to stop treatment or never produce a coherent story at all. The patient is therefore not simply a text. The patient is an author who can refuse our reading. Narrative competence, as Charon develops it, is not the power to tell the patient what her story means; it is the discipline to notice form, rupture, silence, repetition and competing plots without assuming authorship [40].

A text can be interpreted without answering. A Thou answers back, and that possibility of reply is morally decisive: the patient must retain the capacity to surprise the clinical account.The second caution concerns synthesis. Every serious illness generates at least two accounts.One is biomedical: the tumour is 3.2 centimeters, the ejection fraction is thirty per cent, the MRI shows new lesions, the EMG shows denervation, the creatinine is rising. The other is existential: I can no longer lift my granddaughter; my husband sleeps in another room now because the oxygen concentrator is too loud; I am afraid to fall asleep because that is when the seizures come; I survived, but I did not return. These accounts are not competitors, and the trouble begins only when one claims jurisdiction over the other. A scan cannot invalidate grief, and a narrative cannot dissolve a pulmonary embolus.

Medicine loves synthesis, and synthesis is intellectually elegant, but some truths must remain adjacent without being merged. The patient is a body and more than a body. The physician must objectify and must refuse objectification. Death is biologically explicable and existentially incomprehensible. Suffering may acquire meaning for a patient without having been inflicted for the sake of that meaning. The urge to resolve such tensions is Faustian, because resolution promises possession. The second rule of the third grammar, irreducible antinomy, lets them remain true together [4].

Figure 10

Three Anti-Faustian Disciplines: Not Knowing, Consent and Hope

If the pact is a refusal of limit, the counter-practice must consist of disciplines that build limit back into the daily work. Three already exist within ordinary medicine, though we seldom recognise them for what they are.The first is the honest management of uncertainty. Renée Fox showed two generations ago that medical education is, among other things, a training for uncertainty: students must learn to distinguish the limits of their own knowledge from the limits of medical knowledge as such, and to act anyway [41]. What the culture adds, unofficially, is a reward for concealment. Trainees learn quickly that "I don't know" had better be followed at once by a plan. That is sensible as far as it goes; uncertainty should generate investigation when investigation is indicated. But there are several kinds of not-knowing, and they are not morally equivalent. We may not yet know the diagnosis. We may know the diagnosis and be unable to predict the individual course. We may not know whether the intervention will work. These three, epistemic, probabilistic and technical, are medicine's proper adversaries. But we may also not know what the illness means to this person, and beyond that, neither physician nor patient may know what any of it finally means. These last two, interpretive and existential, are not defects awaiting repair. They may be conditions of an honest relationship.

Faust cannot bear not-knowing because he experiences uncertainty as deficiency. Simpkin and Schwartzstein have suggested that the tolerance of uncertainty may be the next revolution medicine requires, and I have argued that both medicine and religion need what I called a sacred epistemology of not-knowing, in which doubt is a foundation for humility rather than a threat to authority [42,43]. This is not ignorance celebrated. It is accuracy protected from vanity. The physician who says "I do not know why this happened to you" may be more trustworthy than the one with a metaphysical answer. "I do not know whether you will be the one who responds" is more honest than converting a population probability into a personal destiny. "I do not know exactly what dying will be like for you, but I can tell you what we anticipate and what we can treat" respects both expertise and limit. The Faustian physician fears that admitting uncertainty will destroy authority. The trustworthy physician discovers that properly bounded uncertainty creates it.The second discipline is consent.

It is usually handled procedurally: disclose risks, benefits and alternatives, and obtain a signature. Jay Katz showed long ago how thin that procedure is, and how much of the history of medicine has been a history of silence in which the doctor decided and the patient complied [44]. At its root, however, consent is an anti-Faustian institution. It places a boundary around the physician's capacity to act. I may know what can be done, may recommend what I believe should be done, may possess the skill to do it, and still the body is not mine. Consent is therefore not simply a transfer of information. It is the institutional acknowledgment of alterity; the patient remains other to the medical project. This is clearest when patients refuse. The refusal of a beneficial treatment arouses powerful feelings in clinicians because it interrupts the expected movement from knowledge to action. We know the diagnosis, the intervention, and the likely consequence of declining it. Why will she not agree? The ethical response is not passive neutrality. Capacity must be assessed, misunderstanding corrected, coercion looked for, reversible fear explored, and one's recommendation stated plainly. But after all that, a competent refusal teaches something medicine must never forget, which is that knowledge does not create ownership. Mephistopheles offers Faust a world of pure availability. Consent reintroduces the boundary he promised to remove. Gretchen in her cell, declining the magic horses, is its patron.The third discipline is a chastened account of hope. Hope is among the most morally complicated words we use. Physicians fear destroying it, families ask us to preserve it, patients accuse us of taking it away, and we sometimes deploy it as though it were a physiological variable requiring maintenance. Hope turns Faustian when it means sustaining the fantasy that every limit will yield to the next intervention.

