Adam Kayal

La Veille

Adam Kayal — Thu, 08 Oct 2026 23:30:51 GMT

There was a message from the office when I got up: the president would like to see me at ten. I had already said no to Renaud. In nine years I had never refused the president anything except in his office, and I did not intend to start with this.

I walked down to Old Montreal. Saint-Denis all the way down, across Sherbrooke, past the university to Sainte-Catherine, across the long flat width of René-Lévesque, and after that the buildings went from brick to stone and the traffic went quiet, because past a certain point only the people who work there still come. The fog had burned off by nine. It was an ordinary September morning, cold enough for a coat and bright enough that I did not need one.

The last time I had walked this street was the morning after the review. Once a year, in September, the ones they rank highest sleep a night in the building so that the office can watch how we work. I had slept there, and walked home up Saint-Denis at seven, and found her gone. I did not call the police. A woman who leaves is not missing; they would have taken her name and a description and put them in a drawer, and finding things is what I do. I came back down Saint-Denis at ten and told them instead. I had walked it every day since in my head and not once on foot. It was the same street.

The building on Saint-Sulpice had no sign. It had never had one. The brass plate beside the door was worn down to a rectangle of dull metal that had once had letters on it, and nobody had replaced it in longer than I had been coming here, because the people who needed to find the door already knew where it was.

Inside, the hallway smelled of stone, radiator heat, and under both of those the paper smell that never leaves a building where paper is kept. The three portraits were where they had always been, going up the first flight of stairs in order. Brassard hangs at the bottom, in a dark suit, looking at something past the painter’s shoulder rather than at whoever is looking at him. Desrosiers is halfway up, younger than I have ever managed to picture him, with his hands folded on a desk that no longer exists anywhere but in the painting. Cartier is at the top, under the landing window, the only one photographed rather than painted, black and white, not smiling, looking directly into the camera as if she had asked for it to be pointed at her rather than the other way around.

Under each portrait there was a small brass plaque. Under Brassard’s: A child who believes the door is locked will not try the handle. The whole of teaching is convincing them the door is a door. Under Desrosiers’s: You already own what you need. Under Cartier’s: Watch the cat. She sleeps most of her life, and she always lands on her feet. It is the first sentence of the Playbook, and the only one every trainee can say without looking. I had read all three often enough, in nine years, that I no longer read them most mornings. I only registered that they were there, the shape of the words familiar enough to skip without losing anything.

A door on the second floor stood open, and a trainer’s voice came through it, mid-sentence.

“...realizing, navigating, remembering, reading. In that order and never any other, and if you tell me you remembered before you realized, I will tell you that you are describing something else.” A pause, not unkind. “Adrien has been doing this for nine years and is still the fastest first discipline anyone in this building has ever clocked, and he will tell you himself that he does not know why, so do not ask him to explain it to either of us.”

I did not stop walking. I have heard my own name used this way before, in rooms I was not meant to be listening from, and it does the same thing to me every time, which is that I walk a little faster past the door.

The reading room was on the ground floor, past the stairs, behind a door with a small window of frosted glass. I did not go in. The room is for members carrying an assignment, and I was not carrying one, so the door was, as far as I knew, locked to me, but I stood in front of it for a moment anyway, because I knew what was on the other side: the parish registers, and the gazetteer with every name a place has ever had, and somewhere in the cabinets, if it existed anywhere on paper at all, a town on the water that I had never been able to name.

I went up.

The president’s office is on the third floor, above the training rooms, at the end of a corridor with more doors than the building seemed to have room for. He was already standing when the assistant let me in, which he did to make you feel considered rather than kept waiting, and it worked on me every time even though I knew exactly what it was.

“Adrien.” He did not shake my hand. He gestured at the chair across from him, and I sat, and he sat down after me, behind a desk with nothing on it but a single closed folder. “Renaud told you about Thursday.”

“He did.”

“And you said no.”

“I did.”

He nodded, as if this were the answer he had planned around, which it probably was. “The Institute does not need you to want the job,” he said. “It needs the client in Toronto to have an answer by Friday morning, and you are the person who gets it to them clean. I am not going to pretend otherwise.”

I noted, as I always did, that he said the Institute, and that in nine years the word had never once come out of my own mouth.

“I am not taking jobs,” I said, “until I find her.”

“I know.” He opened the folder without looking down at it, a habit that used to unsettle me and now only told me that he wanted his hands to have something to do. “Which is why I am not asking you to take this one for its own sake. The reading room is for members carrying an assignment, and you are not carrying one. I can open it to you today. Every record we hold, for as long as you need it: whatever it is you have been trying to get at without us.”

I sat with that for a moment. It was exactly what I needed, and what I had told myself, since she left, that I would not ask for.

“For the job,” I said.

“For the job. Three days to read, one night, one report. You have done it countless times.”

“And after Friday, the room stays open.”

“For as long as you are carrying assignments.” He closed the folder again, having found nothing in it that he needed. “I would rather you were tired from real work than tired from nothing at all.”

I did not answer that. He was right, and I was not going to tell him so. I looked at the folder instead of at him, and I heard myself say the day before I had decided to say it.

“Thursday,” I said.

“Thursday.”

Renaud was in the kitchen on the ground floor when I came down, standing at the counter with two cups already made, one of them clearly poured before he knew whether I would want it, which was the kind of thing Renaud did without seeming to notice that he was doing it.

“You said yes,” he said.

“I said yes to the room.”

“That’s the same thing to them.” He passed the cup across the counter. The coffee was strong enough that I smelled it before I noticed how tired he looked, which told me something about the last two days that he had not said on the phone. He had taken something on, an extra night, or two, and he did not mention it, and I did not ask, because that was the arrangement we had arrived at over nine years without ever discussing it.

A colleague I had worked jobs with for years came in behind me for the pot, and poured her coffee, and leaned against the counter with the air of someone planning to stay for the whole conversation.

“We were talking about you yesterday,” she said, speaking to Renaud so that she could say it to me. “About the day you came in and told us. None of us had any idea you were even seeing anyone. One week you were single, and the next you had a wife, and she’d gone.”

Renaud did not say anything. He was looking into his cup.

“It happened quickly,” I said, which was true, and which was also the sentence I always gave to people who asked, because it ended the question without answering it.

“I’ll bet.” She smiled wistfully into her cup and took it back down the hall.

I finished mine standing there with Renaud, and neither of us said anything else about it.

I left a little after eleven, with the room open to me again and Thursday marked, and it was not until I had crossed back over René-Lévesque that I understood what I had agreed to. I was going to work for them again so that I could look for her in their building.

The fog was gone entirely by the time I reached Saint-Denis, and the street looked ordinary at noon: buses, a delivery truck double-parked outside the pharmacy, people with somewhere to be.

