An argument, three failures, and what the failures leave
The conclusion at the end of this shouldn’t be accepted on its own. Stated flat, it sounds like a shrug: whether a preference is really yours is not a question you can answer, and the harder it presses the less answerable it gets. Said cold, that reads as defeatism with a philosophy degree.
The reason to believe it is an argument in Part III, and that argument needs no biography. So why the failures?
Because the argument is easy to state and nearly impossible to see, and three careful probes had to die before the thing they had in common was visible. The sequence isn’t evidence — it’s teaching. Pedagogy needs no reliable narrator: it needs probes presented at full strength and killed by valid arguments, and both of those are on the page, checkable by someone who believes nothing about my week. Watch three good instruments break. The pattern is the point.
Part I is the argument, which stands. Part II is the failures, at the strength I believed them, in the order I believed them. Everything after is what they turn out to have been.
I. The argument
A stranger is at your door with a clipboard, asking for money for a cause that may well be good. You say: I don’t give money to people I don’t know at my door.
You decided that years ago, in a quiet room, with nobody’s face in front of you. What happens on the step isn’t a decision. It’s a retrieval.
Almost everything worth knowing about how a life gets made is in the gap between those two rooms.
Three instruments
There are three basic ways to change what a person does.
You can change what they want. Call it formation. You can change what an option costs, and leave the choice with them. Call it pricing. You can change whether the choice reaches them at all. Call it jurisdiction.
Steven Lukes cut power roughly this way in 1974 — decision-making, agenda-setting, and the shaping of preferences — and he thought the third the most insidious, because it prevents the grievance from ever forming. People don’t resent what they can’t imagine wanting. The cut is his; the fit is rough. My jurisdiction is close to his agenda-setting and my formation is close to his third face, but pricing isn’t his first face at all. His first face is winning an open contest. Pricing isn’t a contest, it’s a tax.
Lukes was writing about power over other people. The move is to turn it around. You run all three instruments on yourself, constantly, and mostly you are the only one running them. The door policy is jurisdiction: it takes the decision away from the version of you standing in a doorway with a human face six feet off, and hands it back to the version in the quiet room. That isn’t a diminished case of political power. It’s the same instrument with the operator and the target folded into one body.
The arrow
Here is the part that makes this a model rather than a list. The instruments interconvert, and they interconvert in one direction. Everything runs downhill into formation.
Jon Elster named the road in 1983. He called it adaptive preference formation, after Aesop’s fox — sour grapes, and its overlooked twin, sweet lemons. We downgrade what recedes and upgrade what arrives. Price something out of reach for long enough and people stop wanting it. Nobody experiences not flying private as a foreclosed option. They experience it as not caring about planes.
That’s pricing arriving at formation by the slow road, and the arrival is total: the price is still there, and it has stopped being felt as a price.
Jurisdiction takes the same road on a longer clock. Keep a choice off the agenda for a generation and it stops being a choice withheld and becomes a thing nobody thought of. That last step is mine rather than Lukes’s, and it’s the load-bearing move in this section, so it should be marked as a claim: I’m proposing that his third face is what his second becomes if you leave it running, which is stronger than anything he says.
Social norms are the fast lane. A norm begins as pricing — the cost is a look, a silence, being the person who did that — and it converts to formation faster and more reliably than money ever does. Which is why your prices still feel like prices decades on, and your norms feel like you.
One caution, because the arrow gets misread as fate: not every price runs downhill. Some people want at seventy exactly what they couldn’t afford at twenty, and the wanting never adapted. The claim isn’t that constraint always converts, only that when it converts, it converts this way. What the arrow lacks is a rate. I can’t tell you how often, or under what conditions — which is survivable for a claim about direction, and turns out later not to be survivable for something built on top of it.
The mast
Elster gave us the other half, and the other half is why this isn’t only a theory of oppression.
Ulysses at the mast is self-jurisdiction. He doesn’t resist the sirens. He arranges, in advance, for resistance not to be required, because he knows the man who will be there is not the one who should decide. That’s the door policy in worse weather.
And the mast isn’t only for refusing. Consider an introvert who becomes an excellent reference librarian — better than the job requires. He isn’t a shy man talking to a stranger. He’s a librarian answering a question. The desk holds the jurisdiction, and what the desk decides is yes: it starts encounters his temperament would never have started. Same instrument as the door. Opposite direction. Both work by taking the decision away from the self who is standing there.
