O Futuro
Em Suas Mãos

I

In 1994 there was a punk band in Brasília called Vernon Walters.

Sit with the name. Everything I have to say about teaching is downstream of it.

Vernon A. Walters was a US Army officer, military attaché in Brazil from 1962 to 1967. He spoke eight languages. In the Second World War he was interpreter and liaison to the Brazilian Expeditionary Force in Italy — the only South Americans to fight the Axis — ending that campaign as head combat liaison to their commander, Mascarenhas de Moraes. He made friends there that lasted his life. One was Humberto Castelo Branco.

In April 1964 the Brazilian armed forces took the government. Castelo Branco became president. The dictatorship lasted twenty-one years. Walters went on to become Deputy Director of Central Intelligence from 1972 to 1976.

So: a band in the capital, thirty years after the coup, named itself after the American they held responsible. That is the whole argument delivered as a band name, before a single note. You can put an accusation on a cassette insert and hand it to someone at a show.

The cover was a photocopy of a face behind a pair of open hands. It was called O Futuro Em Suas Mãos — the future in your hands.

I was fourteen and I played bass. M.O.F.O. was the four of us — Patricio Verissimo singing, Álvaro Dutra on guitar, Jason on drums. Felipe let us get up on stage that night and use the gear, and we played four songs in ten minutes before Vernon Walters closed the gig.

There is footage of a show from 1993, the year before, if you want to see what the room looked like.

That is the single most consequential ten minutes of my education, and nobody in that room was an educator.

II

I have started this section four times, because the sentence that usually opens it is three words long and I do not agree with it.

The flat announcement puts a full stop where I do not think there is one. This entire essay is an argument about who gets to decide that a matter is closed, and I am not going to hand that authority over on the first line of the part about him. He is not here. The lesson is not finished. Those are two separate facts and only one of them is about him.

Here is what a search returns. A music database lists two members of Vernon Walters: Felipe Caduco and Junior Minduim. Two nicknames. No surname, no dates, nothing else indexed.

I nearly wrote that this was the entire record, and I am leaving the correction in rather than quietly fixing it, because the error is the essay's subject. What a search returns is not a record of a life. It is a record of what somebody got around to indexing. There were decades of working life after that tape, some of it in rooms with a great deal of power in them, and none of it is searchable, and all of it happened. I am not going to be more specific than that, and the reason is the whole point of the last third of this essay: he has people living, and a name, an employer and a date are not three facts. They are a location.

One thing I will say, because it is craft and not coordinates. Late on he took up building — earth, and whatever is already on the site. Geodesic domes. Triangles: a shell that stands because no single member carries the building on its own, assembled from struts light enough for two people to lift, put up by people with no training and nobody's permission.

And now I have to be accurate about what he was to me, because this essay has been drifting into a shape I do not recognise.

He was not my father. We were not close in that way, we did not hang around together, and he is one person out of a great many influences — plenty of whom were nowhere near Brasília and some of whom I have never met.

What is actually true is smaller than the version I keep sliding towards, and better. He was there. He was local. He had the gear.

That is the entire mechanism, and it is the reason a local scene is worth more than the names that fill rooms. A famous band comes through your city, plays, and goes, and you were in the audience, and nothing was handed over. The local ones you can reach. You can borrow their instruments. You can get up on their stage. You can storm their show — we did, and they let us — and that is not one man's personal kindness to a boy. It is what a scene is for, and it is the only thing it is for.

So I am not going to claim the geodesic domes as a lesson he taught me. He built them. That is all, and the symmetry is not mine to bank.

And then it ran the other way, which is the half I nearly left out because it is the embarrassing half.

Not long after, once S.L.U. had formed, we were the band in the room with the gear, and there were younger kids watching us. The criticism that came back — I heard it more than once, and it was meant to land — was that we were so hopeless that if we could do it, anybody could.

They meant that as an insult.

It is the thesis.

That sentence is the entire argument of this essay, handed to me at sixteen by people trying to take me down a peg, and I spent years hearing it the way it was meant. If we could do it, anyone could. Yes. That is the point. It was always the point. And every gate this essay has walked past — the transcript, the grade point average, the credential, the indexed record, the venue, the committee — exists to make that sentence sound like an insult instead of an invitation.

And nobody called it self-taught, either. I should say that, because I have been leaning on the word for several pages now.

Self-taught is not a description anyone applies to a fourteen-year-old in a rehearsal room. It is a compliment handed out afterwards, and only to people who have already been let in through some other door. Applied to us at the time it would have sounded like a joke. We were not prodigies and nobody was ever going to call us prodigies. We were kids who could barely play — which is also, every single time, exactly what being self-taught looks like from the outside at the start.

The act is identical. A person works it out alone, from whatever is lying around. In one body that gets called precocity and in another it gets called not knowing what you are doing, and the variable is never the work.

Which is this whole essay standing in a rehearsal room. A credential is a proxy for having checked. When there is no credential in the room, people do not check instead. They guess, from what you look like.

And a check needs an opportunity. I keep forgetting to say that, and the reason I forget is that I was given one.

We had four songs. We had practised them past the point where they could come apart. And then we were put on in the middle of a line-up, into a full room — full of people waiting for the band that was closing, which is the best audience a fourteen-year-old will ever get, because every one of them is present and not one of them is there for you. Every older punk in the city was standing in it. (We also had to fight a number of martial artists that day, which is a different story and stays in a different book.)

Readiness with no slot is invisible. A slot with no readiness is a disaster you carry around for a decade. The scene supplied the slot, the four of us had supplied the readiness, and the check took ten minutes in front of exactly the people whose opinion was the thing at stake.

Which puts where I am standing now in an uncomfortable and very exact light.

I have the check. I built the check. What I do not have is the ten minutes. A kernel will verify anything I hand it, free, at any hour, forever — and the corpus has not been found wanting by anybody, because nobody's judgement has been applied to it at all. There is no slot. That is not the same condition as not being good enough, the two are confused constantly and sometimes on purpose, and I have spent this entire essay pointing that out about other people.

Caduco will do. It was his nickname. It means senile — an old man losing his grip. He was the oldest person in the room and he took the joke for a name.

And the work is still there. The title track of that cassette is on Spotify, under the band's name, three and a half minutes long, and you can play it right now from anywhere on earth. One of the other songs has a video. Four of them pressed it onto a single-sided tape in Brasília and handed it to people at shows, and it has outlived both of them — Felipe, and the man he named the band after.

He was not my age. He was older — a grown man when I was fourteen, with a band, and gear, and an address printed on a sleeve. Caduco had come up from the south, where he had been in bands before moving to Brasília, so by the time any of us met him he had already been doing this for years somewhere else.

Old enough that he had been a boy himself when the armed forces took the government and Castelo Branco moved in — about the age I was in 1994. He did not get his politics from a record. He was there for it. That is why he talked about the CIA in Brazil the way other people talk about the weather, and it is why, thirty years later, four teenagers in Brasília playing his songs were putting an accusation on a sleeve that their singer had lived through as a child.

Here is how he got from a photocopied cassette to inside the state. I'll lay it out flat, because the shape of it is the argument of this essay.

There were two of them — Felipe and a friend, a drummer out of another band. They were teaching themselves computer science at the university, not enrolled in it, and what they were teaching themselves was how far an institution's account of its own security could be trusted. So they went and found out. They got inside a government system from personal computers, in a country that barely had the word yet for what they were doing, and they were among the first in Brazil to be arrested for it.

They always said the same thing about it, and I give you their account marked as theirs: they were studying, and doing what grey hats do, which is find the limit and report where it is.

And then someone hired him.

Notice what just happened, because it is the argument of this entire essay, thirty years before I had words for it. Nobody asked him for a transcript. Whoever hired him did not check his credentials, because he had already produced the thing a credential substitutes for: a demonstration, run on the very people who would be judging it, verifiable by the party with most reason to doubt it. He was hired on the proof, and he spent the rest of his working life inside the state.