The superintendent of my building was in the lobby, on a stepladder, doing something to the fixture over the mailboxes that had needed doing since summer. He does not work quickly, and he likes to talk while he works.

“You’re back early,” he said, without turning around.

“I had a meeting.”

“Any news of your wife?”

I had asked him, on the morning she left, whether he had seen her go out. He had not, and he had been kind about it.

“She’s away,” I said. It was, of all the things I had said about her since she left, the closest to true.

He climbed down to move the ladder, and while he did he asked, over his shoulder, whether we had any wedding photos up yet, because his daughter had just gotten married and he had already run out of wall for the frames; did I want the number of the frame shop on Mont-Royal, they gave a good rate to anyone in the building.

“We do not have any,” I said. “From the wedding.”

He looked at me then, surprised, deciding whether to ask about it.

“No photographer?”

“No photographer. No guests. We wanted it to be the two of us and nobody else, and a priest to say the words, and he did not ask why we wanted it that way, which was the only thing either of us asked of him.”

I told him about the beach, because once I had started I did not know how to stop in the middle of it, and because it is the easiest story I own, having told it enough times that it comes out of me now the way a held breath comes out, all at once. It was white sand, and flat water, and no wind at all. Mia was in a plain dress too long for the sand, holding it up with one hand, laughing at me because I had not thought to take my shoes off before the water reached them, and there was an old priest with his book open and his shoes as wet as mine. I told it the whole way through, standing in the lobby with the ladder between us, and I did not leave anything out.

He listened the way people listen to a story they can tell is true, his weight on one foot and the ladder forgotten in his hand, and when I finished he said it sounded like the right way to do it, and that he wished he had thought of something like it himself, twenty years ago, instead of the reception his wife’s family had insisted on.

I agreed that it was the right way to do it. Nobody has ever told me otherwise.

I went upstairs and put the kettle on. There had been no dream worth keeping the night before, only the ordinary static of a night’s sleep with nothing shaped enough to write down, so the notebook stayed on the floor where I had left it. The sheet with her name on it was still on the desk, with nothing under the name. I told myself that this was not the same as deciding not to do it.

Thursday was three days away. I would spend them reading about a company in Zurich I had never heard of before yesterday, and on the third night I would prime for a client I would never meet, and all that time the reading room on Saint-Sulpice would be open to me, with every register on the coast in it, and I had not yet opened one.

Ninety Degrees

Adam Kayal — Tue, 06 Oct 2026 22:14:38 GMT

I’m writing a novel. The plan is twenty-one chapters of about ten pages each, roughly two hundred and ten pages in all. When I made that plan, I had written none of them.

Some evenings I sat with the manuscript for three hours and came away with half a page. Other evenings I wrote a full scene in forty minutes. The number of hours spent on the task didn’t predict the amount of pages I produced.

Everyone knows a book takes work, and what most people mean by work is time. You put in the hours and the pages come. That view accounts for part of what happened at my desk. It doesn’t explain how three hours could produce less than forty minutes.

When I studied mathematics and physics, I learned a stricter definition of work. It fits this problem better than the everyday one.

Time is one factor out of two

In physics, when a force moves something, it transfers energy to it, and that transferred energy is called work. Work is what takes a system from one energy state to another. When power stays constant, the formula is short: work equals power multiplied by time.

Writing a book is a change of state too. At the start I’ve written zero books, and at the end I’ve written one. The progress between those two states is the work. In the formula, work has two factors. Time is the one everyone already counts. Power is the one that decides what an hour is worth.

So what is power, for someone sitting at a desk?

Only the part of the push that points forward counts

If you’ve ever pushed a car, you’ve felt the answer. Push from behind and it speeds up. Push on its door and you can strain as hard as you like; it rolls on no faster.

Physics writes this as

Force is the push: how hard, and in which direction. Velocity is how fast the object moves and where it’s heading. θ is the angle between the push and the motion. When you push straight along the motion, θ is zero, its cosine is 1, and all of the push counts. At ninety degrees the cosine is 0, and none of it counts, however hard you push.

For mental work, I map force to focus. The size of the force is how deep your focus is: how much of your attention sits on one thing. Its direction is what the focus is pointed at. The direction of motion is wherever the goal needs you to go next, which for the novel is the next scene. Progress, then, is the depth of your focus, times how closely it’s aligned with the goal, times the time you spend.

Deep focus at ninety degrees moves nothing

Picture an hour of scrolling on your phone. You’re absorbed. You don’t notice the time go, you don’t hear your name the first time someone says it, and when you look up the hour is gone.

You can argue that this counts as deep focus. However, no matter the depth of focus, the angle between the direction of that focus and the goal of writing a novel is clearly ninety degrees, so the hour adds nothing to the book.

Most misalignment is harder to spot. An hour spent on the book can still point slightly the wrong way: rewriting a scene that already worked, or drafting a chapter before knowing what it has to accomplish. The cosine of a small angle is close to 1, so that hour isn’t lost, but part of it goes sideways.

A small angle is where a plan earns its place. Before drafting chapters, I wrote a sheet of world rules and planned a beat sheet for each chapter: what happens in it, and what the reader needs to know by the end. A beat sheet is the direction of motion written down. Without one, the first part of every session goes to working out which way is forward.

In a previous piece, I argued that direction is the only lever the conscious mind has. The formula shows how much rides on that lever: aim is the factor that can multiply everything else by zero.

Once the aim is right, the question becomes how hard you can push.

Depth has to be built before it can be used

At its best, depth is what people call flow: the stretch where the next sentence is ready before you finish typing the last one, and an interruption is hard to recover from. It rarely arrives in the first few minutes, and one interruption can reset it.

There’s a physical side to focus that is well documented. When a region of the brain becomes more active, blood flow to that region rises within seconds, bringing it more oxygen and glucose. The response is called functional hyperemia, and it’s what fMRI scans rely on to show which regions are working.

The more interesting finding is what happens before the work starts. In a 1999 study led by Sabine Kastner, people in a brain scanner kept their eyes on a point in the centre of a screen and were told that something would soon appear off to one side. Before anything appeared, activity rose in the part of their visual cortex that handles that side of the visual field. fMRI tracks activity through blood oxygen, so what the scanner picked up was oxygen-rich blood arriving at the regions about to be needed, ahead of the task itself. Expecting the task was enough to start preparing for it.

The people in that experiment were waiting for a spot of light, not writing a chapter, so applying the result to a writing session is my reading rather than what the study showed. If it carries over, some of the depth I reach in a session is decided before I sit down, by whether I know what I’m about to work on. Sitting down with the next scene already chosen gives the preparation a target. Checking my phone halfway through starts it over for something else.

The practical point doesn’t depend on the study. Depth takes time to build, so a session that keeps getting broken up spends most of its time building and little of it at depth.