Note what the arrow does to the mast, though, because the mast doesn’t get an exemption. Jurisdiction held long enough becomes formation. Keep the door shut against strangers for thirty years and you don’t end up as a man who declines to answer it; you end up as a man for whom no answering arises. Ulysses is untied at the end of the afternoon. The door policy isn’t.
So the whole moral weight lands on one distinction: preferences your circumstances shaped without your consent, versus preferences you shaped on purpose. Much of Elster’s project is the attempt to keep those apart, and the distinction is real.
Which generates the question this essay was built to answer. Standing here, now, holding the preferences you hold: is this one yours?
Everything to this point I still hold. Everything after it is wreckage.
II. Three failures
First: look inside
A precommitment that still costs you willpower hasn’t worked yet. You’re paying, every time, at exactly the moment the rule existed to route around. The rule is working when it’s free. When the face at the door doesn’t pull. When the cupboard doesn’t contain a cookie — it contains a shelf.
And free is precisely what a sour grape feels like. Same interior: no pull, no ache, nothing foreclosed. The fox is not lying. He isn’t pretending the grapes are sour; he has come to find them sour, sincerely, and if you ask him he’ll say he never much cared for that kind.
The natural repair is that there’s a residue. Adaptation leaves a faint defensiveness — I didn’t want it anyway — where a chosen rule leaves endorsement: I’m glad I don’t want it. Two different aftertastes. You’d know.
You wouldn’t, and the reason is structural: the residue is a report, and reports are the thing under suspicion. Confabulation is generous. It will supply an origin story for any preference you hold, on demand, in a voice indistinguishable from memory. And it supplies endorsement preferentially, because endorsement feels better than apathy. The fox who knows he chose to stop wanting grapes is the fox the mechanism produces.
Introspection isn’t a sensor. It’s a press office.
Second: look outside
So go outside. Behavior can’t confabulate.
Take two vegetarians. One chose it. The other spent a decade somewhere meat was scarce and expensive, and stopped wanting it. Ten years on, both report the same thing: don’t want it, don’t miss it, that’s just who I am. Now move them both somewhere meat is cheap and ordinary and unremarked. If the deliberate one holds and the other drifts back, the difference was real all along — it only stopped being visible from the inside and started being visible in what people do.
Introspection is destroyed; the evidence isn’t; it moved.
Hand the fox a ladder.
If his adaptation is complete — and completeness is what this essay says these instruments achieve — he climbs, and he doesn’t eat. He passes. Not from stubbornness or pride. Because he genuinely doesn’t want grapes now, which was the entire point. The test reads him as authentic.
Re-exposure measures depth, not origin. It sorts shallow changes from deep ones. A deeply installed adaptation and a deeply chosen preference are both deep, and no ladder sees between them.
Watch what that does at the bottom. Sen’s examples of adaptation are people at the floor of a society — the beggar, the labourer scraping subsistence, the servant worked around the clock, the wife who has made her peace with her position. What they have in common, he says, is that they come to terms with their respective predicaments. Hand any of them the option back and they decline. Sincerely. They pass. The test certifies their preferences as their own, and the cleanliness of the result is the adaptation reporting success. The probe gives its most confident answers about exactly the people it was built to protect.
The confabulation argument was run against introspection and stopped there. It applies to behaviour identically. Behaviour is what adaptation produces.
Third: look for the date
Strip out the inside and the outside and something remains; it looks like the last honest thing in the building.
The preference arrived at a particular time. That happened. It’s fixed. It doesn’t move when you move, it doesn’t care who’s asking, and it was the same fact before you thought about it and will be after you stop.
Better: origin isn’t only a date, it’s a date and a circumstance, and adaptation is defined by a correlation — the new taste arrives as the alternative leaves. The fox’s tell was never what he’d do with a ladder. It’s that he stopped wanting the grapes at four in the afternoon, which is when he learned he couldn’t reach them. So don’t ask what a preference costs you. Ask when it showed up, and what else showed up that week. A taste you had before the option closed is yours. A taste that arrived exactly on schedule — the year the money ran out, the year the body changed, the month you understood it wasn’t going to happen — that one wants looking at.
And unlike the other two, this one has a referent that isn’t you. Ask how a preference feels and there’s no fact outside the report; the report is the whole object, which is why confabulation ate it. Ask when it arrived and your report is evidence about something else, and the something else left traces.