I am not going to tell you he sold out. That is a cheap sentence and not what happened. He got caught, in that country, in those years, they offered him the one road that did not end badly, he took it, and he was good at it for the rest of his life.

But I have to tell you what he used to warn people about, in a warning about table manners, of all things: watch the menu, when the garbage man is a waiter.

He got caught, and they put him in the pigsty, and he stayed there for the rest of his life, and he never once stopped telling me to read the menu.

I heard it as being about hygiene, and I want to be careful here, because I think that is what it was: the hands that handle the rubbish are the hands that bring your plate. That is what he meant by it. It was enough.

Everything I have built on it since is mine and not his, and I am going to say so rather than let you assume otherwise. He gave me a sentence about a restaurant. I have spent thirty years reading a state into it. Flagging that is not modesty; it is the same discipline as the rest of this essay, because the alternative is doing to a dead friend's warning exactly what I once did to a hexagon — taking a particular claim for a general truth and never checking which one I was standing on.

With that said, and owned: here is what I have read into it.

III

Every book like this one has Paulo Freire in it. You'll read his name a dozen times before you reach my essay, and rightly.

Here is what I can add. Freire was from Recife, and he built a literacy method for cane-cutters and their families, and it worked. On 21 January 1964 the government of João Goulart made it national policy by decree — sixty thousand culture circles, one and a half million adults in the first phase alone.

The military took the government on 31 March. A second decree killed the programme on 14 April, four months into its life, and ordered that all the equipment and materials be collected back in.

Freire was arrested on 16 June and held seventy-five days in a barracks in Olinda. He took asylum in the Bolivian embassy that September, went on to Chile, and did not come home for sixteen years. Pedagogy of the Oppressed was written in Chile.

So when he says — and I am paraphrasing, because I want it in the form I actually carry it — that the education of the children of the poor has to be made by the children of the poor, that is not a slogan about empowerment. It is a description of what had just been done to him. A literacy programme for poor adults was dangerous enough to destroy. Teaching people to read was a security matter. Nobody was being paranoid. They were being accurate.

To say punk was born somewhere is to misunderstand it. It was not born, and it is not going to die. There is no line of descent to trace, because it is not an inheritance — it is what happens wherever people with nothing make something without asking, and each time it happens it is not the descendant of the last time. It is the same event again.

London in 1976 was an instance of it. Brasília in 1994 was an instance of it. The oldest one I know of sits in the senzala, under the big house, among people who had been denied instruments, denied assembly and denied their own religion, and who made the music anyway out of hands and boxes and whatever was in the room. A great deal of this hemisphere — its population and its music both — descends from what was done to indigenous and African women who were in no position to refuse, and every country here has spent a century finding gentler words for it. The music did not stay. Almost everything the world has danced to since came off that ground. Abolition reached Brazil in 1888, last in the hemisphere. The music was already there.

Abolition changed the law and not much else.

Two years later the republic proved it, and the clearest instance I know of the thing this essay is about turns out to have a criminal record.

Capoeira was made in Brazil by enslaved Africans — a fight carried inside a dance, by people forbidden to assemble and forbidden to fight, which is the senzala answer to being denied everything, exactly as the music was. It was never only Bahia's. In Recife, when the police went after capoeiristas, they are said to have carried umbrellas instead of knives and called the moves dancing, and the moves became frevo, the street dance of Recife's carnival, which dates from around 1907 and is still danced every year. Then the new republic named capoeira. On 11 October 1890 the penal code, Decree 847, gave it a chapter of its own: Dos Vadios e Capoeiras, of vagrants and capoeiras. Article 402 made practising it in a public place an offence of two to six months, doubled if you led a group. That stood for forty-seven years. From 1937 the state began to legalise it and set about turning it into a national symbol; Mestre Bimba had opened his school in 1932; and by 1972 the Ministry of Education had recognised it as a sport.

Read that sequence again, because of where it ends. A crime of the enslaved. Then, in the Estado Novo year, a national symbol — legalised not by a democracy but by a dictator who wanted one. Then a sport with a federation. Then an academy, a monthly fee and an abadá, and abadás are not cheap.

And Brasília is a government town. The families who could pay the fee were the families the city exists to house, and we knew exactly who the boys in the white trousers were, because everyone there knew: they were service kids. Children of the armed forces.

So an art invented by people who owned nothing, criminalised by the republic, licensed by a dictatorship, arrived one afternoon on the sons of the officer class, who came to use a room full of poor kids as punch bags.

And the poor kids were playing in a band named after the American attaché who had helped hand the country to the government those families served. Nobody arranged that. Nobody in the room would have described the afternoon in those words, least of all me at fourteen. It was just what was standing on the floor.

A master came to fight. There was a martelo rodado, spinning and genuinely beautiful, and one of us stepped into it and pushed, and the body went over in the middle of its own rotation, and the room did not applaud the dance. What happened after that belongs in the other book.

The mechanics are worth stating, because the mechanics are the argument.

A kick like that needs the room you give it. Step back and you are still inside its geometry — that is the space it was designed to cross, and your retreat is already accounted for in the shape of it. Step in and the rotation has nothing left to finish against. The whole beautiful thing goes down.

Punks do not step back. They push back. That is what it takes to drop a body in rotation.

And it is the same motion as everything else in this essay. Do not give the claim the distance it needs to work. Walk into it, and check it.

I am not telling it as a story about capoeira being no good. It is magnificent and those men could do things I cannot. I am telling it because of what had happened to the door. The kids on that floor — no money, whatever clothes they had, making a noise they had taught themselves — were standing a great deal closer to what capoeira was originally for than the paid-up members were. The art had not changed at all. The price of admission had.

In 1993 a woman named Pureza Lopes Loyola, a potter from Maranhão, went looking for her youngest son. Antônio Abel, nineteen, had left for a gold camp in Pará and stopped writing. She searched for three years carrying clothes, a Bible and his photograph. He came home on his own in the end, after thirty-eight months, sick with malaria and hepatitis — but that is not why people know her name. To get onto the farms, she took work on them as a cook. And on farm after farm, looking for one boy, she found men whose documents had been taken, who owed the owner for their own food, who could not leave.

She was not an inspector or a journalist. She had no standing of any kind. She went and looked, which was all it took, and what she found was what the country had spent a century declining to see.

Brazil now runs inspection teams for what its own statutes call work analogous to slavery. They freed 2,772 people in 2025 and more than sixty-five thousand before that, since the inspections began in 1995 — brickworks, construction, coffee, quarries, domestic service, and, last year, three hundred and four of the people freed were found working in public administration.

Brazil is not the only place. It is the place that counts, which is a different thing. The cocoa in your chocolate comes mostly from Côte d'Ivoire and Ghana, where a survey commissioned by the United States government found 1.56 million children working in cocoa in 2018 and 2019, and about ninety-five per cent of them at tasks it classed as hazardous. The industry had signed up in 2010 to a seventy per cent reduction in the worst forms of child labour by 2020. It helped set the target itself. It missed, and the only reason anyone can say so is that somebody went and counted.

Nobody owns it, and nobody can be kept out. As long as somebody is making rules there will be something that survives them, and that is all it has ever been.

And punks make rules too. I never had any patience for the part where people started organising — committees, a scene deciding who counted, a movement founded against dress codes issuing dress codes. Give it long enough, I used to say, and the organised ones will want a political party. Organise it and you have built the very thing it exists to get around, and now it has to survive you as well.

What it is good at is not organising. It is seeing the mechanism: who set the rule, and what the rule is actually doing. That is a systems habit and it is the one I still use every day. It is why I ended up building a gate that checks the work, rather than a committee that vouches for it.