That leaves time, the factor nobody needs convincing about.

An hour multiplies whatever you bring to it

Two hundred and ten pages don’t get written in one deep, well-aimed evening. Nothing in the formula lets you skip the hours.

What the formula changes is what an hour is. It’s a multiplier. Multiplied by deep focus pointed along a beat sheet, it moves the book. Multiplied by an hour of scrolling, it gives zero, however many of those hours you stack. Three hours at a bad angle can lose to forty minutes at a good one, which is what I’d experienced at my desk many times before.

Most advice about goals is about the hours. The other two factors are harder to see, and they decide what the hours are worth.

Somewhere ahead is the state where the book exists, two hundred and ten pages I can hold in my hands. Between here and there is a fixed amount of work, and I can’t make it smaller. What I can choose is which way I’m facing when I sit down.

He Didn't Have to Earn It

Adam Kayal — Thu, 01 Oct 2026 00:40:56 GMT

My mom recently sent me an old photo of myself as a kid.

My first reaction was the one anyone has to the sight of a small child: I loved him and I wanted to protect him. It didn’t occur to me to ask what he had done to deserve that. He was a kid, and a kid’s worth doesn't depend on anything he does.

Then the obvious fact hit me: that kid is me. I’m the same person, grown, and for most of my life I’ve refused myself the verdict I gave him without thinking. So why did I believe that the grown version had to earn the same love and protection?

I paid for my worth in achievements

Like a lot of high achievers, I’ve struggled with low self-worth for as long as I can remember, and that struggle is where most of my drive came from. I set a goal and work toward it until I hit it. And for a very short while afterward I feel like I’ve earned my place. Then the feeling fades, and the only way I know to get it back is a new, bigger goal.

I remember the first time I benched 225 lb, or the first time I landed a six-figure job. These felt great in the moment, but shortly after I needed something else. By the next day the new targets were set: a 315 lb bench and seven figures a year.

Wanting to achieve things is healthy, and I still want to. The problem was where the wanting came from. I didn’t achieve in spite of low self-worth. I achieved because of it. Every goal I reached proved I was good enough for that goal, and the proof expired once I set the next one.

Psychologists call this contingent self-worth: self-esteem that rises and falls with how you do in whatever you've staked yourself on. Jennifer Crocker’s research on university students found that the ones who based their worth on grades felt bigger swings with each result, and didn’t earn better grades for it.

For many people, that stake is their job, or how much their partner seems to love them that day. Either way, your mood depends on someone else's verdict, and it changes whenever theirs does. How can you call yourself free when your worth is set each day by peers and partners?

In code, a loop exits when its condition becomes true. If the condition moves every time you get close, the loop never terminates, however fast each iteration runs. Feeling worthy was my exit condition, and each goal I reached moved it one goal further. It was always when I get this, or when I achieve that. Running the loop harder only produced more iterations.

What would change if the worth came first?

Worth has to come first

In the reversed order, you start out feeling worthy, and because of that you trust you’re able to go after what you want. Confidence comes from the worth, and achievement comes from the confidence.

The difference shows up the first time something fails. In the old loop, a missed goal counts as evidence against you, so every attempt puts your worth at stake. That makes hard attempts expensive, and it makes quitting early look sensible. When your worth is settled before you start, a failure costs you the attempt and nothing else. You can take on something harder, miss, and try again.

Knowing the right order is one thing. Feeling it is another.

The kid never had to earn it

The photo gave me that order. When I looked at that kid, I didn’t check what he’d accomplished. He hadn’t accomplished anything yet, and it made no difference. The verdict came first, before any evidence, which is the order I’d been failing to build for myself.

And that kid is me, at an age when I had achieved nothing. The worth I saw in him on sight was already there before any of my achievements existed. Nothing I’ve done since added to it, and no goal I miss can take it away.

One photo won’t shut down a loop that has run for most of my life, and I don’t expect it to. What changed is that I’ve seen the verdict I kept chasing, when it was already mine, long before I’d done anything.

If anyone told that kid he had to earn his place, I’d want to stand between him and that person. For most of my life, the one telling him was me.

When I see that picture, I want to hold him and tell him that I love him unconditionally. I want to make him smile and giggle just to see him happy. Because he’s a kid… because he deserves it.

I still want the goals. I want to go after them the way I’d want him to, already sure he’s enough before he starts.

That kid is me. I am worthy.

Eleven Days

Adam Kayal — Fri, 18 Sep 2026 13:54:00 GMT

The fog on Saint-Laurent had come down with the morning and decided to stay. It filled the street from the ground to the second floor and it did not move at all; the shop signs glimmered inside it without lighting anything around them. I walked through it for six blocks and it parted for me and closed behind me and I felt nothing, and then I stopped in front of the bakery, and it was suddenly all there was. The street ended twenty feet from me in either direction. The window of the bakery was the only thing with an edge left in the world, and everything I could see was arranged around it. How had I walked through this for six blocks without once looking up?

The bakery was full and warm and smelled of the inside of bread. I took my place behind a man reading a newspaper, and while I waited I read over his shoulder, a habit I have never managed to lose and have never much wanted to. He was on the second page. A photograph of a bridge at dusk, and under it three columns about something the city had promised and not done. I read the first column with him. He turned the page.

He was still on the second page. The photograph of the bridge was there again, and the three columns under it, and I had already read the first one. I read the first column again, slowly this time, giving every sentence the attention you give a thing you suspect of hiding something. It hid nothing. He turned the page again, and I let my eyes drift to the window, to the fog, because I had decided he had lost his place, and that a man losing his place in a newspaper was the most ordinary thing in the world. It was. I lost mine constantly.

The baker’s hands were long. I mean that they were a lot longer than normal hands should be, so that when she reached for the loaf on the top shelf she did not need to stretch, and did not need to look, and brought it down as easily as if the shelf had been at eye level. I thought she must have been very tall. Then I looked at her, and she was not tall, and I did not think she was tall anymore, because when you look at a person you see the person and not the sum of the things you had assumed about them. She was a short woman in her fifties with flour on her forearms, and light from the window on the flour, and that was all there was to see, so that was all I saw. Then it was nearly my turn, and I had not decided what to buy.

The woman two places ahead of me was wearing Mia’s coat.

The grey one with the torn lining she kept saying she would fix, and never fixed, because she liked the way it felt when she put her hand in the pocket and found the thread. I knew the coat the way you know a word in your own language. I knew it the way I knew her hand on my chest in the morning before either of us was awake. There was no moment of recognizing it. It was hers before I had finished seeing it, and something in me went toward it and toward her, all of her, at once.

She paid and went out into the fog, and the fog closed over her without a sound, and I let her go.