You will misremember, of course. Memory runs on the same engine as endorsement. But that’s an access problem, not a target problem: stable target, unstable access. There’s a fact; you’re just bad at reaching it.
Now consider the fox who never approaches the vine.
He looks up, reads the height, doesn’t bother. Nothing ever receded from him, because he never reached. There was no four o’clock, no arrival to co-time with a closure, because the closure was a forecast rather than an event.
That fox isn’t a hard case. He’s the terminal one — he’s what the arrow says the deepest formation produces: a choice that stops being a choice withheld and becomes a thing nobody thought of. No withholding event. No before. Nothing to date.
Stable target, unstable access was a defence against misremembering. It presupposes a target. Here there isn’t one.
Third instrument, same population, third road.
III. What the three failures were
Three careful probes, three deaths, one graveyard. Bad luck, or a result.
It’s a result, and getting to it means defining a word this essay has been leaning on for three sections without ever saying what it means.
Yours. When I ask whether a preference is yours I’m not asking whether you endorse it, or whether you’d keep it under pressure, or whether it fits the world. I’m asking about cause: is this preference here because the alternative closed? The librarian’s taste for starting conversations is his — not because it’s deep, not because it’s correct, but because what explains it is his taking the desk rather than a door shutting. The fox’s conviction that grapes are sour isn’t his, because the closure explains it and nothing else needs to.
That’s a causal claim. And a causal claim of that shape is a counterfactual: would this be here if the alternative hadn’t closed?
Which is the entire result, and it isn’t an induction from three failures. Counterfactuals need an otherwise. Formation’s success condition is the elimination of the otherwise — that’s what the arrow says and it’s the only thing the arrow says: a price stops being felt as a price, a foreclosure stops registering as foreclosed, a withheld choice becomes a thing nobody thought of. Complete formation isn’t a strong preference. It’s a preference with nothing on the other side of it.
So the question can’t be answered in proportion to how much the answer would matter. Where formation is shallow, every probe works and you didn’t need one. Where it’s total, no probe works and none can.
The three failures weren’t three accidents. They were the three places an otherwise can live — in the feeling, in the behaviour, in the history — and the process that makes the question urgent is defined by clearing all three. That’s why each death looked like a bad instrument when the instrument was fine.
And the population route isn’t the escape, though it looks exactly like one.
Sen compared self-reported illness across Indian states. As he read the figures in the 1990s, Kerala — near-universal youth literacy, life expectancy then around 74 — reported the highest morbidity of any state; Bihar, with low longevity and poor medical and educational provision, reported the lowest. Self-reported morbidity ran almost exactly opposite to life expectancy across states. Nobody was asked to introspect. Nobody was handed a ladder. The verdict falls out of the shape of the distribution, and no otherwise is needed inside anyone.
That instrument is real. It is not a fourth probe, and the reason is exact.
Sen has a non-report measure of health. He does not have an otherwise for the wanting. What the divergence licenses is a verdict about a scale: self-reported morbidity is an invalid measure of health, because the reports are calibrated to expectation rather than to the state. That’s a finding about an instrument, not about anyone’s authorship of anything.
Read it as an authenticity result and you’ve quietly adopted a theory — that a preference is yours when it’s calibrated to the objective state of the world. The librarian kills that in one line. His preference is calibrated to nothing, tracks no mortality, has no non-report measure within a mile of it, and is this essay’s central case of a want that is his. Calibration doesn’t make a preference yours, and miscalibration doesn’t make it stolen. Adapted perception and adapted preference are cousins, not the same animal, and the transfer needs them to be identical.
So the population instrument answers a real question and it’s a different one: is this report a valid measure of the thing it’s about. Which is worth a great deal — hold it; it comes due in Part V. It just isn’t this, and the question itself has one move left.
IV. The otherwise you can’t erase
Go back to the third failure, because it broke in a way worth studying. It asked for a date — the moment the preference arrived. But a date is a point, and a point was the wrong shape for the question. The temporal move isn’t when did it show up; it’s fix the criterion and follow it. The Ship of Theseus stops being a puzzle the instant you refuse to let “the ship” drift between matter and form while you reason: name the criterion at the start, hold it, and the mystery flattens into a plain report of a plank replaced. Naturalization is that same unmarked drift one deck down — the criterion for what I want slides while the label who I am stays nailed in place. Hold the label still and trace, and the slide reappears as a crossing, with a year on it.