True hope can survive the death of cure, because its object changes. Hope for remission becomes hope for time at home. Hope for walking becomes hope for an independent transfer. Hope for recovery becomes hope for recognition; hope for longevity becomes hope for reconciliation; hope for another year becomes hope for one lucid afternoon. Groopman's clinical portraits make the same distinction between hope and mere optimism, the first of which requires the truth and the second of which avoids it [45]. The physician's task is not to manufacture optimism but to discover, with the patient, what future remains imaginable without falsifying the present. Prognostic honesty and hope are therefore not opposites. False reassurance preserves expectation for a while and destroys trust in the end. Hope is ethically mature when it can survive the truth.

Figure 11

The New Mephistopheles? Artificial Intelligence and the Homunculus

No contemporary discussion of Faustian medicine can avoid artificial intelligence, and the analogy is almost too easy, which is a reason to handle it carefully. AI is not Mephistopheles, and machine learning is not demonic. Automated interpretation may improve diagnosis, reduce clerical burden, find subtle patterns, extend specialist expertise to places that have none, and predict deterioration before any human notices. Topol has argued, persuasively, that its best use would be to give clinicians back the time that the last generation of information technology took away, and my own experience with AI-assisted diagnostic tools in a large population with post-traumatic stress persuades me that such tools can serve a holistic practice rather than displace it [46,47].Goethe, as it happens, anticipated the figure. In Part II, Faust's former assistant Wagner, the pedant who stayed in the laboratory, succeeds in crystallizing an artificial man in a glass flask. Homunculus is dazzlingly intelligent from the moment of his making. He reads the sleeping Faust's dreams briefly, knows the way to classical Greece, and outthinks everyone in the room including Mephistopheles. But he has no body.

He is pure mind under glass, and he knows it; his whole desire is to come into being properly, and in the end, he shatters his vial against the sea-shell chariot of Galatea and pours himself into the ocean to begin life where life begins. It is a remarkable image to find in a poem finished in 1832: an intelligence that exceeds its makers in cognition and is aware that cognition is not existence. The danger, then, lies not in the machine's intelligence but in an old epistemological temptation that the machine intensifies: the belief that what can be rendered as data is what is most real. The algorithm receives representations, laboratory values, images, codes, clinical text, waveforms, genomic sequences, medication histories, and its strength is precisely that it can discover relationships among represented variables at scales no clinician can match. But a representation is not an encounter, and the distinction only seems sentimental until one sees its consequences. Obermeyer and colleagues showed that a widely used commercial algorithm, which took past health-care expenditure as its proxy for health need, systematically under-identified sick Black patients, on whom the system had simply spent less [48].

The machine was not malicious. It faithfully mistook the record of what had been done for the reality of who was suffering, which is the Faustian error in its purest form.The same gap opens at the bedside. An algorithm may predict that a patient is likely to deteriorate; it cannot thereby determine what deterioration means to that patient. It may estimate survival; it cannot decide what makes the remaining time worth inhabiting. It may flag a pattern of opioid requests; it cannot tell from the flag whether the person before us is manipulative, undertreated, frightened, physiologically dependent, addicted, traumatized, or several of these at once. It may generate the differential diagnosis; it cannot relieve the clinician of responsibility for the person who must live inside whichever diagnosis proves true. The Faustian danger of AI is not intelligence exceeding ours. It is representation masquerading as presence.If machines become better than physicians across ever larger domains of detection and prediction, medicine will have to confront a question it has long postponed: what is the physician for, when the physician is no longer the best information processor in the room? That question can produce panic only if information processing was secretly our final definition of ourselves. A relational account gives a different answer. The physician is not merely the one who knows. The physician is the one who becomes responsible for what the knowing means here, in this body, for this person, and who, unlike Homunculus, has a body of his own with which to stay.

Figure 12

Sorge: What Remains When the Bargain Expires

One of the odd linguistic habits of modern medicine is the distinction between "curative" and "supportive" care. The terminology unintentionally discloses our hierarchy. Cure is the substance; care supports it; and when cure ends, supportive care is what is left, as though care were the scaffolding still standing after the real building has gone. Historically and morally the relation runs the other way. Cure is one of the things care sometimes does. Care is not what begins when medicine can no longer cure; medicine begins with care, and the Faustian clinic is the place that has forgotten the order. That is why the grey woman at Faust's keyhole matters so much. Colvin locates Sorge close to the redemptive centre of the drama: in a world given over to achievement and acquisition, Care returns as the one thing that accelerated activity cannot master [13]. Faust will not acknowledge her and is blinded. The clinician who will not acknowledge her is blinded in the same way, able to see the next intervention and nothing else.

Patients eventually reach the limits of intervention, not always dramatically. Sometimes the limit is metastatic cancer; sometimes progressive neurodegeneration; sometimes severe chronic pain after every reasonable alternative has been tried; sometimes an irreversible cord injury, a treatment-resistant depression, or a frailty in which each successful intervention produces the next vulnerability. At such moments the physician faces a crisis of identity. The question "What can I do?" becomes dangerous if doing is the only category within which clinical worth can be imagined. There is by now good evidence that the alternative is not second-rate medicine. In Temel's trial, patients with metastatic lung cancer who received early palliative care alongside oncological treatment had better quality of life and less depression, chose less aggressive care at the end, and lived longer [49]. Care was not the residue. It was an active ingredient.