I did not decide not to follow her. I simply did not follow her, and it did not occur to me that I should have wanted to, even though everything I felt for her was there. Nothing was missing. The coat was there, and whose coat it was, and everything I would have said to her. The wanting was missing. And a missing want leaves no space behind it, because the only thing that could have felt the space was the want.

I bought a baguette, still warm, and carried it out into the soft grey street like something alive. I do not remember the walk home.

I woke up because the bed was cold on one side.

For a few seconds I lay there with the bread still in my hand. Then there was no bread, and there had never been any bread, and the fog came back to me, and the newspaper, and the hands, and the coat. I had been dreaming. I had not known it.

Mia had been gone eleven days.

The notebook was on the floor by the bed, where I had left it, with the pen inside it holding the place. I did not sit up. Sitting up is the first thing you want to do and the first thing that ruins it, because the body moves and the mind moves with it, and whatever was there goes wherever it goes. So I stayed on my back, reached down without looking, and opened the notebook on my chest.

I wrote the fog first. I always write the weather first, before anything else, before I have decided what the dream was about. Everything else in a dream changes while you write it down. You remember a man and by the time the sentence is finished he has become a man who was watching you. The weather does not do this. The fog was fog before I had any opinion about it, and it stayed fog. Not moving. Not thinning. I wrote that it had been in front of the bakery and that I had only seen it once I stopped.

Then the newspaper. The bridge, the three columns, the man turning to the same page. I wrote that I had explained it to myself and that the explanation had satisfied me. That goes in the notebook, always: not just what was wrong, but what I told myself so that it would not be wrong anymore. The mind is astonishingly good at offering an explanation, and better still at convincing you that what you are looking at is ordinary. Most nights it takes me a few seconds to hear it, put it down, and look at the thing again. That is all the training is. That morning, in that bakery, in that fog, I had accepted what the mind offered me three times and questioned it no further.

Then the baker. The hands, the shelf, the loaf. I wrote that I had looked at her and stopped seeing the hands. I did not write why. I had already written why, in other notebooks, on other mornings, and it was always the same reason, and I had never found a better way to say it than that when you look at a person you see the person.

Then the coat.

I wrote the coat carefully. The grey, the lining, the thread in the pocket. I wrote that the woman wearing it had paid and gone out, and that I had let her. My hand was not steady on that line and I let it not be steady. There is no point writing a dream down if you are going to correct it into something you can bear.

Under the coat I drew the line I always draw, across the page, and under the line I made the list. The man with the newspaper: yes. The baker: yes. The woman in the coat: yes. The other customers, the ones in front of the man and behind me, the ones whose faces I could not now have described: no. I did not have to think about any of it. You never do. You know who mattered in a dream, and nobody I have ever worked with has been able to say why, and by now nobody asks.

Three important, the rest decor. That was a normal count for a short dream. It told me nothing on its own. It never does.

Then, at the bottom, where I keep the last line for what I actually did, I wrote that I had not realized.

I read it back. I had walked down Saint-Laurent in fog that did not move, watched a man read the same page three times, been served by a woman whose hands could reach the top shelf without her lifting her arms, and seen my wife’s coat walk out the door, and I had bought a baguette. I had not realized once. Every time I had found something to tell myself, and I had believed it, and I had gone on.

I closed the notebook and lay there with it on my chest.

It had been years. There is nothing to be proud of in the first discipline; it is the one you learn first, and the one that the people who train you assume you will never lose, as they assume you will not forget how to read. I had lost it, in the fog, in front of a bakery, and I had not lost it in years. I knew what that meant and I did not want to know what it meant, lying there, with the cold on my left side and the bread that had never been in my hand.

I got up at seven forty. I put the notebook on the desk, on top of the others, and went to the window.

There was fog on the street. It was ordinary fog, the kind the river sends up in September when the nights turn cold before the days do. It sat over Saint-Denis at the height of the second floor. A bus went through it with its lights on. A man in a yellow vest was doing something to the base of a streetlight across the road, and he did not look up at the bus, and the bus did not slow down for him. I could see the pharmacy sign, the bank sign, and the sign of the restaurant that had closed in March and never taken its sign down.

I noticed that the fog was in the dream and the fog was outside the window, and I decided that this was a coincidence. It had been foggy the night before. I had seen it when I closed the blinds. My mind had used what it had, as it always did. There was nothing in it.

I stood there for a while longer. Then I decided that there was something in it, not because I had new information, but because I had been doing this for a long time, and in a long time you learn which of your own explanations to trust. The one about the fog was not one of them.

I made coffee. I did not eat the bread, because there was no bread.

The phone rang at eight ten. It was Renaud. He works the same jobs I work, or worked, and he is the only person from La Veille who still calls me at home instead of leaving a message with the office and waiting.

“I’m calling about the Thursday thing,” he said.

“I know what you’re calling about.”

“They asked me to ask. I told them what you’d say. They asked me to ask anyway.”

“What is it?”

He told me. It was a retrieval, a company in Toronto that wanted to know what a company in Zurich was going to do before the company in Zurich had decided. Three days of reading, one night, one report. It was the kind of job they gave people they trusted, and it was the kind of job they had been giving me for nine years.

“No,” I said.

“I know.”

“Not until I find her.”

“I know, Adrien. I told them. I’m telling you I told them.”

There was a pause. I could hear his kitchen behind him, a kettle, the radio. He did not ask how I was. He had asked on the second day and the fourth day and I had answered both times, and after that he had stopped asking, which I was grateful for and had not told him.

“If you need someone to read for you,” he said, “I have the weekend.”

“I’ll call you.”

“You won’t. But I have the weekend.”

He hung up. I finished the coffee. Outside, the man in the yellow vest had finished with the streetlight and was walking to the next one, and the fog was where it had been.

There is a difference between recognizing that something was wrong and realizing that it is wrong, and the whole of what I do lives in that difference.

Recognizing is easy. I can do it now, awake, with coffee. Set out on paper, the things I saw in that bakery are obviously wrong, and anyone could say so. Recognizing is what you do afterward, with the answer in front of you. It costs nothing.

Realizing is what you do inside, without the answer, while the thing is still happening and while your own mind is offering you, one after another, perfectly good reasons to leave it alone. He lost his place. She is tall. It is fog; fog does not move. Every reason is true somewhere. What makes it hard is that the reason comes from me. I trust it as much as I trust anything else I think, and I have to set it aside anyway and look at the thing again.

That is the first discipline. Everything I was trained to do afterward, the navigating, the remembering, the reading of what the dream returns, depends on it, because none of it can begin until the moment you stop and say, out loud if you have to, that this is not right. I had not said it. I had bought bread.

I put the cup in the sink.

