The Beige

He opened the app at the coffee shop on Linn Street and I looked over his shoulder, not on purpose, just the way you look at a screen that catches the corner of your eye. White background. Soft gray icons. Rounded rectangles.

A single accent color, a kind of muted teal, doing the work of a logo. I could not tell you what the app was. I still cannot. He scrolled, tapped, scrolled.

The interface looked like the interface of the app I had opened that morning to check the weather, which looked like the interface of the app I had opened the night before to refill a prescription, which looked like the interface of the bank, which looked like the interface of the parking meter on Washington Street. Same font. Same gentle curvature. Same beige courtesy.

I walked the river loop afterward and thought about it. The leaves were down. A woman in a Carhartt jacket was throwing a tennis ball into the water for a dog who would not retrieve it. Across the river a backhoe sat idle next to a pile of broken concrete from the old footbridge.

None of these things looked like each other. The dog did not look like the backhoe. The Carhartt did not look like the leaves. The screens look like each other.

This is the observation. I want to be careful with it because it has been made before, mostly by designers complaining about other designers, and the complaint usually ends in a portfolio site that itself looks like every other portfolio site. The thing is, the complaint is not really about design. It is about what the design is for, and what it does to the person holding the phone, and what it implies about the company on the other end of the rectangle, and what it forecloses.

The beige is doing work. The work is the subject. Two summers ago I went into a Verizon store to replace a phone and the woman behind the counter, who could not have been more than twenty-three, walked me through the setup. She had done it five hundred times.

She tapped the screen the way a person taps a screen they have given up on, with the index finger held almost flat, a kind of dismissive prod. At one point she said, without looking up, “they all do the same thing now.” She meant the phones. But she could have meant anything on them. I asked her what she meant and she shrugged and said, “you know.

They all do the same thing.” That was the entire exchange. I have thought about it more than the exchange deserved. What I keep coming back to is that the convergence is not accidental and it is not, exactly, a conspiracy. It is the residue of a thousand individual decisions made by people who were not in the room with each other but who were reading the same blogs, hiring from the same five schools, copying from the same two or three reference companies, and answering to investors who wanted to see, on a quarterly call, that the product looked “mature.” Mature, in this context, means it looks like the other ones.

It means a customer can pick it up without instruction. It means the friction has been sanded off. The sanding is the product. I do not want to write the easy version of this essay.

The easy version is that everything used to be interesting and now it is boring, that MySpace pages had character and Instagram does not, that the early web was a garden and the current web is a strip mall. The easy version is partly true and entirely useless because it cannot say why. It treats the visual flattening as a matter of taste, when taste is downstream of something else. Something is making the decisions.

Something wants the screens to look like this. Here is one piece of it. The dominant interface pattern of the last decade is the feed. The feed wants you to keep scrolling.

To keep scrolling you must not be interrupted by the interface itself. Anything visually assertive — a strong color, a hard edge, an unusual typeface — becomes, in feed logic, an obstacle. It pulls the eye off the content and onto the chrome. So the chrome retreats.

It becomes white, gray, soft. The accent color shrinks to a small dot on a notification. The typeface becomes a neutral sans-serif designed by a committee in San Francisco or Berlin. The feed is the thing.

The feed must not be disturbed by the frame around it. This is, by itself, a reasonable design choice. A picture frame in a gallery is also unobtrusive on purpose. The problem is what happens when every app becomes a feed, or aspires to be one, or imitates the visual language of one even when it is not one.

My bank is not a feed. There is no reason for my bank to look like Instagram. But my bank looks like Instagram because the people who designed my bank’s app trained at the same places as the people who designed Instagram and because the bank’s executives, when shown a mockup, recognized the Instagram visual language as “modern” and approved it. The beige spreads not because every product needs it but because every product manager wants to be seen as having the thing other product managers have.

On the river loop there is a bench near the Park Road bridge where I sometimes stop. From the bench you can see the back of the water plant and a row of cottonwoods and, in the morning, a great blue heron that stands in the shallows for forty minutes at a time. The heron does not move. It is waiting for a fish.