Which brings me to a question I suspect more than one person reading this book has quietly asked themselves. Can you call yourselves punks?

Of course you can.

There is no register. There is no committee. Nobody has ever held the standing to rule on it, and anyone who claims they do has just told you which side of the argument they are on. That was the point before it was anything else.

I can say that because it was tried on us, and it failed inside ten minutes.

The older ones called us names. Punk laranja — orange punk. I never did find out what they meant by it, and my guess is unripe: green, not yet, not really. Whatever it meant, it was meant to tell four fourteen-year-olds they did not belong.

Then we played. And the same men — the ones we looked up to, the ones who had been calling us that — were jumping and dancing, and happier than they were for anything else on the bill that day. They respected us for the music, and for taking on the whole town, which is a different story and a different book.

Nobody admitted us to anything. There was no mechanism to admit us with. We played, and the question stopped being a question.

And the register nobody keeps was being waved at other people in that scene a great deal harder than it was ever waved at four of us.

One of the bands we played with was Bulimia — all women, out of Brasília, in those same years — and they had a song whose title is the entire argument and needs nothing added to it: Punk rock não é só pro seu namorado. Punk rock is not just for your boyfriend. Another was called Nosso corpo não nos pertence.

That is the same gate, worked from a different side. We were told we were unripe. They were told that the thing they were standing in belonged to whoever they had arrived with. Neither is a rule, because there are no rules. Both are somebody without standing behaving as though they had some.

And I have had to correct this essay twice for the same habit — writing men across rooms that were not all men. Which is the small, textual version of exactly what that title is complaining about, committed thirty years later by somebody who was on the same bill.

Olho Seco came out of São Paulo around 1980 and sang about its streets, which in those years meant the police. Their 1983 record was called Botas, Fuzis, Capacetes — boots, rifles, helmets — which tells you whose streets.

Cólera started in 1979 in Vila Carolina and ended up pacifist, anti-militarist, ecologist — a lonelier position in that city than rage. In 1986 they sold eighty-five thousand copies of Pela Paz em Todo Mundo independently. The record before it was Tente Mudar o Amanhã — try to change tomorrow — out in 1985 with the generals still in the building. Redson wrote them a song called Medo — fear — on Pela Paz em Todo Mundo in 1986, and it carries a line I have had in my head since I was a boy: that the hand held out to you is not help, that you are given a hand precisely in order to be pushed. Which means it is not an observation about power. It is an admission of being frightened.

Garotos Podres formed in 1982 in Mauá, in the ABC — the belt where three years earlier the metalworkers had struck and the modern Brazilian labour movement was born outside the auto plants. Some of their songs had to be retitled to get past the censors: the song was permitted, the name of the song was not. Twenty years later they recorded “A Internacional” — the anthem that has spent most of its life being sung slowly by officials at ceremonies — and played it fast, in the accent of the ABC, at the tempo of a song instead of a rite.

And in Brasília, in the planned city, eight hundred miles from all of that and sitting on top of the government itself, four kids named their band after the attaché.

IV

Now the part I have to be honest about, and it is the reason this essay exists rather than the thing that spoils it.

I cannot prove it.

Walters's diaries are held at the National Intelligence University. A historian, James Lockhart, went through them. They demolish one allegation outright: a historian had placed Walters in Santiago on 11 September 1973 watching, even directing, the Chilean coup, and the diaries put him in Florida: Palm Beach from 8 to 12 September, with a drive to Miami for dinner on the 11th. He was not on the continent.

On Brazil, Lockhart's argument is about the shape of the reasoning. The case against Walters is: he was there, he knew Castelo Branco, and he predicted the coup accurately. Lockhart points out that this is the inference McCarthy ran on John Service — he was there, he met them, he reported what was coming, therefore he caused it — which we rejected then for good reasons. Walters's own denial has a deflating plausibility: he said there was little advice an American colonel with no experience of coups could offer Brazilian generals who had deposed two presidents in five years.

I do not think that settles it. Lockhart's own piece contains a move I'd flag in a referee report — that too many historians are uninterested in any frame but an angry anti-imperialist one, a claim about motive, unfalsifiable as stated, doing rhetorical work his documents don't. The standard cuts both ways or it isn't a standard.

But he is right about the logic, and I have to say so.

What the record supports, at the grade the evidence carries: the United States positioned a carrier group and a fuel convoy off the Brazilian coast under the name Operation Brother Sam, ready to supply the coup if it stalled. It didn't stall, so nothing was landed, and people who want to minimise it lean hard on that. The plan existed. The ships sailed. That is damning enough on its own.

And nothing had to be landed. That is what the argument over the matériel keeps missing. A carrier group off your coast is not an intervention, it is a statement of the terms, and everyone ashore can read it. The ships went to stand watch.

Read the calendar and you can see what was being watched for. In August 1961 Jânio Quadros pinned Brazil's highest decoration on Che Guevara; six days later he was out. In January 1964 João Goulart handed Paulo Freire's method to the country; ten weeks later the army had the government, and the programme was dead inside a fortnight. If you want to keep your country, you do not give Che Guevara a medal, and you do not give Paulo Freire a programme to run. That was the lesson, and it was taught twice in three years.

“Walters directed the coup” is a different sentence with a different evidence grade, and running the two together is not a small thing. It is the exact error I named in my own work last month, in public, under its own heading.

V

Twenty years later I was living in the United States and I found his book.

Walters published a memoir in 1978, with Doubleday. It is called Silent Missions. I held it in another country, a long way from the room where I first heard the name, and the first thing I did was ask Felipe about it.

We had stayed in contact the whole time. By then it was not a scene acquaintance. We had shared bread and shared a family.

What he told me was this: the man was a spy who blackmailed the United States government with what he knew, and was promoted so that none of it would ever be published. Writing a book was itself the leverage. The manuscript was the threat. He got what he was after, which was advancement, and the rest stayed silent.

It took me until this year, writing this, to hear what he might have been saying.

Felipe was a man who got caught with what he knew, and was hired for it, and spent his life inside the thing he had broken into. And he is telling me, over a book, in another decade, about a man who knew things and traded them and rose. I do not think he was only talking about Walters. I think he was telling me the shape of a life, and I was too young and too far away to notice he was describing his own, and he would not have said so directly, because nobody does.

I have checked that story against the dates, because I check things, and the machinery does not hold. Walters left the Agency in 1976; the book came out in 1978. There was no post to be promoted into at publication. He was already out.

And yet.

After the book: ambassador-at-large under Reagan from 1981. Ambassador to the United Nations from 1985. Ambassador to West Germany from 1989, through reunification. A former Deputy Director published his memoirs and was handed three of the most senior diplomatic appointments the country has. That is not blackmail demonstrated. It is also not nothing, and I can see how thirty years and one retelling compress it into the sentence Felipe gave me.

One more thing about that book, and it took me most of my adult life to see.

A memoir by a CIA officer goes through the Agency's own review before it can be printed. So the object I was holding — a primary source, in his own hand, about his own missions — is a document whose omissions were approved by the institution under discussion. You cannot see what was taken out. There is no command you can run on it. It is the purest possible case of it is well known that, and it is called Silent Missions, which is the redaction wearing its own name on the cover.

VI

Because that is my other life. I build machine-checked proofs — a corpus of a couple of thousand formal statements, verified by a program called a kernel that reads an argument and either accepts it or does not.

In August I published a theorem and then published its refutation. The theorem was fine; the kernel had checked it. The English sentence I had wrapped around it was false. I had claimed a certain structure was the only one of its kind, and it was only the only one because I had been counting in base ten. Anyone working in base twelve would have found the hole immediately. So I named the defect, logged it publicly, and corrected the paper. I did not quietly patch it. The log is the point of the exercise.