I would prime tonight. Properly, this time. For eleven days I had been priming as a tired man primes, with the problem stated badly and the faith half there, and the dreams had returned exactly what half faith returns, which is fog. So I would do it as I had done it for nine years for other people’s problems. I would sit at the desk and write the seed, everything I knew about her, until every line on the page pointed at the same woman, and then I would sleep, and the mind would build from it whatever it was going to build.

That evening I sat down at the desk and took a clean sheet and wrote her name at the top.

Then I sat. The name was there and nothing came after it, not because there was nothing, but because I knew what a seed written properly costs. It costs an afternoon at the least, and all of the attention a person has, and I did not have that in me that night. I had done it a thousand times for people I would never meet. I put the pen down. Tomorrow, or the day after, I would do it for her.

Permutations can arrive for free

Adam Kayal — Thu, 03 Sep 2026 12:01:35 GMT

1. The door

Imagine a door with a keypad. The keypad has three buttons: 1, 2, and 3.

The door opens when you press all three buttons in the correct order. You don’t know the correct order. And here is the annoying part: there is no enter key, no clear button, nothing. The door just watches the last three buttons you pressed and opens if they happen to be right.

There are six possible codes: 123, 132, 213, 231, 312, 321. So you could just type all six in a row.

123 132 213 231 312 321

That is eighteen presses, and it definitely works. But you can do better, because the door doesn’t know where one attempt ends and the next begins. It only ever looks at the last three. So codes can share presses.

Try this one:

1 2 3 1 2 1 3 2 1

Nine presses. Slide a window of three along it and watch what appears: 123, then 231, then 312, then 121 (not a code, that one is wasted), then 213, 132, 321. Every single one of the six codes shows up, and you used half as many presses. Some presses are doing double duty, ending one code and starting the next.

Nine is the best possible for three buttons. Nobody can do it in eight. That has been checked.

Now the obvious question. What about four buttons? Ten buttons? A hundred?

And that question, which a curious ten year old can ask in one sentence, is unsolved.

2. The pattern that looks like an answer

Let’s collect what we know for small keypads.

For a single button, the shortest sequence is obviously of length 1. For two buttons, the shortest is 3 (you can figure this one out by yourself). For three, we just saw that it is 9. For four buttons, 33, and for five, 153.

Stare at 1, 3, 9, 33, 153 for a moment. They look random. They are not.

1 = 1
3 = 1 + 2
9 = 1 + 2 + 6
33 = 1 + 2 + 6 + 24
153 = 1 + 2 + 6 + 24 + 120

Those numbers being added are 1, 2, 6, 24, 120: the number of ways to arrange 1, 2, 3, 4, 5 things. Mathematicians write them 1!, 2!, 3!, 4!, 5! and call them factorials. So the pattern is: add up all the factorials up to your number of buttons.

Better still, there is a recipe. If you have the best sequence for four buttons, there is a step-by-step method that turns it into a sequence for five buttons of exactly the right length. Then five into six, six into seven, forever. The pattern isn’t a coincidence you spotted. It’s a machine, and you can watch it work.

In 1993 two mathematicians, Dan Ashlock and Jenett Tillotson, wrote this up and conjectured that the recipe always gives the shortest possible answer. Other people found the same recipe independently over the years and believed the same thing. Why wouldn’t you? It matches every case anyone could check, and it comes with an explanation.

3. Something odd about those strings

Go back and look at the small answers again.

121

123121321

Read each one backwards.

1 backwards is 1. 121 backwards is 121. And 123121321 backwards is, character by character, 123121321.

They’re all palindromes. Like racecar, or level, or never odd or even. They read the same in both directions.

That is a genuinely lovely thing to notice, and it feels like it must be a clue. Maybe the shortest sequences are short because they’re perfectly symmetric. Maybe the symmetry is the trick.

Hold that thought. We’re going to come back and check it, and what happens when we check it is the most useful part of this whole article.

4. The one character that broke everything

In 2014, Robin Houston decided to attack the six-button case with a computer.

He noticed that this puzzle can be disguised as a completely different famous puzzle: the travelling salesman problem, where you have to visit a bunch of cities using the shortest possible route. There are good solvers for that. So he translated the keypad question into a travelling salesman question, handed it to a solver, and let it run.

On the 9,228th attempt, it came back with a sequence of length 872.

The recipe says 873 (1! + 2! + … + 6!).

One character. After thirty years, and only by one, but that is all it takes. The conjecture was wrong, and not just for six buttons: because the recipe builds each answer from the previous one, that single saved character means every case above six is beatable too.

Nobody had predicted this. Nobody had a reason for it. A computer found it, and the character it saved has never really been explained.

5. So where did the palindromes go

Now let’s check the symmetry idea from Section 3, because it makes a clean prediction: symmetric sequences should be the champion ones.

I ran the recipe and looked at what it produces.

For 1, 2, 3, 4, 5 and 6 buttons, the recipe outputs sequences of length 1, 3, 9, 33, 153 and 873. I checked all six (Yes, I have way too much time on my hands). Every one is a valid answer, and every one is a palindrome, including the 873.

That’s the first crack. The 873 is symmetric, and the 873 is not the champion. The palindrome shows up precisely in the case where the recipe loses.

It gets worse. For five buttons, there are eight different shortest sequences, all of length 153. I checked two of them. Both are perfectly valid, both are length 153, and one is a palindrome and one is not. So even at five buttons, where the recipe genuinely is unbeatable, being a champion doesn’t require being symmetric.

And Houston’s 872? I checked. Not a palindrome.

So the palindromes were never a fact about the answer. They were a fact about the recipe. The recipe happens to build symmetric things, so everything it makes comes out symmetric, whether or not it’s any good.

This is exactly the same mistake as the original conjecture, just smaller and easier to see. Somebody noticed a real, checkable, repeatable pattern, and the pattern was about the method rather than about the thing being sought. It is a good mistake to have made once, on purpose, before you make it by accident.

6. But there is a real symmetry hiding in here

Here’s the thing that survives.

Take any valid sequence and reverse it. You get another valid sequence, the same length.

Why? Reversing the whole sequence reverses every three-in-a-row window inside it. 123 becomes 321. 231 becomes 132. But 321 and 132 are codes too. The full set of codes is a mirror of itself, so a sequence that catches all of them, run backwards, still catches all of them.

That means every answer has a mirror twin. And palindromes are the special ones that are their own twin.

So the picture is: the recipe always parks itself exactly on the mirror line, and the improvement at six buttons is the first time anyone stepped off it. That is a familiar shape in this kind of problem. The most symmetric solution is the best one while the puzzle is small, and then, at some size, the space gets big enough that being lopsided pays.

Which turns the childish observation into a grown-up question: at what size does symmetry stop being free, and does how much you can save track how far off the mirror line you have to go?

That is a much better question than the one we started with, and we only got it by taking a wrong idea seriously enough to test it.