The fish does not know the heron is there. I have watched the heron strike maybe four times in three years. Each time the strike was so fast I could not see it as a motion. There was a heron with its neck folded back and then there was a heron with a fish in its beak.

The intermediate moment was not available to me. I bring this up because I want to say something about what an interface used to do, before the beige, and I cannot quite say it abstractly. An interface used to be a place. A place has a back of the water plant and a row of cottonwoods.

A place has a heron in it sometimes. A place implies that someone chose it, that someone decided this was the place and not some other place. The early web was full of places like this. Geocities pages were places.

Forums were places. The blogs of the early 2000s were places. You knew when you had arrived somewhere and you knew when you had left. The current interface is not a place.

It is a transparent pane through which you look at content. The content is the heron and the cottonwood and the water plant, supposedly. But the content is also flattened, because the content is also designed to be feed-compatible, which means it is also designed not to assert itself too strongly against the surrounding stream. The post in your Instagram feed is calibrated to look good next to the other posts in your Instagram feed.

The article on a news site is laid out to look like the other articles on the other news sites. The beige is not only on the chrome. It has seeped into the content. No one decided this.

That is the strange part. If you asked the designers individually whether they wanted the entire visible internet to look like a single product, they would say no. Most of them are, I think, sincere people who like color and texture and difference and who own books about Saul Bass and Paul Rand and the Bauhaus. They have opinions about kerning.

They have, in their own homes, walls painted unusual colors. The flattening is not a function of their preferences. It is a function of the system they work inside, which rewards certain choices and punishes others, and which the individual designer cannot meaningfully resist without leaving the field. A friend of mine who used to work at a large software company in Seattle described the process to me once.

He said the team would mock up a new feature in three or four visual directions. One would be conservative — close to what existed. One would be slightly more adventurous. One would be quite different.

The mockups would go through review, and at each stage of review, the more adventurous options would be eliminated, not by any single person’s decision but by the accumulated weight of objections. Someone would say the bold version “tested poorly with older users.” Someone would say the unusual typeface had “accessibility concerns.” Someone would say the unfamiliar layout would require “user education.” None of these objections were wrong, exactly. Each was a real consideration. But at the end of the process, the version that survived was always the version closest to the existing industry standard.

The system did not select for difference. It selected against it, at every stage, with a thousand small filters. He said it was like watching a river deposit silt. Each particle small.

The bank, over time, smooth. I have been trying to think of what we lose. This is the harder part of the essay because the losses are mostly invisible. You cannot point at the thing that did not get made.

You cannot say, with any confidence, that if interfaces had remained as varied as they were in 2005, we would now have richer interior lives or more interesting opinions or stronger friendships. Probably we would not. The interface is not that consequential. People had interesting opinions before the internet and will have interesting opinions after it.

But something is being trained, slowly, by the visual environment we spend our hours inside. I do not know what to call it. A kind of expectation. A sense that experiences should arrive without drag, that the surface of a thing should not impose itself, that the proper relationship between a person and a screen is one of unimpeded consumption.

The beige teaches this. The beige is the curriculum. Two doors down from my apartment there is a hardware store that has been at the same location since 1947. The sign is hand-painted.

The interior is dim and the floors are wood and the shelves are crowded with things I cannot identify. The first time I went in I asked for a particular kind of washer and the man behind the counter, without saying anything, walked to a drawer in the back, opened it, pulled out a small paper envelope of washers, and set it on the counter. He did not ask me to confirm the size. He did not ask me what the washers were for.

He had decided. The washers were the right washers and they were eighty cents. This is not a sentimental observation. I am not saying the hardware store is better than the app.

The hardware store would not work for everything. You cannot order a hardware store at 11 p.m. on a Tuesday from your living room. You cannot ask the hardware store to remember your address.

The app does things the hardware store cannot do. But the hardware store imposes itself. It has a smell. It has a particular man behind the counter who has opinions about washers.

It has, in its physical existence, evidence that someone chose to make it this way and not some other way. The choosing is visible. The choosing is part of the experience of being inside it. The app does not impose itself.