Here is the thing I did not expect to find out about myself while writing this: I learned that from the band named after Vernon Walters.

Not the epistemics — Felipe would have laughed at the word. What I learned was that you say the thing out loud, with your name and address printed on the back, and take what comes. They accused a former Deputy Director of Central Intelligence of installing a dictatorship, and then printed where they lived so you could write to them about it.

The kernel does the same thing and calls it #print axioms. There is a command that prints exactly which unproved assumptions your result is standing on. Most published mathematics does not do this. Most published mathematics says “it is well known that” and moves on. The machine will not accept “it is well known that.” It itemises what you leaned on, the list is public, and anyone can run it and get the same list.

That is not a retreat from the politics. It is the politics. The people who have spent years lying about immigrants, about students, about what was done in Brazil and by whom, have relied every time on nobody checking. Against an information regime, accuracy is the one weapon that does not turn in your hand.

So I will not tell you Walters ran the coup. I will tell you a band named itself after him, that they were not being frivolous, that Brother Sam is documented, and that the strongest honest version of the accusation is still strong.

VII

Everything I know how to do, I taught myself, and I learned how to teach myself from those records.

That is not a figure of speech and it is why I am in this book. I have no graduate degree in mathematics. My formal mathematics ended at Calculus I, which I passed with a B-minus, and Calculus II, which I enrolled in twice and finished neither time. My combined undergraduate grade point average is 2.62, earned across two degrees at a city college while working. Everything after that — the proof assistants, the operator theory, the verification tooling, the corpus — is self-taught, in the evenings, out of documentation and mailing lists and other people's source code.

I did not choose that method on principle. It was the only one available, which is how most people arrive at DIY, and you notice afterwards that it was better.

Felipe taught himself the same way and they arrested him and then put him on the payroll. I taught myself the same way and a transcript from 2015 says I am not admissible. Same act; different institution, different year.

A kernel does not know where you studied. It does not know about the 2.62. It will not read your letters. None of that is an input. You hand it an argument; it checks the argument; it says yes or no.

That is a higher standard than peer review, not a lower one, and it costs nothing. The compiler is free. It runs on a laptop. It is a four-track recorder for proofs — except the demo is not an inferior version of the real thing. It is the same thing, checkable by anyone, forever.

I have been reading Grothendieck's notebooks this week, and they settled an argument I had not noticed I was still having.

The University of Montpellier has about eighteen thousand pages of his archive online at grothendieck.umontpellier.fr — free, no login, no letter of introduction, no affiliation. They are the papers he handed to a former student, Jean Malgoire, in 1990, which sat in Malgoire's house for twenty years before they reached the university at all.

Go and open Travaux, Cote n° 122: Point singulier isolé de la fibre — notes manuscrites. Thirty-four pages, undated, somewhere around 1968 to 1971. I am giving you the shelf mark rather than describing it to you, because everything else in this essay you have to take from me, and this one you do not.

Here is what you actually see when you open it. He wrote on the backs of other people's paper. The versos are a Bourbaki seminar handout and a conference announcement — typed exercise sheets, page 42, page 45 — and his mathematics runs across the clean side, and in the scans the typed exercises ghost faintly through the sheet from behind. There are crossings-out on nearly every page, whole passages struck through with a single line. Figure 7 is drawn freehand: dashed circles, hatched surfaces, the same quality of drawing I put on a whiteboard for a woman who had no English for what I was saying. And the last page stops in the middle of a line. He has written Q = and then nothing. Nobody ever finished it, and it is on the internet, and you can go and look at it in about twenty seconds.

That is not a great man's rough work, kept for sentimental reasons. That is what the work looks like while it is being done — by anyone, at any level, forever. He had a photocopied handout, a fountain pen, and whatever was on the desk.

I have been calling my papers zines with an axiom checker bolted on, and I half meant it as a provocation. It is not one. The only difference between a scrap and a document is whether somebody kept it, and the only difference between a private notebook and a public method is whether somebody scanned the back of the page too.

And I am going to take that last sheet as instruction rather than as an accident. He left it open. Q equals — and then it is somebody else's turn.

The same argument runs the other way round in the same discipline, one generation later. Grothendieck's notebooks are work nobody published. Perelman's proof is work nobody submitted.

Between November 2002 and July 2003 he put three papers on arXiv, about sixty-eight pages in total, and never sent them to a journal at all. Other people wrote the peer-reviewed versions. The community did the thing a community is actually for, which is check, and it held. He then declined the Fields Medal in 2006 and the Clay million in 2010, gave no reasons, and is owed no explanation by anyone, including me.

That story usually gets told as temperament — there is a rebel in every genius. I think that version costs nothing and teaches nothing, because a temperament is not transferable and nobody can act on it. Read as method it is this whole essay delivered from the top of the discipline. The proof was public and checkable. The checking had already happened. A prize certifies something; that something was already certified; the offer arrived afterwards and added nothing to it. Declining is not defiance. It is noticing the order the events came in.

And I have to say plainly what those two men were not, because letting them stand wrong here would be the same error this essay has already confessed to once.

Neither of them was outside. Grothendieck was at the IHÉS, taught and surrounded by the best interlocutors alive, at the dead centre of the subject. Perelman was at the Steklov, a gold medallist as a schoolboy, with postdoctoral years at Berkeley and in New York. They did not lack the check. They had it in the highest concentration available anywhere on earth. What they declined was the certificate.

So they are not evidence that a person can work without being checked, and I will not use them that way. They are evidence of something narrower and far harder to argue with: the venue was never the check. And that counts for more coming from them than it ever could from me, because nobody can say they were shut out and made a virtue of it. Grothendieck published none of Cote 122 and the pages are on a public server. Perelman submitted nothing to anyone and the proof held.

Mine is the other half, and it is the half this book is about. They had the interlocutor and walked away from the prize. I never had the interlocutor, so I had to build one — and it caught me inside a month, in public, on a hexagon.

I taught myself a language and then taught it, and I taught myself mathematics. Everybody assumes the language was the harder one. It was not, and the reason is exact. A language self-teaches badly because you need other people. Mathematics self-teaches badly because you don't.

Get a language wrong and somebody's face tells you inside a second. The correction is constant, unpaid and unavoidable, because the entire point of the thing is another person. Get mathematics wrong and nothing happens at all. There is no face. You can be fluent, and confident, and wrong, for years, alone in a room, and the subject will not stop you, and no one else will either — because nobody is watching, and the few who are will mostly nod.

That is exactly what happened to me with the hexagon. I was not confused. I was fluent. I had a theorem that compiled, a sentence about it that was false, and no interlocutor of any kind.

So the kernel is not a shortcut for people who could not get a supervisor. It is the missing interlocutor. It supplies — mechanically, free, at two in the morning, to anybody — the one thing self-teaching in mathematics cannot generate on its own: something that will not nod along.

Which is the complaint four fourteen-year-olds put in a song, thirty years before I understood it. A room agreeing, heads going up and down, with nobody in any position to follow. We thought at the time that this was the audience's failure and that we were making fun of them. It is not their failure. It is the ordinary standing condition of anyone working alone on something difficult, and I have lived inside it since. The fix is not to go and find a better room.

The fix is to build something that cannot nod.

Computer languages. Music. Other languages. Mathematics. I have taught myself all four, in that order, and I am not finished with any of them.

I am not done.

About two thousand formal statements, built this way. Nearly four hundred through the kernel with their assumption lists published. Fifty-three depend on no assumptions at all. Nobody gave me permission. There was nobody to ask.

VIII

I have been working since I was thirteen. Commercials first. Then for my mother, at La Force. Then for myself, with Ganesh Emporium — which in practice meant standing outside the church on a Sunday selling incense, because money had to come into the house and it was going to come in through me.