7. Two more numbers, and nobody knows why

Six buttons: recipe says 873, best known is 872. Saved one press.

Seven buttons: recipe says 5913. Greg Egan got it down to 5908. In February 2019, Bogdan Coanda found 5907 and announced it in the comment section of a YouTube video. Days later Egan and Houston tweaked his idea and reached 5906. Saved seven presses.

One, and then seven. That’s the entire dataset. No formula connects them. Both were found by computers grinding away, not by anybody working out why the savings should exist.

This is a strange situation to be in. Normally in mathematics someone has a theory and a computer confirms it. Here the computer sprinted ahead and left the theory behind, and the theory hasn’t caught up in over a decade. We have two answers and zero explanations.

Meanwhile the general question is fenced in but not pinned down. There is a proof that no sequence can ever be shorter than a certain formula, and a recipe that achieves a slightly bigger one. The two don’t meet. The gap between them grows as the keypad grows.

One thing worth saying plainly, because it often gets stated too confidently: even six buttons isn’t officially closed. The published proof only rules out anything below 867, not below 872. There’s an unpublished argument by Cole Fritsch that would finish the job, along with a search that found nothing at 871, but the write-up was left mid-revision. So: almost certainly 872, not formally proved.

8. Counting the waste instead of the presses

Here is a way to think about it that I find much clearer, and it needs no algebra.

Every press you make, after the first couple, reveals a new “last three buttons.” Some of those are real codes. Some are junk, like the 121 we hit earlier. Junk presses are pure waste.

So any sequence, no matter how clever, breaks into exactly three parts:

total presses = codes you must contain + a short run-up to get started + wasted presses

The first two are fixed. You cannot contain fewer codes than there are codes, and you cannot avoid the run-up. So the entire puzzle is: how few presses can you waste?

Now the numbers say something concrete. At six buttons, the recipe wastes 148 presses. Houston’s 872 wastes 147. That’s the whole difference. At seven buttons the recipe wastes 867 and the best known wastes 860. The published proof says you can’t get below 142 waste at six buttons, or below 838 at seven.

This framing also kills a tempting wrong idea, and I want to flag it because I believed it for a while myself. It’s natural to guess that the savings come from some code sneaking in twice, appearing for free as a lucky byproduct. So I checked both the 873 and the 872 (again, too much time on my hands). In each of them, every single one of the 720 codes appears exactly once. No duplicates anywhere, in either. The strings differ only in that one wastes 148 presses and the other wastes 147.

The free lunch isn’t a repeated code. It’s one fewer wasted step. Small difference in words, big difference in what you go looking for.

9. The merry-go-round

To go further we need one structural fact, and it’s the sort you can verify on paper in a minute.

Take a code, say 1234. Move the first button to the back: 2341. Do it again: 3412. Again: 4123. Again: back to 1234. Four spins and you’re home.

Call that a merry-go-round. Every code sits on exactly one, they never overlap, and each has exactly as many codes on it as you have buttons. For four buttons there are six merry-go-rounds of four codes each. For n buttons there are (n−1)! of them, and the reason is a one-line argument that any curious teenager can follow with a little help.

Here’s why they matter. Spinning is the cheapest possible move. Going from 1234 to 2341 costs a single press, and it wastes nothing at all, because the new window is immediately another real code. Riding a merry-go-round is free.

Every other move costs you. Getting from one merry-go-round to a different one means pressing buttons that produce junk windows on the way.

So the recipe, seen properly, is just: hop on a merry-go-round, ride it all the way around for free, pay to jump to the next one, repeat until you’ve ridden them all. Add that up and the sum-of-factorials formula falls out. The recipe was never a mysterious pattern. It’s the cost of the obvious tour, and every wasted press in it happens during a jump.

10. Getting off early

Now the mechanism behind the savings has a name you can picture.

The obvious tour rides each merry-go-round all the way around before jumping off. But nothing forces that. You could hop off halfway, go do other things, and swing back later to pick up the codes you skipped.

Usually that’s a bad deal, because coming back means paying for another jump. But every so often your route was going to pass right by those leftovers anyway, on a jump you already needed. Then the pickup is free, and you’ve shaved off waste that the obvious tour would have paid.

That’s it. That’s where the missing character at six buttons comes from, and presumably the seven at seven buttons.

I should be clear this is not my idea. Zach Hunter and Cole Fritsch have both worked on formalizing exactly this, calling them “early exits” and “exitless paths,” trying to count how many such early hops a route can support. Their work lives in a public Google Group and in unpublished drafts rather than in journals, which is probably why it’s less famous than the record-breaking sequences. Anyone starting on this should read them first.

11. Why nobody has finished it

Here is the wall, and it’s worth understanding because it’s the real reason this is still open.

Whether an early exit is free is not a local question. It doesn’t depend on the merry-go-round you hopped off, or the codes you skipped. It depends on whether some other part of your route, somewhere completely different, happens to swing past them later. Which depends on where the other merry-go-rounds got placed. Which depends on their early exits. Which depends on a third thing.

Everything depends on everything. That’s why 1 and 7 don’t sit in any obvious pattern: the number isn’t built from independent pieces you can count separately and add up. It emerges from the whole arrangement at once. And “count the pieces and add them up” is the main tool mathematics has for questions like this.

Fritsch’s draft handles a single early exit reasonably well and gets stuck exactly where several of them start interacting, where two merry-go-rounds end up depending on each other. In the group discussion he describes the machinery as holding together but needing more work, and there it was left.

That distinction matters. Nobody proved that case is impossible. Somebody ran out of time. Those are very different, and the second one is an open invitation.

12. What you could actually go and do

This problem has a strange history. The best known limit on how short sequences can possibly get was worked out by an anonymous person on 4chan, on a message board, in a thread about what order to watch a TV show in. A record-breaking construction was announced in a YouTube comment. The serious discussion happens in a public Google Group anyone can read. There is very little gatekeeping here, and the entry price is a laptop and some stubbornness.

Three things that are genuinely open and genuinely small:

Explain the 872. The sequence is published. Anyone can download it and check it in a few lines of code. Work out which merry-go-round got left early, and why the pickup cost nothing. Not “one press was saved,” but “it hopped off here, and this later stretch grabbed the leftovers on a jump it needed anyway.” Right now that explanation doesn’t exist in writing, and everything else depends on it.

Then do the same for the 5906, and ask whether the seven savings share a reason. This is the big one. If all seven come from one repeating pattern, there’s a mechanism to generalize and the problem might crack. If they’re seven unrelated flukes, then “waste” isn’t one phenomenon and half of this article needs rewriting. Both strings are public. This is answerable by hand, by anyone, this week.