The app is designed not to impose itself. The app aspires to a kind of invisibility, a transparency without resistance, in which the user encounters only their own task and the system’s response. There is no man behind the counter. There is, if you look hard enough, a customer service chat staffed by a person in another country reading from a script written by a manager in a third country, and behind that person there is a software layer designed by a team in a fourth country, and behind that team there is a venture capital firm in Menlo Park that has never seen the product and does not care to.

The choosing has been distributed across so many people in so many places that no one feels responsible for it. The app is no one’s. It is the average of a million decisions made by people who do not know each other. The aesthetic of the app expresses this.

The beige is the visual signature of distributed, deniable authorship. It is what something looks like when no one will admit to having made it. I have been calling this algorithmic loneliness, in my own head, for a few months now. The term is imperfect and I should explain it.

By loneliness I do not mean the loneliness of having no friends. I mean a more specific condition: the experience of moving through an environment that has been calibrated for you by something that does not know you, and that does not know it does not know you. The algorithm has decided, on the basis of your past behavior and the past behavior of people it thinks resemble you, that you would like this particular thing in this particular order. It is often correct.

It is often correct in a way that feels uncomfortably close to actual knowledge. But it is not knowledge. It is statistics applied to your shadow. The thing it knows is not you.

It is a representation of you derived from the residue of your clicks. This is lonely in a specific way. It is lonely the way being among a large crowd of strangers is lonely, when each stranger nods at you with a polite familiarity that does not extend past the nod. The system is friendly.

The system anticipates. The system remembers what you bought last time. But the system is not your friend, and the friendliness, repeated across every interaction with every app, accumulates into a kind of low-grade unease. You are being treated kindly by something that does not exist in a moral sense.

The kindness is a product feature. It has been designed and tested and A/B-rolled out. The beige expresses this too. The beige is the visual equivalent of the polite nod from the stranger in the crowd.

It is friendliness without commitment. It is the absence of any particular face. To put it less politely, the design language of contemporary consumer software is the design language of a system that does not want to be held responsible for anything. It does not want to assert.

It does not want to be remembered. It wants to be felt as a kind of ambient courtesy, present everywhere, locatable nowhere. When something goes wrong — when the charge appears on your card that you did not authorize, when the feed shows you something disturbing, when the chatbot misroutes you for the fourth time — there is no face to address the complaint to. The interface has been carefully designed not to have a face.

The facelessness is, in the legal sense, an asset. It limits liability. It enables scale. It permits the company to grow to a hundred million users without ever having to feel like a hundred-million-user company.

The part that catches me is that this is not what design has historically been. Design, in the older sense, was a way of saying who made the thing and what they cared about. A chair by George Nakamura — I mean Nakashima, George Nakashima, the woodworker — looks like a chair by George Nakashima. The wood is selected for its figure.

The joinery is visible. You know, looking at it, that someone stood in a shop in Pennsylvania and decided to leave the live edge intact. The chair has a face. The face is Nakashima’s face, or his shop’s face, or the face of the particular tree the wood came from.

Design announced authorship. The current design language has inverted this. It does not announce authorship. It conceals it.

The goal is not to make the work of the designers visible but to make it disappear, so that the user feels they are interacting with something natural, inevitable, almost geological. The app simply is. The app is not anyone’s idea. The app, in its calm beige courtesy, presents itself as the way things are.

I am not sure I have made the case yet for why this matters. Let me try. In 2009 I was in a steel fabricator’s shop outside Rockford reporting a story about the closing of three nearby suppliers. The owner of the shop, a man named Ed, took me on a walk through the floor.

He showed me a press brake he had bought in 1978 and which he was still using. He showed me a small CNC mill he had bought in 2004 and which had, he said, paid for itself in eight months. He showed me a section of the floor that was empty, where two machines had been, and which he now used to store pallets. He did not get sentimental about the empty section.

He just walked me past it. When we sat down in his office afterward, he said something I have not been able to forget. He said the trouble with the new buyers, the ones who placed orders through the online portals the big OEMs had started using, was that they had no idea what they were asking for. They would submit a drawing with tolerances specified to four decimal places when two would have been fine, and they would not understand why this drove the price up, and when he tried to explain it on the phone they would not really listen, because for them the part was just a line item in a spreadsheet.