That is a better instrument than dating a point, and at the bottom it dies the same death, for the same reason. A trace needs its first frame on the record, and the terminal fox never made one, because he never reached. Nothing was written at the first hour, so there is no first hour to return to. The most thoroughly formed man is once again the one the instrument can’t read — the third failure in a fourth coat.
But look at what every probe so far has taken for granted: that to catch a formed preference you have to recover its history — a feeling, a behaviour, a date, a trace, all of them backward-looking, all of them things formation is licensed to erase on its way out the door. Part III counted the places an otherwise can live and got three, and the count was right about the wrong set: three places a historical otherwise can live. There is an otherwise that does not live in the past.
Consider Euclid. For two thousand years the parallel postulate read as a fact about space — not chosen, not choosable, simply how extension is. It was invariant across every geometer who ever checked it, and the invariance was taken for the mark of necessity. Then someone negated it and the negation held: a whole consistent geometry on the far side. The postulate hadn’t moved and no one could date its arrival, but it had been a choice the entire time — and what exposed the choice was neither a memory nor a measurement. It was that an otherwise had been standing there, coherent, the whole while, simply never tried. Invariance is not innocence. A thing can sit unmoved for a lifetime and still be a decision, so long as a coherent alternative stands beside it unbuilt.
That hands you a question you can put to the terminal fox with no date in it. Not when did you stop wanting grapes — he has no answer, and neither would a saint. Ask instead whether a coherent otherwise stands beside the not-wanting: a fully livable life in which the grapes are still wanted. For the fox there plainly is one — foxes eat grapes wherever the vine hangs lower; the wanting is a going concern everywhere but here. For a real feature of his nature there would not be: there is no livable fox who longs to walk into the wolf, no otherwise to point at. The tell was never in his history, which is gone. It is in whether the world one step to the side of his preference is coherent or empty — and that you can read off the present, with no archive at all.
So the preference is made, not found: it sits where a decision sits, beside a real alternative, not where a mountain sits with nothing on its far side. And that is the smaller, sharper distinction the essay should have been drawing the whole way, because there were always two questions folded into is this mine, and they have different fates. Made or found — is this a decision-shaped thing or a fact of my nature — is answerable, structurally, in the present tense, by whether an otherwise stands beside it. Authored or installed — granting it’s made, did I make it or did a closing door — is the counterfactual one, and it stays unanswerable exactly where Part III left it: at the bottom, where the door has finished closing and taken the history with it, because that otherwise is the historical one, and erasing it is the whole of what formation does. The librarian’s ease with strangers has its otherwise too — the shy life he never led — and it is every bit as made as the fox’s sourness, and it is his. Both are constructions. Only which of them did the constructing stays behind the wall.
Say what the criterion is, though, because it forks, and one tine is a trap. Read coherent otherwise as actual variation — foxes eating grapes somewhere — and the test is a census. Run the census in a world where every vine hangs high and every fox has stopped wanting, and the not-wanting reads back as nature: the probe hands its most confident verdict to the most totalizing formation, which is the ladder’s failure again, one step up and one step later. So the criterion cannot be somebody somewhere has it. It has to be coherence proper — livability judged against the creature’s constitution the way the negated postulate was judged against the remaining axioms: grapes are digestible, the wanting is biomechanically available, a fox who wants them is a fox and not a contradiction. That is the commitment, and its price goes on the same line: Euclid’s otherwise came with a consistency proof, and preference-space has none. The judgment that a life is livable is a judgment, not a theorem. If it turns out the only way to make it is to take the census after all, this probe dies where the ladder died, and Part IV goes into the graveyard it was built to walk out of.
And notice what the criterion does when you run it on a person instead of a fox. Human constitutions are wide. Nearly every preference you hold has a coherent neighbour — a livable you who wants what you don’t, priced differently, formed in another room. Run made-or-found honestly and almost everything comes out made; the test discriminates only at the far edge, where no livable alternative exists at all. That looks like a leak, and it’s the finding. Almost nothing about you is found. The self is decision-shaped nearly all the way down, and the mountain-facts — the ones with genuinely nothing on their far side — are a thin rim at the bottom of the map.
