The Floor and the Ceiling

Saul Leiter spent sixty years in a single East Village apartment, painting in the morning and photographing in the afternoon, and for most of that time almost no one knew he was doing either. There were boxes of color slides in his rooms, thousands of them, photographs he had made on the corner near his building or through the rain on his own window, kept and not shown, made and not sent out into the world. He had been briefly known for fashion work in Harper's Bazaar in the 1950s, and he had let that fall away. He stayed in the neighborhood. He kept painting. He kept making the photographs he was not pushing anywhere. The first serious book of his color work was published when he was eighty-three.

The chronology is startling only until you understand what it means. Leiter was not a man who failed to seek attention. He was a man for whom attention would have been the wrong direction, the deformation of the thing he was actually doing, which was looking, slowly, every afternoon, at the way light fell through fogged glass and the way a yellow taxi moved through rain and the way the city revealed itself to a person who refused to chase it. The camera, for him, was an instrument of patience more than assertion. Most of the work was the work of staying. The frames he made and did not show were not failures of ambition; they were the practice itself, kept private because the practice was private, kept because the keeping was the point.

I have been thinking about Leiter this year because I have been watching something familiar happen to a craft that is not mine, or not only mine. A tool has arrived that appears to make the craft trivial. Code, which used to require years of careful accumulation, now answers to a sentence in English. The junior developer, the designer who always wanted to ship a small thing, the founder with an idea and no engineer: all of them can now produce, within an afternoon, what would once have taken a team a quarter. The flood has begun. The accusations have begun, too: the craft is over, the profession is ending, everyone will build. The shape of the noise is recognizable. I have heard it before.

When the camera became cheap, first with the Kodak, then with 35mm, then with the phone that is not yet even called a camera, the photographers panicked. Mass photography was going to dilute the art beyond recognition. Anyone could now make an image. The serious photograph, the one made with intention, would be lost in a tide of snapshots. And something like this did happen, if you count the tide. Billions of images are now made every day, most of them never looked at by a second person, most of them not even looked at by the first. The floor of photography, which for most of its history had been the price of equipment and the time required to learn it, dropped away.

What did not happen was the disappearance of photographers. The people who had been photographing for reasons the tool could not touch, for the long patience of seeing, for the private discipline of staying with one corner, one window, one quality of afternoon light, for whatever it is inside a person that makes them willing to walk the same three blocks for forty years, those people were still there. They were, in fact, more visible than before. Against the flood, the photograph made with attention now stood at a certain distance from the rest, the way a single sustained note stands against static. The ceiling had not moved; only the floor had fallen. And the people who had never needed the floor remained where they had always been, doing the difficult thing for its own sake, their craft made legible, perhaps for the first time, as a matter of character.

I want to be careful here. I am not saying the tool did not change photography; it changed it enormously. The commercial middle was hollowed out. The photojournalist, the wedding specialist, the studio portraitist: many of those livelihoods compressed or vanished. What survived, at the level of the art, was not the profession but the disposition. The willingness to look at something for longer than it asked to be looked at. The willingness to not take the easy frame. The willingness, often, to take no picture at all, or to take it and keep it in a box for fifty years.

The camera could not capture any of this. The camera could only capture what the photographer had already decided, silently and over years, was worth capturing. A tool cannot confer the instinct that tells you the moment is not the moment. A tool cannot produce the thousand small refusals that are the quiet architecture of a body of work.

It is two in the morning and a person I know is rewriting, by hand, something a model produced for him in eight seconds. The code works. The tests pass. He is rewriting it anyway, because the architecture is wrong, or will be wrong in six months, when a thing he has not yet named will need to live inside it. He cannot explain, in a sentence a manager would accept, why he is doing this. He is doing it because he can feel the shape of the system the way Leiter could feel a New York afternoon through fogged glass, and he knows that the generated code, however correct, is a frame taken for the wrong reasons, from the wrong angle, at the wrong fraction of a second. He is refusing it. He is doing the work the tool cannot do, because the tool, as always, has produced the artifact and not the attention that would have produced the artifact if the artifact had been worth it.

He is the developer the flood is going to reveal.

There is a tendency, when a tool arrives that lowers the threshold to an art, to believe the art is over. The tendency is almost always wrong, and wrong in the same way. The tool is visible; the art was never only visible. The tool touches the output; the craft was never only the output. What the camera did, over a century, was not to kill photography but to strip it down to the people for whom photography was a way of being in the world. What the AI coding assistant is doing, now, at speed, is the same thing to development: stripping the work down to the ones who were never in it for the threshold, who were in it for the problem, for the architecture, for the slow accumulation of taste that tells you, without a rubric, that this dependency is wrong and that refactor can wait and this system, if it is to last, must remain smaller than it wants to be.

Everyone will be a developer, now, in the sense that everyone has become a photographer: the way of producing the artifact has been handed out. The photographers, though, are still on their corner, still at their window, still painting in the morning and walking the same blocks in the afternoon, still keeping the slides in boxes because the keeping is the point. The developers will be in the same place, doing the same thing, under different light. The tool has not made them obsolete. It has made them, at last, clearly themselves.

Perhaps this is what tools have always done. Perhaps mastery was always character, and the threshold (the price of the camera, the years of learning to compile) was what let us mistake the craft for the threshold. When the threshold is gone, what is left is what was there beneath it: a person willing to keep doing the difficult thing, because the difficult thing is what they are for. The frames they take are in the world, and we see them. The frames they do not take, or take and do not show, are not in the world, and we do not see them. But the not-taking, the not-showing, the staying with the work past the point where the world has stopped asking, is the craft. It always was.