The drawing came out of a CAD program. The order went into a portal. The portal looked like every other portal. The buyer had never been in a shop.

The buyer did not know what a press brake sounded like. The buyer did not know that a four-decimal-place tolerance required a different machine than a two-decimal-place tolerance and that the different machine cost three times as much per hour to run. Ed said: “they think it’s all the same metal.”

I wrote that down. I did not use it in the piece because I could not figure out how to use it without making it sound like an old man complaining. But it has stayed with me, and I am bringing it up now because I think it is the same observation. The portal flattens the underlying reality.

The portal makes a four-decimal-place part look the same as a two-decimal-place part. The portal does not show you the press brake. The portal does not show you the empty section of floor where two machines used to be. The portal shows you a form to fill out, designed to look like every other form to fill out, calibrated to go down as easy as possible, with the beige courtesy that signals: do not worry about what is happening on the other end.

We have taken care of it. Just submit the order. The cost of this, in the case of Ed’s shop, is not abstract. The cost is that the buyers do not know what they are buying, and so they cannot make good decisions about what to buy, and so the market sends the wrong signals back to the fabricators, who then make the wrong investments, and over the course of a decade or two the entire industry deteriorates because the layer of mutual knowledge that used to connect buyer and maker has been replaced by a portal that flattens everything into a line item.

The portal is efficient. The portal is also corrosive. Both things are true. I think the beige is corrosive in the same way, on a larger scale.

It is corrosive because it conceals the actual conditions of the things you are interacting with. The food delivery app conceals the conditions of the kitchen and the driver. The dating app conceals the conditions of the matching algorithm and the people it has decided you are not allowed to see. The bank app conceals the conditions of the bank, which include, somewhere, a series of decisions about who gets loans and who does not, made by models trained on data sets the bank does not publish.

The beige is the visual signature of this concealment. It is the smooth surface over the complicated machine. I want to be fair. There are good reasons for some of the smoothing.

Accessibility is a real concern. A button needs to be readable by a person with low vision. A typeface needs to scale across devices. A color palette needs to work for someone with red-green color blindness.

The standards that have taken shape over the last fifteen years have, in many cases, made software more usable for people who were previously excluded from it. This is a genuine gain and I do not want to dismiss it. But I notice that the appeal to accessibility, when made by the people who design these systems, almost always cashes out in favor of less variation, less idiosyncrasy, less authorship. Accessibility, in the mouth of a product manager, becomes a reason to use the same template as everyone else.

The actual disability community is not asking for this. The actual disability community is asking for specific features , screen reader compatibility, adjustable font sizes, alternative input methods , which are entirely compatible with idiosyncratic design. You can have a website with a hand-painted header and full screen reader support. The two are not in tension.

But the appeal to accessibility gets weaponized, internally, to shut down any visual choice that someone in the meeting finds unfamiliar. The beige is not, mostly, an accessibility win. It is a managerial win. It is what software looks like when the people approving the mockups are afraid of being blamed for a choice that turns out to be unpopular.

On my walk this morning the river was steaming. The temperature had dropped overnight and the water was warmer than the air and a low fog was running along the surface from the dam down toward the bridge. There was a man in a wool coat standing on the bank with a camera on a tripod, taking the same picture I have seen people take a hundred times. I have taken it myself.

There is no reason not to take it. The fog is beautiful and the camera knows it. But I thought, watching him, about the thousand other people who would, over the next several days, post similar pictures from similar bridges with similar fog, processed through similar filters, on similar feeds. The pictures would all look like each other.

Not because the photographers had copied each other but because the tools had narrowed the range of what was easy to do. The default filters on the default apps produce a particular look. The look is fine. The look is, in some narrow technical sense, good.

But it is one look, and it has displaced ten thousand other looks that were available to a person with a film camera and a darkroom and a particular set of choices to make about exposure and development and printing. Each of those ten thousand looks expressed a person. The current look expresses an app. This is not a complaint against digital photography.