Which is what the verdict buys, and why the section doesn’t end at taxonomy. Found would mean fixed: no adjacent life, nothing formation could reach. Made means the preference sits inside formation’s jurisdiction — the instruments built it and the instruments still run. You can’t learn from the present whether you or a closing door did the building; that stays behind the wall. You can learn that the wall has a door in it. A made thing is a remakeable thing, and the slowest clock in Part V — choose rooms and roles, then wait — is how remaking is done. The audit of authorship fails at the bottom. The audit of remakeability doesn’t, because it was never an audit of the past.
V. What survives
The change of subject was the discovery.
Sen didn’t find a third-person authenticity probe. He stopped asking — and the reason he stopped is the reason Part III gives. Adaptive preferences are precisely his argument against using satisfaction as the measure of a life: if you come to terms with your predicament, your contentment stops being evidence about it. The capability approach was built in part as the answer — evaluate what people are actually able to do and be.
Capability needs no otherwise. What you can do is a present fact about the world: reachable, checkable, indifferent to whether the wanting was installed and indifferent to who authored it. It doesn’t ask you to compare yourself to a version of you the process erased, because it doesn’t ask about the wanting at all.
Three probes for authenticity died on the same people, and got a graceful paragraph of apology. The apology was the discovery, misfiled. The problem was never that those preferences might not be theirs. It’s that the lives are small, and would be small whoever’s preferences those are.
So the load-bearing result is negative, and Part IV fixed its scope: stop auditing who authored your preferences. Made or found stays open for business — cheap, present-tense, answerable. Authorship is the audit to drop. Not because the question is meaningless — it has an answer, there is a fact about what caused what — but because it’s unanswerable exactly when it’s urgent, and a neighbouring question is answerable exactly then.
Two things survive because they never needed the question.
Depth. The ladder failed as an authenticity test and works fine as what it always was: a test of whether a rule is deep. Worth knowing on its own terms, since a shallow rule is one you’ll break under a pressure you haven’t met yet, and finding that out on purpose is cheaper than finding it out by accident. Note the shape — a present-tense fact requiring no historical otherwise. Same class as capability, and same class as Part IV’s probe, which needs an otherwise but takes it from the space of alternatives rather than the erased record. Three survivors of the same shape isn’t luck; it’s the diagnosis confirming itself: what formation kills is backward-looking, and what survives reads the present.
And the instruments and the arrow, which are descriptive. They say how wants get made and don’t care whose they are. The doorstep is jurisdiction. The norm is pricing on a fast clock. The settled life is what the arrow produces.
A design rule survives, with its warrant attached.
A rule can be well-made without anyone knowing whose it is. A precommitment works when it’s harder to revise than the pressure it defends against and easier to revise than your life. Too fast and it revises under the pressure it was built for — that’s not a rule, it’s an intention. Too slow and it outlives its reason and grinds whoever’s attached: indissoluble marriage is revision slower than a bad decade against a pressure that runs for life, and it needn’t be exploitative to destroy someone, because nobody holds the authority to change it.
Which sorts the instruments by the clock of the pressure.
Seconds — doorstep, deadline, last call, checkout page. You will lose, because the encounter is engineered to make the you who’s standing there worse at deciding than the you who isn’t. Use jurisdiction; don’t be reachable.
Months — the diet, the budget, the habit. Use pricing. Keep the choice, move the cost, because you don’t yet know if the rule is right and jurisdiction destroys the evidence by removing the case.
A life — who you become. Only formation works, you can’t aim it directly, and you aim it by choosing rooms and roles and then waiting. The librarian didn’t decide to be outgoing. He took a desk.
Take that as a heuristic that has never been tested, not a rule that has passed. It needs pressure duration to be measurable independently of the rule being evaluated, or match the clock is circular and the right tempo is whatever tempo prevails, flattered. The obvious design — vary revision speed, hold circumstance constant, compare divorce waiting periods across jurisdictions — is confounded, because whatever sets the waiting period is usually also shaping the pressure. Nobody has a clean design, including me. Use it as scaffolding and don’t cite it.
VI. The instrument still in the building
Everything above discredits one faculty: first-person reports about your own mental history. Confabulation supplies them on demand, in a voice indistinguishable from memory, and supplies endorsement preferentially.
This essay’s frame is a first-person report about my own mental history. I built three probes and caught myself. Same class, same discount, and no version of me can audit it — the man who knows he caught himself is what the mechanism produces. Which is why the sequence is pedagogy rather than evidence: the argument doesn’t need a narrator, and the narrator can’t be trusted, and those are the same sentence.
Four things I’d expect to be wrong about.