Try to count the early exits directly. How many early hops can coexist without tripping over each other? That smells like a scheduling problem, the kind where you fit as many overlapping bookings into a calendar as possible. Those have standard tools. If a scheduling argument predicts 147 waste at six buttons and 860 at seven, it’s probably the right way to think and worth pushing to every n. If it predicts the wrong numbers, the size of the error tells you exactly which rule you forgot, which is more useful than being vaguely right.

And one bonus, from Section 6. Since the recipe always sits on the mirror line and both improvements stepped off it, here’s a guess worth testing: maybe no palindrome is ever the champion, once you have six or more buttons. If that’s true it’s a free gift to anyone running a search, because it lets you throw away half the possibilities before you start. I don’t know if it’s true. I haven’t seen anyone check.

13. The moral

Twice in this article a beautiful pattern turned out to be a fact about the method rather than about the answer.

The palindromes were real. Every sequence the recipe produces is symmetric. But that was the recipe’s fingerprint, not a property of good sequences, and the moment somebody found a better sequence it was lopsided.

And the sum of factorials was real too. It’s exactly right, it’s provable, it comes with a construction. It’s just that it is the exact cost of the obvious tour, and for thirty years everyone assumed the obvious tour was the best one.

Both times, the mistake wasn’t sloppiness. It was answering a slightly different question extremely well and not noticing the substitution.

The question was never really “how short can the sequence be.” It’s “how little of it can be wasted.” Somewhere out there is a rule that says exactly how much you can get for free, and as far as anyone knows it’s still sitting there, unwritten, waiting for whoever gets curious enough about a door with a broken keypad.

You Don’t Get What You Want, You Get What You Are

Adam Kayal — Tue, 01 Sep 2026 11:31:37 GMT

Missed the first part? Start here:

A goal is a query you enter once. You set it, the search runs, it resolves, and you move on.

Identity is a query that never stops running. It doesn’t resolve, because it isn’t a question about the world, it’s a standing assumption about what kind of results are available to someone like you. Every situation gets filtered through it before you’re aware of the situation. That’s why a goal can direct your attention for an afternoon while identity directs it for a decade.

I found this out in the gym, over years I’d rather have back.

The stall

For a long stretch I made no meaningful progress in the gym. Not because I was doing something wrong on paper. I knew about progressive overload. I knew about volume, about recovery, about eating enough protein. If you had tested me I would have passed.

I trained anyway, consistently enough to feel like I was trying, and I stayed roughly the same.

The explanation I’d have given at the time was genetics, or that I had already gotten what there was to get.

What I actually believed

The real explanation was something I never said out loud, and never had to, because I’d never examined it: I didn’t think the physique I wanted was achievable without drugs.

I’d never articulated it that way. It sat underneath the training as an assumption, and assumptions don’t announce themselves. Everything I saw seemed to confirm it. The people who looked how I wanted to look were, as far as I could tell, on something. So the results I was after belonged to a category I wasn’t in.

Notice that this is not a belief about hypertrophy. It’s a belief about me. That distinction turns out to matter more than anything else in this piece.

The loop

Here’s what a belief like that does.

If the result isn’t really available to you, then full commitment is irrational. So you don’t fully commit, and not from laziness. You commit at the level the belief warrants, which is most of the way, and the last part gets withheld quietly enough that you don’t notice you’re withholding it.

Then you don’t grow. And not growing is exactly what the belief predicted.

So the belief gets confirmed by an outcome that the belief itself produced. It generated its own evidence. This is why beliefs about yourself are so stable, and why arguing with them doesn’t work. They aren’t floating free waiting to be corrected. They’re being fed, every day, by the results they cause.

What broke it

What broke it wasn’t an argument. It was seeing two people.

I came across Alex Leonidas and GVS online, the first genuinely muscular naturals I’d encountered who I found credible. Nothing about my training knowledge changed that day. I learned no new technique. All that changed was that the result moved out of the category of things unavailable to me.

I didn’t decide to believe it. I saw evidence and the belief updated on its own, which is how most real belief change works.

What followed

Some of what changed afterward was visible and boring. I ran a program properly instead of drifting between them. I logged. I stopped treating sessions as optional when the week got busy. I ate like someone who expected the food to turn into something.

But the change that mattered doesn’t appear in a training log.

I started training with an intensity I had never trained with before. Same exercise, same weight, same body. The difference was where the set ended. Before, stopping early felt like good judgment. There was always a reason, and the reason was always sound: form was breaking down, the joint felt off, tomorrow’s session mattered more. After, the same moment felt like quitting.

Nothing physical had changed. What changed was what I thought the last three reps were buying me. When you don’t believe the outcome is available, hard effort is a cost with no return, and your body is very good at finding respectable ways not to pay it. When you do believe, the same effort is an investment, and you find you can pay it.

This is why willpower is a poor explanation for anything. You cannot decide to take a set closer to failure. Intensity isn’t summoned by the conscious mind. It arrives, or fails to, depending on what sits underneath it.

The chain

What happened there follows a sequence, and it’s worth walking through slowly, because each step is doing work.

Belief sets the index. This is the mechanism from the first piece. What you believe determines what gets surfaced from underneath, because it determines what the search is running against. Two people facing an identical setback retrieve completely different sets of options, not because one thinks harder, but because they’re querying against different assumptions about who they are.

Thoughts produce emotions. Feeling follows interpretation rather than events. The failed set means “I’m not built for this” or it means “that’s the edge, I found it,” and those two readings produce entirely different states in the same body on the same day.

Emotions produce actions. This is where the first piece’s asymmetry pays off. The conscious mind is weak. It cannot force sustained behavior against your emotional state, which is why knowing what to do is so reliably insufficient. What moves you is how you feel about the thing, and how you feel about it came from what you believe about yourself.

Actions produce reality. One session changes nothing. The point isn’t the single action, it’s that a stable belief produces the same action repeatedly, without negotiation, for years. That compounding is what shows up in the mirror.

And then reality feeds back into belief, which is where the loop closes. This is the same structure that kept me stuck, running in the other direction.

Identity is the load-bearing kind

Most of what you believe is inert. Your opinion about interest rates or a restaurant isn’t shaping anything.

Beliefs about yourself behave differently. Your subconscious treats them as constraints rather than as data, and it enforces consistency against them. Behavior that contradicts an identity gets corrected back, usually without your involvement and usually with a reasonable explanation attached.

An identity is a group of beliefs about who you are, held together and operating as a unit. And a goal held against your identity is a one-off query fighting a standing one. The standing one wins, because it was running before you set the goal and it will still be running after you’ve forgotten it.

Which means the sentence people repeat about wanting something badly enough is backwards. You don’t get what you want. You get what you are.

The obvious question is how you change what you are, given that you can’t argue yourself into it and I only changed by accident, by seeing two strangers online. That has an answer, and it’s the subject of the next piece.