Digital photography is a remarkable thing and I would not give it up. The complaint is against the convergence, the way the tools, through their defaults, gently and persistently herd everyone toward the same handful of outputs. The herding is not coercive. No one is forcing the photographer to use the default filter.

But the default is what most people use, because the default is what is in front of them, and the default has been chosen by a small team at a company who optimized for what most people would not actively dislike. The default is the beige of the visual world. What I keep coming back to is the question of what gets made, and what does not get made, when this is the prevailing aesthetic. The things that get made are the things that fit the template.

The things that do not get made are the things that do not fit the template. Over time, the people who would have made the non-template things either learn to make template things or stop making things. The range narrows. The bank of available work, the collective inheritance from which the next generation will draw, becomes thinner and more uniform.

This is a slow loss. It is not visible in any single year. But over a decade or two the difference becomes apparent, the way the difference between an old-growth forest and a tree farm becomes apparent only when you walk through both. I am aware that I am writing as someone who came up in a particular moment and who has, naturally, an attachment to the textures of that moment.

The Rockford Register Star newsroom in 2002 had a particular smell , old carpet, coffee, paper , and a particular sound, and the sound and the smell were part of the work. When the paper consolidated its operations in 2009 and moved most of the editing to a different city, the smell and the sound did not transfer. They could not. The work that came out the other end was, technically, the same work.

But it was thinner. Anyone who had been in the newsroom could tell. The thinness was not, by itself, an argument for the old way. The old way had its own pathologies, including the smell, which was actually unpleasant.

The thinness is just the thinness. It is what is left when you take the place out of the work. The current consumer software environment is, I think, the thinness at industrial scale. The places have been taken out.

The smells have been taken out. The men behind the counters have been replaced by chatbots that have learned to apologize in a tone calibrated to test well in user surveys. The work that comes out the other end is, technically, the same work. The bank transfers happen.

The packages arrive. The matches are made. But the thinness accumulates, in the user, as a kind of low background sadness that the user often cannot name and so attributes to other things , to the weather, to the political moment, to their own failings. I do not want to overclaim.

I am not saying the beige is the cause of the broader unease. The broader unease has many causes and most of them are larger than design. But the beige is one of the surfaces along which the unease moves. It is one of the registers in which the unease becomes visible, if you look at it directly, which most people do not, because the whole point of the beige is that you do not look at it directly.

You look through it. You are meant to look through it. I want to come back to the man at the coffee shop on Linn Street, scrolling through whatever app he was scrolling through, in the soft gray and the muted teal. I have thought about him several times since.

I do not know what he was looking at. I do not know if he was happy or unhappy, engaged or bored, present or absent. The interface gave me no information about his inner state, which is exactly what it was designed to do. The interface presented him as a generic user performing generic actions on a generic surface.

The genericness was the achievement. But he was a specific person. He had a specific coat and a specific posture and a specific way of holding the phone , slightly canted, his thumb working faster than his eyes seemed to be moving. He was somewhere in his fifties.

He had a wedding ring. He drank his coffee black. He looked, to me, like someone who had been doing this for a long time and who no longer remembered when he had started. The gap between his specificity and the genericness of the surface he was interacting with is the gap I have been trying to describe.

It is the gap the beige produces. The beige does not know him. The beige cannot know him. The beige is the visual signature of a system that has scaled past the possibility of knowing any individual user, and that has, in the absence of knowing, produced a surface that pretends not to need to know.

The pretending is the design. He finished his coffee and left. I watched him go. The barista wiped down the table where he had been sitting and did not look up.

Outside the snow had started, a thin dry snow that was not sticking to the sidewalk, and somewhere across the river the heron was probably standing in the shallows again, neck folded back, waiting for the strike I would not be able to see. The phones in our pockets were all running the same handful of apps. The apps all looked like each other. The looking-like-each-other was not, in itself, the problem.

The problem was what the looking-like-each-other concealed, which was the actual conditions of our lives, which were not, despite the surface, the same. We were not the same. We had never been the same. The beige was a lie about us, told gently, in a voice calibrated to be unobjectionable, by a system that had no face.