The counterfactual result is a philosopher’s argument, and it stands or falls on the definition. If yours means something other than not caused by the closing of the alternative — if there’s a defensible sense of authorship that isn’t causal — Part III is about a word I chose rather than a thing that’s true, and the whole graveyard is a stipulation.
The body might be the otherwise. The fox declines the vine and his cortisol spikes as he passes it: a non-report measure, in the tissue, of a want he denies. It’s the best candidate anyone’s offered, and I think it dies where the ladder died — arousal registers live conflict, which is depth, and a shallow adaptation spikes while a complete one doesn’t. That’s the same failure a fifth time, which is either confirmation or a rut. If a physiological signature turns out to discriminate origin rather than depth, this essay is wrong at the spine.
Part IV’s criterion is the newest commitment and the least tested. Coherence-not-census is a judgment with no consistency proof behind it, and I don’t have a procedure for making it that couldn’t be accused of consulting the population out of the corner of its eye. If livability turns out to be decidable only by looking at who actually lives that way, the modal probe collapses into the ladder, and the essay’s one forward-looking instrument goes into the ground with the other three.
And the arrow’s missing rate. I said it was survivable for a claim about direction and fatal for anything built on top, and the date probe was built on top: co-timing is evidence only if conversion is the default, and I never checked the dependency because I was busy defending the probe from the wrong objection. The section I never wrote is the one that matters — what predicts non-conversion. Counterfactual vividness. A visible reference group who has the thing — which is Part IV’s otherwise, standing where you can see it. Whether you said it out loud before the door shut.
VII. What to write down
That last one is a practice, not just a prediction, and it’s the only prescription this essay has left.
If the self can’t be inspected, the advice was never check your dates. It’s leave dates. Say the thing out loud before the closure; write down what you wanted while you can still feel it, because the record has to be made before the thing it records disappears.
But hold that to what Part III allows, which is much less than it sounds like. A diary is a first-person report with a timestamp on it. The timestamp is a fact about the world; the content is the press office. And a note written the same afternoon the wanting stops is written while the mechanism is running — I looked at the vine, judged it too high, turned away is not a check on confabulation, it’s confabulation’s first draft, filed on deadline.
What a dated record can do is smaller and real. It can establish that a want existed at a time, against later denial. Not this preference was yours — that stays unavailable — but a preference was here, and the man who says it never was is wrong about a fact. It’s protected against backward contamination, because a note written before you hold a preference can’t have been shaped by the preference you don’t yet hold. It’s fully exposed to everything else. At three o’clock the grapes were there, and he wanted them. That’s the whole of it, and it’s more than the other three probes managed.
And it doesn’t reach the bottom. The terminal fox doesn’t decline the vine; he doesn’t see a vine. There’s no assessment because there’s no object, and nothing to write at any hour. That’s Lukes’s third face proper, and it’s the same people a fifth time — though it’s worth marking that this failure isn’t the same kind as the other four. Those were probes that couldn’t read. This is advice that arrives too late.
Which turns the thinnest paragraph in every draft of this into its main prediction. Not always, often thin was how I used to write about traces — a caveat, apologising for the instrument. It isn’t a caveat. Record density tracks power. The people whose wants were most administered are the people with the least paper: no letters kept, no diaries, no one writing down what they wanted before it became impossible. A criterion whose evidence is systematically absent for the foreclosed makes the knowability of a self a property of the documented, which is close to the same list of people as the well-supplied. Checkable in the ordinary historical way — whose interior lives can we reconstruct at two hundred years’ distance, and does the answer track literacy, property, and who was worth writing to. I’d guess it tracks them almost perfectly. I haven’t checked; historians of literacy would already know; and if archives turn out more democratic than I think, this is wrong and the piece ends at the graveyard.
The fox who never approaches the vine has no four o’clock. Ask him and he’ll tell you, truthfully, that he never wanted grapes — and he isn’t wrong, exactly, because there’s no moment where the wanting stopped and no other fox to compare him to.
He is not the hard case. He’s the ordinary one, and he’s what the mechanism is for.
So the self is not something you look into. It’s something you look up. What the present tense can carry — made or found, deep or shallow, capable or not — it carries for anyone, letters or no letters. What it can’t carry is authorship. And the question of whose authorship can be known was never a question about attention, or honesty, or courage. It’s a question about who kept the letters — and about who was ever written to.
cafebedouin.org 