That is the other half of what a scene is, and I left it out of the earlier section because it is a good deal less romantic than being handed a guitar. Everyone knew me. It was a small city and they saw me out there most weeks. They were not admiring my work ethic. They were buying incense from a kid on the one day of the week they had to themselves, and then buying it again — and that is what support actually looks like where nobody has any money. Not encouragement. A transaction nobody had to make.

Teaching English came at nineteen. Five years of it in Brasília, where my own teachers became the people I taught beside. It was also the first job I took in New York, in 2005. After that came banking, business, hospitality, retail, trade. I came back to a classroom this year, in Elizabeth, as academic coordinator.

The students were adults. They worked jobs that started before class and continued after. They paid to be there. They arrived at the level the Common European Framework calls A1 — on the first day we had almost no words in common — and left at levels where they could argue with me, which is the only measure I trust.

I want to be precise about who they were, because the word “student” has been doing a lot of concealing lately. They were adults who crossed a border legally, enrolled in a school the federal government certified to enrol them, paid tuition to a private business, and sat down to learn the language of the country they were standing in. Every part of that is what this country says out loud it wants people to do.

There was a woman in an A1 class and I was trying to tell her that her own hands were fractals. I couldn't say it — no words to give her, none for her to receive them with. So I drew it on the board, badly, and pointed at her fingers branching into knuckles branching into creases, and she arrived before I could finish the sentence I couldn't build. She sat forward in her chair. I have taught at much higher levels since and I have never seen anything better.

There were carpenters in the same room who already knew what a fractal was. They knew it from the grain. Nobody had handed them the word, and nobody had ever suggested the word was for them.

That is Freire, in a room, on a Tuesday. Not as theory — as the daily finding that the ceiling on what a person can learn is almost never the idea. It is almost always something that happened to them before they reached it: some earlier authority, holding some earlier credential, having communicated that this material belongs to other people.

I have thought about that cassette cover more than once while teaching. A face behind a pair of open hands, and underneath it, the future in your hands.

IX

The school did not close, in the ordinary sense of the word. It stopped, and told nobody.

The staff were told it was shutting for renovations. The students were told they were on vacation. The first true statement anyone made about it was a For Lease sign that went up on the building, and it was not addressed to us. Notice reached the staff on 17 June, a week after the sign.

Months later students were still paying tuition to a school that did not exist. They could see the sign. Some of them photographed it. They went on paying anyway, because nobody with authority had told them otherwise.

The man whose job it was to tell them was the academic coordinator. He had been let go that same week, by email, with no notice and no severance, and he had nothing to tell the teachers either when they rang. That was me. Earlier in the spring I had been out on Elizabeth Avenue handing out flyers at twenty-four dollars an hour, recruiting students into a school that was about to close without telling them.

They were Ecuadorian, Paraguayan, Brazilian, Colombian. They worked all day and came at night, because the language was the last thing standing between them and everything else.

At fourteen I was in a band that sang in Portuguese, with one exception. That one song we wrote in English, and it was about not liking to. All four of us sat down to make it — Patricio and Álvaro, Jason doing the drum part, me doing the bass — and the joke was that English let you say absolutely anything you wanted, because nobody in the room would understand a word of it. It was a hiding place. We chose English for that song on purpose.

It was called You Don't Understand Me. And what the song complained about was the crowd nodding along. Heads going up and down in agreement, to words not one person there could follow.

Everything I have written here is that complaint. A room agreeing with a claim nobody in it has checked. We set the experiment up ourselves — picked a language the audience did not have — and then were furious at the result, which was that they agreed anyway. At fourteen I thought the song was about being misheard. It was not. It was about what agreement is worth when nobody is in a position to check it.

At fourteen I thought all that was funny. At nineteen I was teaching the language.

Not better than the people who taught me — that thought does not survive contact with itself, and I want to strike it before it settles. I knew a fraction of what they knew. What I had was not more. It was a position.

To cross a phase transition you have to be at the cusp. Not down inside one phase or the other, where everything is settled and nothing is moving, but on the border, which is the only place the crossing is actually happening. At nineteen I was standing on that line. I had come over it the year before and could still see it behind me. My teachers were far into the other phase, where the difficulty had gone invisible — which is not a failing. It is what mastery is for.

What took me another twenty-five years to understand is that the position is the whole thing, and you have to stay in it. Permanently. Keep walking yourself back to the edge of what you do not have. That is not a teaching technique and it is not a personality. It is where the work is, and it is the only place it is.

Thirty years later I was standing on the other side of the same language, in a storefront in Elizabeth, teaching it to people for whom not being understood was the entire problem. The thing the four of us used as cover was the thing my students could not afford to be without. I have thought about that song more in the last two years than in the thirty before them.

The pay had already stopped. Nobody was paid for anything after May, and the teachers were left to work it out among ourselves — ringing each other to establish whether there was still a school. A month of that, then a sign, then a letter.

I was paid for forty hours a week however many I worked, which is lawful here. It is not lawful in Brazil. The constitution there fixes overtime at no less than half again the hourly rate, the law caps it at two hours a day, and an employer who did to its staff what this one did would be answering for it in a labour court, with the back pay running years.

I want to be plain about the shape of that, because it is the shape of this whole essay. The country whose government was removed in 1964 with American help, on the argument that it was drifting somewhere dangerous, protects a worker better than the country I was standing in when mine stopped paying me.

I want to stay on that For Lease sign, because it is this entire essay standing on a street in Elizabeth. The evidence was on the building. It was photographable, and it was photographed. What it was not was authorised. And so people with every reason to act on it went on paying, month after month, because a check they could run with their own eyes did not carry the standing of a statement from someone entitled to make one.

This is not a story about an unusual company.

I live in Newark. Mars invented the M&M in this city in 1940 — chocolate sealed in a sugar shell, the trick taken from soldiers eating coated chocolate in the Spanish Civil War, which is a coincidence I would cut if I were making it up. The company was in Newark about eighty-five years. It is not gone; it is still very much in business, just in another state. This year it announced that it will close its United States snacking headquarters here — three hundred and seven jobs, under the state's layoff notice, with the building to go by the end of 2027 — and consolidate in Chicago, in Illinois, where the global head office already sits. The plant out in Hackettstown stays. Newark made the fortune. Newark got the notice.

There is no hidden hand in that and it is not a conspiracy. It is the plainest arithmetic there is: a firm arrives, makes its value out of a place and the people standing in it, and leaves when the numbers change. I followed that chain backwards once, from a padlocked Newark loading dock to a cocoa grove — one candy bar, with a child at one end of it and a laid-off line worker at the other, and the fortune travelling light between them. Sugar, cacao, tea, pepper: four mild drugs a civilisation learned to call food. The genius of a drug-food is that the person who craves it never has to see the end where the cost is paid.

A storefront language school in Elizabeth is that same move, three hundred times smaller, with one feature added. When Mars left, the workers lost their jobs. When the school left, the students lost the paperwork that made them lawfully present — because the thing that walked out was also the thing holding their standing.

Here is what people outside this work do not know.

When a school like that closes, the students do not merely lose a school. It is not only a place where classes happen — it is the institution that holds a person's standing. There is an official at every such school whose title is Principal Designated School Official, and that person signs the records by which a student is lawfully present. The school is the paperwork. The paperwork is the permission. When the school ends, the permission ends with it, and what was a tuition receipt yesterday is a countdown today.

And now Felipe's line, which I did not understand until this year.

The record that makes an international student legal is the same record that makes them findable. One database. The document that proves you are complying is the document that locates you the instant compliance stops being possible — including when it stops through no act of your own, because a private business in New Jersey shut its doors. The apparatus that enrols you and the apparatus that removes you are not two systems that happen to share a file. They are one system with two functions, and you were required to feed it accurately, in your own hand, as the price of sitting in the room.