Palliative medicine has institutionalised the insight, but the insight belongs to all of us: the oncologist, the neurologist, the surgeon, the intensivist, the family physician, and above all the doctor of the chronically suffering patient whose symptoms cannot be abolished. Quill and Cassel named the corresponding duty nonabandonment and called it a central obligation of the physician, a continuing commitment that does not lapse when the therapeutic options do [50]. Back and colleagues found that dying patients and their families experience abandonment in exactly the two ways one would predict: loss of continuity with the doctor they knew, and a lack of closure when that doctor simply ceased to appear [51]. The opposite of cure is not failure. Sometimes the opposite of cure is fidelity.Death is where Faustian medicine becomes most visible, because death is the one limit no progress has abolished. Medicine has transformed when, how and of what most people die. It has not altered mortality, and Hans Jonas argued that we should not wish it to mortality is a burden and the condition of there being new lives and any urgency to our own [52].

Yet our idiom treats death as if it belonged in the same category as a preventable complication. The patient "failed" chemotherapy. The heart failed, the graft failed, resuscitation failed. Each usage is defensible. Repeated often enough they build a world in which dying itself is an adverse outcome of medical management, and Gawande has described with painful honesty what that world does to the old and the dying and to the doctors who cannot stop [53]. The burden falls on everyone. The patient feels responsible for "giving up." The family feels that choosing comfort means choosing death. The physician experiences the transition to palliation as defeat. The dying person becomes the stage on which medicine performs its anxiety about finitude.The rabbinic Moses offers another model [6,33]. He does not welcome death serenely. He protests, bargains, tries to evade the decree, and his greatness does nothing to diminish his resistance; the text does not require him to become psychologically exemplary before he may die.

The clinical significance is large. Dying patients are not obliged to redeem death for the comfort of their clinicians. They may be frightened, angry, ambivalent, peaceful one hour and terrified the next, ready and not ready. There is no moral requirement that death become beautiful. The physician's responsibility is not to produce a good dying subject. It is to protect the dying person from unnecessary suffering, coercion, abandonment and falsehood. The Jewish traditions of the Shekhinah point the same way. Their image is not of a deity standing outside suffering who supplies its reason, but of a presence that goes into exile with the exiled. To assert as metaphysical fact that God suffers with the patient may exceed what a clinician can responsibly claim. To let the image of accompanying presence shape one's conduct is a different matter, and it changes the governing question from "Why did this happen?" to "Who remains here, now that it has?" [38]. That shift is the core of what I have called an apophatic clinical theology.

Apophasis is often misread as silence in the weak sense: we cannot know, so we say nothing, which at the bedside becomes abandonment wearing the mask of humility. The stronger apophatic move is otherwise. Because I do not possess the reason for this suffering, I am forbidden to use explanation as a way of escaping the sufferer. The failure of explanation increases the demand of presence. Job's friends did one thing right: for seven days they sat with him in silence. Their error began when the silence became intolerable and they spoke to protect the intelligibility of the world. Faced with incurable disease we over-reassure; with unexplained symptoms we over-interpret; with grief we offer stages; with rage we offer psychiatry. Sometimes each is appropriate. Sometimes each is speech deployed against silence, and the apophatic clinician asks first whether the next sentence serves the patient or rescues the physician.What would a clinical anti-theodicy look like in practice? It would begin with refusals. It would refuse to tell patients that everything happens for a reason, and refuse the secular equivalent, that every suffering can be redeemed through growth, resilience, insight, compliance or acceptance.

It would refuse to read refractory suffering as personal failure, to confuse the absence of cure with the absence of medicine, to let prognostic statistics erase individual uncertainty, or to call a technological possibility a moral necessity. Above all it would refuse to treat the clinician's own discomfort with helplessness by escalating intervention merely to preserve the feeling of agency. But it would also require positive disciplines, and I would name seven. Diagnose completely, without confusing diagnosis with comprehension; anti-theodicy must never become an alibi for diagnostic laziness, and the mysterious is not to be declared sacred because we have failed to work it up. Treat what can be treated, vigorously: pain, dyspnoea, infection, depression, seizures, inflammation, malignancy; the refusal of omnipotence is not a refusal of potency. Ask what the disease has done to the person: not only where the pain is but what it has taken; not only how many seizures but what she is now afraid to do; not only whether he can walk fifty feet, but where he no longer goes.

Permit protest, for the patient does not owe medicine gratitude for reality; anger is not automatically pathology, refusal is not automatically denial, and grief is not automatically depression. Distinguish reassurance from abandonment, since "I'm sure it will be fine," said by someone who knows it may not be, is a way of leaving the room. Preserve the second person: speak with the patient and not merely about her, especially when family members answer first, when technology fills the room, and when cognition declines, for the person with dementia remains a Thou long after ceasing to be a reliable historian [54]. And remain after failure: when the scan worsens, when the operation does not work, when the drug causes harm, when the diagnosis stays uncertain, when death approaches. I have described the physician's role in that room as closer to the chaplain's than we usually admit, a ministry of presence, witness and meaning rather than of technique [55]. The relationship should not vanish merely because the mastery has.