There Is No Index for "A"

Adam Kayal — Wed, 26 Aug 2026 02:39:50 GMT

Recognizing something is far easier than recalling it. I noticed how far apart the two are while on a date.

We were playing a game: name every country that starts with a given letter. We took A first.

Argentina. Australia. Austria. Afghanistan. Algeria. Then we stalled. Between the two of us we managed maybe eight countries, and the well went dry.

Then we looked up the full list. Afghanistan, Albania, Algeria, Andorra, Angola, Antigua and Barbuda, Argentina, Armenia, Australia, Austria, Azerbaijan. Every one of them obvious the moment we saw it. Not a single country on that list was unknown to either of us… we had simply been unable to get to them.

What surprised me wasn’t the failure. It was how effortless recognition was compared to recall, and how wide the gap between the two turned out to be. We kept going through other letters, and it was the same every time: six or seven produced with real effort, then the full list read back with instant recognition of all of them.

We knew all of them. So why could we only produce eight at best?

The conscious mind points, it doesn’t produce

Nothing failed in your memory during that exercise. The countries were all there. What failed was access.

Your conscious mind did not generate those names. It couldn’t. It can only direct attention toward them and wait for something to come back. That is, more or less, the only thing it does. Every idea it appears to produce, including the ones that feel like your own deliberate thinking, arrives from underneath, already formed.

A man holding a gun cannot produce the force of the bullet. He has no capacity to generate that kind of energy himself. What he can do is aim. The entire contribution of the conscious mind is direction, and direction alone.

Or, in more current terms: the conscious mind is the prompt, and the subconscious mind is the model. The prompt contains no knowledge. It holds no data, performs no computation, and stores nothing. It is a short string of text pointed at something enormous. All the capability sits on the other side.

The asymmetry is easy to underrate. Everything you have ever read, heard, watched, or lived through is in there, including material you have no memory of encountering. What reaches conscious awareness is a filtered fraction of it, selected for relevance and discarded the moment it stops being useful.

Stuck because of the effort

When I studied mathematics and physics, I would sometimes hit a proof I had no way into. Not a hard one; an impossible one. I’d sit with it for countless hours and produce nothing but the same three failed approaches, each one a little more elaborate than the last. Eventually I’d give up and go to sleep.

And I would wake up the next morning with the solution already in my head. Not a hint of it. The whole thing, clean, obvious, sitting there before I’d fully opened my eyes.

The French have a phrase for this: la nuit porte conseil. The night brings counsel. It’s usually said about decisions, but it describes something more literal than that.

Because nothing arrived overnight. No new information reached me while I slept. Every element of that solution was already in my possession at 10pm, when I was staring at the page getting nowhere. What changed was that I stopped looking.

I wasn’t stuck despite the effort. I was stuck because of it.

Watch what happens when you’re stuck on the letter A. You say Argentina, Australia, Austria, and then, without deciding to, you say them again. Argentina, Australia, Austria. And again. You’re not searching anymore; you’re circling the same three names, each pass a little faster and a little more useless than the last. Conscious recall reaches for the nearest association, which by then is whatever you just retrieved. So the third attempt is closer to the second than the second was to the first. The path wears deeper every time you walk it. And Andorra is sitting right there, one room over, unreachable. Not because it’s buried, but because you keep entering through the same door, and that door doesn’t open onto Andorra.

Frustration tightens the loop. On that date there was mild pressure to produce something, and the pressure made it worse, not better. Effort narrows attention toward the familiar and the defensive, which is precisely the wrong direction, because the name you’re reaching for is a faint signal and the three you already have are loud. You end up drowning out the thing you’re looking for with the thing you’ve already found.

Setting the goal starts the machine

And notice what you do next, once the loop breaks. You start going continent by continent: Africa, Algeria, Angola, good, that works. Asia, Afghanistan, Armenia. Europe, Austria, Albania, Andorra. Or you go by language, by cuisine, by places you’ve traveled, by news you half remember.

What you’re doing there is reaching for an index. An index is a way of arranging things so you can find them without checking everything. A phone book is sorted by surname, so looking up Tremblay takes seconds. Ask that same phone book who lives on Rue Saint-Denis and it’s useless, even though the answer is somewhere on its pages, because it was never arranged that way.

Every one of those routes is an index built for a different question than the one you were asked. You’re rummaging through the filing system you happen to have, hoping the answer falls out sideways. That is what makes retrieval expensive: the data is all there, but nothing in it is organized by first letter, so you have to reach it through structures built for something else. Recognition skips the problem entirely. Shown Andorra, you don’t search anything. You just know.

Eventually, you go to sleep, the pressure comes off, and the search continues without you.

It doesn’t run the way you were running it. It stops chasing one thread and starts replaying fragments in parallel, including the ones that seemed irrelevant while you were awake, which you had discarded precisely because they didn’t fit the framing you were stuck in. And with no bias toward what you tried last, it connects a piece of the problem to a piece of something unrelated: a technique from a different chapter, a shape you saw in another context. The connection you need is usually with something unrelated: a technique from a different chapter, a result you saw months earlier in another context. That kind of link is exactly what you can’t make deliberately, because deliberate search keeps you inside the same neighbourhood of ideas.

This is also why answers show up in the shower, or on a hike, or in the middle of a tennis rally. Most people notice the pattern and treat it as coincidence, or as something vaguely to do with being relaxed. But those situations all have the same thing in common: your conscious mind is busy with something trivial and has stopped giving orders. The rest of your mind is free to work.

But note what had to happen first. None of it starts on its own. I had to sit with that proof for hours and fail: that was the query being entered. Setting the goal is what sets the sequence in motion: indexation, then retrieval, then solution. The goal is the instruction that tells the system what to search for. Then the search runs in the background without you having to watch it.

Retrieval is harder than recognition

There’s a reason the sequence has to start with indexation.

Your memory is not filed alphabetically. Nothing in it is. Those countries are stored by continent, by trips you took, by news you half-read, by food you ate, by people you met. There is no index for starts with A, because the category is meaningless to how the data was stored in the first place.

So when you ask for it, the index doesn’t exist yet. It has to be built. That’s what the goal actually does: it constructs an index along a dimension your memory was never organized by, and only then does retrieval become possible.

Recognition and retrieval aren't the same task at different difficulty levels. They're different tasks. Recognition is cheap. Shown Azerbaijan, you verify it in an instant. Retrieval is expensive, because it requires searching a space that hasn’t been sorted for the question you’re asking. It’s the same shape as P versus NP in computer science: checking an answer is easy, finding it is not.

So a goal does real work. It tells your mind what to index for, and until you set one, nothing gets indexed. What you know stays where it is, out of reach, because you never asked in a form your memory could answer.

Everything that follows depends on this. If direction is the only lever the conscious mind has, then the question becomes what determines the direction, and that turns out to be identity.