And the warning was never about who carries the plate. It is about the document — the account the house gives of what it is serving you.

The record is the menu.

Nobody in that classroom did anything wrong. They had complied with every instruction they were ever given. Compliance was not the relevant variable. The hand was real, the invitation was real, the tuition was cashed, and the hand was never holding them up. It was positioning them.

I have a name for what happened to them, and a proof of it, in work that has nothing to do with schools.

A control — a regulator, a credit committee, a court — needs three things before it can act on anything at all. It has to know which rule applies. It has to have some means of reaching whoever broke it. And it has to be able to attribute the act to somebody. Call them ε, δ and α. They are jointly necessary and they are independent, and that is the whole danger: you do not have to corrupt a system to defeat it. Take out any one of the three and the sanction does not become weaker. It becomes addressee-less. This is why raising penalties almost never works.

The most effective defeat of a control is not a lie. It is a true description, correctly filed, that reaches a reviewer whose competence does not cover what is wrong with it.

Read the school against that. Every record it filed was accurate. Enrolment, attendance, hours, addresses — true, in the right format, on time, into the federal system certified to receive them. No field on that form asks whether the business will make payroll in May. The reviewer who reads the immigration file is not the reviewer who would notice a school going insolvent, and nobody routed it to the second one, because nothing in the filing was false. Nobody lied. Nobody had to.

I worked that argument out on a trade-finance structure, where a bank raises capital with money it lends, secured on the capital it raises, and every leg passes review because every leg is conventional — the advance rates ordinary, the collateral good, the goods real and actually shipped. The defect is not in any leg. It is in which desk reads the file. I did not expect to find the same shape in a storefront on Elizabeth Avenue, and I certainly did not expect the storefront to be the clearer case.

I have had that cassette cover in front of me for thirty years and I want to tell you what I did with it, because it is the same failure as everything else in this essay.

I read the words. O futuro em suas mãos. The future in your hands — I took it as a slogan, an exhortation, the thing you say to a fourteen-year-old to get him to pick up an instrument, and it worked, and I have been repeating it ever since.

I never looked at the drawing. It is two hands, raised open beside a face, and the fingers end short. They are drawn as stumps. The banner with the title on it runs straight across them.

So the cover is not an exhortation. It is a question with the answer already in it, and the question is the one I still had not asked when I wrote the paragraph above about a candy bar with a child at one end of it: which hands, and when did they chop them.

Because that is the actual history, not a figure of speech. In the Caribbean sugar mills the cane went into vertical rollers and an exhausted person did not always let go in time, and the standard equipment kept beside the machine was an axe, to take the arm off before the rest followed it in. That is what the sweetness cost, at the machine, in the century the fortunes were made. And in the Congo Free State the rubber quota was enforced by militia who had to account for every cartridge they fired, and the accounting instrument was a severed right hand — one hand produced per bullet, to prove none had been wasted. On 14 May 1904 a missionary photographed a man named Nsala on a veranda in Baringa, looking at the hand and foot of his five-year-old daughter, which was all he had been able to carry away.

Nobody hid that either. It was filed, audited, and photographed.

The song titled O Futuro Em Suas Mãos says it in words, and the words are worse than the drawing. It opens on the slogan, the future is in your hands, and then walks the man through it. He takes a job at a sisal rope factory. They tell him on the first day to pay attention, because the machines cut off hands, and the insurance only gives you a pension if you lose both. In the fifth month the machine takes one. He does the sum out loud: he will cut off the other the day he goes back to work.

That is not a lament. It is arithmetic, done by a man who has understood exactly what the system pays for and what it does not, and has acted on it. It is the axe beside the cane roller and the hand in the basket once more, with one difference: here the man is holding the axe himself.

And the room I have been describing to you was full of hands. There were carpenters in it who could read a fractal off the grain of a board, and I stood at the front drawing branching on a whiteboard to tell a woman that her own fingers were the same object, and I meant it as wonder, and it is wonder. I never once asked, in four months in that building, which of those hands had already paid for something, or what the work had already taken off them. I was too busy being moved by the shape.

The future is in your hands. Count them.

I taught those people the conditional tense. I would like to be able to tell you I taught them something that turned out to be more use.

X

Someone is going to ask whether I therefore want a self-taught surgeon. It is the standard objection and it deserves a straight answer, because it is where this argument is actually tested.

My position is that if a thing pertains to me, it is my business to know it. Not to be briefed on it — to know it. That principle is the basis of informed consent, of the right to your own medical records, of open science, of right-to-repair. It is also how some of the most important corrections in medicine happened: the AIDS activists of the late eighties taught themselves virology and trial design well enough to argue with the FDA, they were right, and the trial system they forced open is the one we all use now. None of them had the credential. All of them did the work.

So yes, I will do clinical work being no doctor, in the sense that matters. I read the primary literature on anything that concerns me or mine, and I do not accept a summary in place of data available to me. I already do it professionally: I am principal investigator on a public-health proposal that issues a daily green, yellow or red call on outdoor activity during wildfire smoke, checked against each jurisdiction's own written policy. No medical degree. It is sound, and for a specific reason.

The reason is the same one that makes the proof corpus sound, and it is the only thing separating DIY from arrogance. A credential is a proxy for having checked. Where you can replace the proxy with the actual check, the check wins. Where you can't, the proxy is doing real work and you had better respect it.

The kernel is an actual check. Trial data, replication, the primary source, a historian going through the diaries — all actual checks, public, available to anyone who will do the labour. That door is open, and the people who say otherwise are usually protecting a position.

But the door is the checking. It is not the feeling of having got past the gatekeepers. Skipping the check and keeping the confidence is exactly the base-ten error, and in mathematics that costs you a retraction, which I paid. In clinical work it costs somebody their body. The stakes change what the error costs. They do not change what the error is.

That is the honest form of the punk claim and it is stronger than the loose one. Not nobody can tell me what's true. Rather: show me the check, and I will do it myself, and then neither of us has to take the other's word for anything.

XI

So here is what I carry into a classroom, and it is not contact geometry.

The student who believes she cannot learn the thing has almost always been told so by a credential. Not by a person — by a structure. A placement test, a transcript, a grade, a form, a status. Something that looked at one number and returned a verdict on a whole human being, and she believed it, because the verdict arrived wearing the clothes of objectivity.

I know how that feels from the inside. When I looked at doctoral programmes this autumn, the honest assessment was that the 2.62 would stop the application before any person read a word of it. The research would not be weighed and found wanting. It would not be weighed.

The kernel does not know about the 2.62. It checked the proofs anyway.

That is the thing I have to teach that is worth more than the material: the difference between you cannot do this and a filter rejected you. They are not the same statement. They are confused constantly, and on purpose, by institutions that benefit from the confusion. Telling them apart is a skill, it can be taught, and it is the precondition for teaching anything else to an adult who has already been sorted once.

The practical version is unglamorous. I build my own materials and give them away — course sites, problem sets, the whole stack for teaching this to kids, self-hosted, open, free, no publisher, no adoption committee, no sales rep. It has a build script and a licence instead of a photocopier and a staple, and structurally it is a zine. My research papers are zines with an axiom checker bolted on.

I did not get there through educational theory. I got there because the older band in town let four kids use the gear for ten minutes.

XII

I don't have a redemptive ending, and I notice the essay wants one, and I am declining, because supplying it would be the same error as the base-ten theorem: a satisfying sentence the evidence does not support.

And there is one knot in this I am not going to untie for you, because I have not untied it for myself.

The man who taught me to say the thing out loud and print my address on the back of it is the same man who handed me the one claim in my life I can never run the method on. What Felipe told me about that book is the least checkable evidence there is: one person, remembered, across twenty years and two countries, and now dead. No diary at the National Intelligence University is going to settle it. Nobody is going to open a file.

I have spent a decade building a machine whose entire purpose is that nobody should ever have to take my word for anything. I built it because of a man whose word I took.

I do not think that is hypocrisy. I think that is just what it is to come from somewhere.

What I have is a method, and I learned it from a cassette.

Say the thing out loud. Print your address on the back. Itemise what you are standing on and mark the parts you cannot prove. Nobody is going to give you permission, and the institution holding your standing can vanish inside a month and take your standing with it.

So build the thing yourself, publish what it depends on, hand it to people for free, and make it checkable by someone who has no reason to trust you — because that is the only durable legitimacy there is, and unlike the other kind, nobody can put it up for lease.

There is a courthouse across the street from where the school was. Broad Street, in Elizabeth, which was the first capital of New Jersey before anyone thought to move it. Carved across the front of the building, legible from the pavement outside a shut storefront school: vox populi, vox dei. The voice of the people is the voice of God.

I walked past it for four months. On one side of that street a building says the people's voice is divine. On the other, a school told its students they were on holiday and went on taking their money, and the first honest word spoken on the block came from a landlord.

And I had asked it before. M.O.F.O. had a song built on that same question — put as a question, because at fourteen in Brasília it did not look like a settled matter. I was the bass player. I did not know there was a building in New Jersey with the answer already carved on it, or that I would spend four months walking past that claim on my way into a school that was lying to its students.

If the future is in our hands, and if the voice of the people is the voice of God, then we ought to get a say in it. That is not a demand. It is just reading the two buildings against each other and noticing that they do not agree.

Fourteen years before that cassette, on the other side of the Atlantic, a generation had already worked this out and pressed it onto a seven-inch.

In 1978 six people were arrested in Britain and charged with conspiracy to cause explosions. The material recovered was weed killer, sugar, flour and some electrical wire. The trial ran sixty-one days; in December 1979 four of them were acquitted, one had pleaded guilty and gone down for nine years, and one had jumped bail. The case is remembered by the phrase in the charge, because they were accused of conspiring with each other and with persons unknown.

Crass and the Poison Girls put out a split single to pay for the defence — Bloody Revolutions on one side, Persons Unknown on the other, 1980. It raised something near ten thousand pounds, which Penny Rimbaud later called a lot of money in those days. The defence did not end up needing it, so they spent it on a building: the Wapping Autonomy Centre. A prosecutor's turn of phrase paid for a room with a PA in it.

What the Poison Girls did with that phrase is the part worth stealing. The state used persons unknown to make an act attributable to nobody in particular, and therefore to anybody it fancied. The song takes the same words and hands them back the other way up: the unknown persons are the woman on the bus, the man in the queue, you, me, each and every one of us — and once the category is that large it is no longer a weapon, because there is nobody outside it to point it at.

That is the third condition again, arriving from the other end. A control fails when an act cannot be pinned to an identifiable person. Here a state stretched attribution deliberately — conspiracy with whoever — and the answer was not to plead that you were an individual and had nothing to do with it. The answer was to make the class so big the charge dissolved in it. Which is the exact inverse of the school on Broad Street, where the paperwork was immaculate and the class of people answerable for it came out empty.

And Vernon Walters, on a photocopied cover in Brasília, does the opposite thing for the same reason. Refuses the anonymity. Prints the block, the building and the apartment number, and asks you to write. Both are answers to one question: whether you are going to let somebody else decide whether you have a name.

One more thing off that label, because it is the best small example I know of what this essay is about. In 1981 Flux of Pink Indians released an EP called Neu Smell on Crass Records. Its second track is Tube Disaster, written out of the Moorgate crash of 28 February 1975 — forty-three people killed on the Northern City Line when a train did not stop, a cause never established, and every dead-end platform on the Underground fitted afterwards with an automatic device still called a Moorgate control. Its fourth track is Sick Butchers, and the sick butchers are us. Scratched into the run-out groove of side one: ALL ANIMALS ARE EQUAL. On side two: BUT SOME ARE MORE EQUAL THAN OTHERS.

You will never find those two lines unless you are holding the record and you tilt it under a light. They are not on the sleeve. They are not in the songs. Somebody put a message in the one place that can only be reached by a person who physically has the object and takes the trouble to look — which is this entire essay, cut into dead wax in 1981 by people who were not trying to make the point.

And here is the part nobody can reconcile, which is exactly why that label kept saying it.

Sick Butchers is not about butchers. It is aimed at the room. Flux stood on a stage and told the people who had paid to get in that they were the ones doing the harm — not the government, not the police, not a villain somewhere else: the person standing there who has never done a wrong thing in his life, and eats.

Everybody thinks of themselves as a good person. I do. You do. The woman who filed the enrolment records did, and her filings were accurate. The landlord was within his rights. The teachers ringing each other through June were good people with no information. And I was a good person on Elizabeth Avenue at twenty-four dollars an hour, handing flyers to people about a school that was going to go out from under them.

A system does not run on bad people. There are not enough of them and they are unreliable. It runs on good people doing correct work in a sequence where no single part is the wrong part. That is what those three conditions describe. It is what nobody lied and nobody had to means. And it is why Felipe's warning was never really about the garbage man. A bad man in the kitchen is a small problem and a findable one. The problem is a good man handing you a true menu.

My own country has the cleanest proof of it I know, and it is twelve days before the one everybody remembers. On 19 March 1964 — the feast of Saint Joseph, patron of families — somewhere between three and five hundred thousand people walked through São Paulo in the Marcha da Família com Deus pela Liberdade, the March of the Family with God for Liberty. It was organised by Catholic women's associations, a civic union and the state industrial federation. They carried a manifesto demanding the president's removal. They were not soldiers and most of them were not cynics. They believed they were protecting their homes from something that was coming to take them.

After the first of April the marches went on, under a new name. They were called Victory Marches. About a million people in Rio on 2 April. Forty-nine marches across the country between March and June, which is the same stretch of weeks in which a literacy programme for five million adults was cancelled by decree and its materials ordered recovered.

And somebody had paid for the manners.

Two organisations had spent the preceding years building the sentiment those families marched on. IBAD, from 1959, and IPES, founded in November 1961 — both presenting themselves as civic and educational, both financed by Brazilian and foreign business. The historian René Dreifuss put the take at around four million dollars inside two years, drawn from hundreds of companies, among them 297 identified American ones: Texaco, Shell, Standard Oil, General Electric, IBM, Coca-Cola, General Motors. IPES made the books, the pamphlets, the magazines and the leaflets and handed them out free, and ran coordinated campaigns through the newspapers and the broadcasters. IBAD financed something like two hundred parliamentary candidates in 1962. A congressional inquiry in 1963 found IBAD guilty of electoral corruption and ordered it closed.

Then the government fell, and the men from those two institutes took the Finance Ministry, the Planning Ministry, the Central Bank and the BNDE.

The papers ran with it. On 2 April 1964 O Globo carried an editorial signed by Roberto Marinho: Ressurge a Democracia — democracy resurges. O Estado de S. Paulo, Folha, Jornal do Brasil and Correio da Manhã backed it too. Forty-nine years later, in 2013, O Globo published an admission that the support had been an error, and that other editorial decisions of the period followed from that original mistake.

And it is worth asking what the threat was, in the units the people who name threats normally use.

The opposing power in this case had no military budget worth the name. It was a president proposing to expropriate land and nationalise oil refineries, and a literacy programme aimed at five million adults, which was cancelled by decree six weeks after the marches began, with an instruction to recover all the materials. That is the size of the thing that bought twenty-one years.

Compare it to the real article, where the numbers are published. Last year the United States spent nine hundred and fifty-four billion dollars on its military, a third of the entire world's total, and two point eight times the next largest spender. And even that is a number with an assumption inside it. In 1976 the CIA revised its estimate of Soviet defence spending from under eight per cent of Soviet output to twelve or fourteen — not because Moscow had doubled anything, but because new pricing data and a defector's procurement figures changed how the existing spending was valued. Nobody lied. The estimate was competent, correctly filed, and reached readers in no position to audit the pricing assumption buried in it, and the number was the argument, and the argument moved budgets.

These things do not contract. Whatever you decide it is — overreaction, procurement, or the ordinary appetite of an institution — it does not get smaller afterwards.

There is a cashew tree on the coast of Rio Grande do Norte that covers somewhere between seven and eight and a half thousand square metres, the size of about seventy ordinary cashew trees, and drops sixty thousand fruits a year. Visitors assume it is a grove. It is one tree. It spreads because the branches grow heavy, bend down under their own weight, touch the ground and root there — and each new root sends up what looks like another tree, which is not another tree.

That cajueiro is not going to get smaller either.

I have a name for this and I have been using it in other work for years without noticing where it belonged. The cajueiro principle: a system tends to complexity. The seed tends to forest. What begins as one root does not stay one thing — it puts out weight, the weight bends the branch down, where it touches it roots, and what stands there afterwards looks like seventy separate trees and answers to nobody as a single organism.

Let me mark the grade on that, since grading claims is the whole argument of this essay. It is not a theorem. It is a named observation. The strongest thing standing behind it is a toy model I have checked mechanically, which crosses a robustness threshold I can state precisely — and that is a result about the toy model. It is not a proof about ministries, or militaries, or databases, or schools. I am telling you the shape I think I see and exactly how far I have checked it, which is as far as anyone gets to go.

But read back over what you have just been told. Two propaganda institutes become a finance ministry, a planning ministry, a central bank and a development bank. A pricing assumption becomes a force posture. An enrolment record becomes a location. A tuition receipt becomes a countdown. Not one of those was a separate tree.

And the word was sitting there the whole time. I distrust a pun that arrives this neatly, so I went and checked the dates before I let myself use it.

They are called branches. On 18 September 1947, under the National Security Act, the air arm of the United States Army got enough purchase and rooted, and what stood up out of the ground was the United States Air Force. On 20 December 2019, under the National Defense Authorization Act, a piece of the Air Force got enough purchase and rooted, and what stood up was the Space Force.

A branch that reaches the ground does not become a longer branch. It becomes a trunk — with a budget line, a procurement pipeline, a chief with a seat at the table, and a permanent institutional reason to exist which must thereafter be maintained. And it is never folded back in, because by then it is not a branch any more. It is a tree, and nobody prunes a tree by pointing out that it used to be part of another one.

Now hold both halves at once, because this is where the argument is either honest or worthless.

The money was real. The organisation was real. The campaign was professional, partly foreign-funded, and it worked.

And the people in the street were not actors. Half a million of them came out with their children on a saint's day because they had been supplied, free, for three years, with a completely coherent account of what was about to be done to their homes — and inside that account, marching was the decent thing to do. Nobody had to lie to them on the day. By then it was their own conviction, sincerely held, and they were marching on it. The persuasion had been finished months earlier, by people they would never meet, and what stood in the street was the finished product wearing its own face.

I did not notice until I wrote this paragraph that it is the same finding as my own: the most effective defeat of a control is not a lie.

Which is worth keeping in mind whenever the word march gets used flatly, as though a crowd were a crowd. A crowd assembled against power and a crowd assembled by it look identical from a helicopter and photograph the same in the paper. The difference is not in the faces and it is not in the sincerity, which may be total in both. It is in who paid for the leaflets, and that is a question with an answer, and the answer is usually filed somewhere.

Not one person in those crowds was the villain of the story as they understood it. They were the good people. They turned out with their families, in daylight, under the name of God and liberty, and what they were turning out for was twenty-one years.

And a Brazilian band has already written the whole of this section into a title. Punkzilla, on Hora de Acordar, 2017: Eu Tenho Medo das Pessoas de Bem. I am afraid of the good people. Not of criminals, not of the regime — of the respectable, the ones who are certain which side they are on, the guardians of everyone else's morals. Fifty-three years after the families walked through São Paulo, and it still has to be said, in Portuguese, by a punk band nobody outside the scene has heard of.

The good people are not the exception to the machine. They are the mechanism.

A handful of teenagers in Britain put that on a record in 1981 by turning round and pointing at their own audience, and the audience could not take it either. Nobody can. That is the reconciliation that does not happen, and it is why the music that says it never got large, and it is why it is still the most useful thing I was ever handed.

One more instance, and it is the one I would put at the head of the whole argument if this ran backwards.

In 1938, on the Gandesa front in the Ebro, Republican soldiers sang a song called Si me quieres escribir — if you want to write to me. The answer the song gives is an address. You know where I am. I am at the front, first line of fire. A Chilean, Rolando Alarcón, put it first on a record of Spanish Civil War songs thirty years later, credited to nobody: popular española.

The tune was not even theirs. It came off Spanish colonial units in the Rif, in Morocco, in the twenties — an empire's marching song, taken and turned around and sung back at fascism by people the empire had used it on. Nobody owned it. Nobody could be kept out. It is the same event again, in 1938, with rifles.

And what it does is give its address. Which is the same gesture as a photocopied sleeve with a street in Brasília printed on the back of an accusation. Here is where I am. Write to me. I am not moving and I am not hiding.

That is the position I have been trying to hold for thirty years and have not always held. Permanently at the cusp. First line of fire. If you want to write to me, you know where I will be.

So here is mine, and a question to go with it.

I was not paid after May. The notice came on the seventeenth of June, a week after the sign went up. New Jersey has a process for this, and it is not a lawsuit: an administrative claim to the Division of Wage and Hour Compliance, which investigates on your behalf, and which the state says plainly an employer may not retaliate against you for using.

I have not filed it. Somebody asked me to and I said no. Not because there is no one to file against: the company is still in business, in another state. Because I am forty-six and out of work, and I am afraid that a man who files a claim becomes a man with a claim attached to him, and that the next employer finds out somehow, and that I cannot afford to be that person this year.

Which is to say I have spent this entire essay telling you to run the check yourself, and then declined to run my own, out of fear, in precisely the terms of a song I have been carrying since I was a boy.

There is a paper of mine about exactly this, and I did not notice the overlap until both were written.

It concerns automated credit — margin calls, collateral triggers, forced liquidation. A system watches a ratio, and when the ratio breaks it acts, inside a window set by the model rather than by the person who has to answer. I called that window the response gap: the interval between a machine detecting a breach and doing something irreversible about it, which can be — and routinely is — shorter than an informed response requires. Then comes the second half, which is the half that does the lasting damage. The institution that built the window describes the loss as the borrower's poor judgment. That is the attribution gap. The first failure is mechanical. The second is narrative, and it works by moving the causal story off the system and onto the person.

I was careful in that paper to mark what I could show and what I could not, because in the general case forced and chosen look identical from outside. That is not a weakness in the argument. It is the finding — and it is why the corrections I proposed are also the instruments that would let anybody measure the thing in the first place.

Now read my own June against it. The clock was never mine. The pay stopped in May, the sign went up on the tenth of June, the letter arrived on the seventeenth. And here I am in the last section of my own essay describing the outcome as a decision — I am afraid, I cannot afford to be that person this year — which is the attribution gap operating on the author of the paper about the attribution gap.

I can write the mechanism down. I formalised it. I published it with a DOI. None of that stopped me doing the thing it describes.

So I will do the thing the essay actually recommends, and put it to a room that can see it from outside, where I cannot.

Should I file?

g6llc@proton.me. You know where I will be.

A man who had been a child when they took the government wrote o futuro em suas mãos on a photocopied cover, and then handed the guitars to a fourteen-year-old for ten minutes to find out whether it was true.

Play it fast.