2026-11-01 21:45:04
My feed is mostly designers talking about AI, and the temperature of all of it is the same: Something is coming. You need to get out in front of it. You need to be the one doing the changing rather than the one being changed.
I don’t disagree with most of the individual advice. But read enough of it in one sitting and what surfaces underneath is competition — ordinary competition, of the kind that has always been there.
The worry takes two shapes. Either the client no longer needs a designer, because the tool supplies the design. Or another designer, using the tool, takes the job you would have had.
Both are worth taking seriously. But neither is new.
There has always been someone faster than me, someone cheaper than me, and someone better than me. This was true the day I started my career and it will be true on the day it ends. It describes a market, not an era.
Designers — actually, all in the professional services — have a folk theorem for living with this reality: Fast, cheap, good — pick two. I’ve heard it said and read it written countless times. I’ve said it and written it, too, to win arguments and contracts.
Today I got curious about where the saying comes from, and went looking. Nobody knows. The phrase itself gets attributed to something called the Common Law of Business Balance, and from there to John Ruskin — an attribution with no evidence behind it whatsoever. The idea underneath has a slightly better paper trail: an engineer named Martin Barnes drew it as a triangle, with cost, time and quality at the corners, for a project management course in 1969. Even that isn’t an origin. Versions of it were in use by the 1950s, and no one can say who put it that way first. This beaten-to-cliché piece of professional wisdom is a cipher.
Which is useful to remember because it’s also, often, wrong. There are as many designers who are fast and cheap and good as there are those who are slow, expensive, and bad. I have worked with some of each. Thinking back on this now, I am surprised that no one ever responded to the “pick two” challenge by saying, “I want all three, I just won’t be getting them from you.”
My entire career has been punctuated by tools that sped up production. Desktop publishing collapsed typesetting and layout into something one person could do in an afternoon. Then Photoshop. Then the rest of the Adobe suite, Illustrator and InDesign. Then Sketch. Then Figma. Each one made producing the work materially faster, and each one set off a race to master it before everybody else did.
Each of them may have changed what a day at work felt like, but none ended the profession.
What makes faster feel dangerous is that we convert it immediately into cheaper, and we do that because most designers price their time. If the work takes four hours instead of twelve, charging the twelve-hour price feels indefensible, and somebody else will happily charge for four.
That’s a pricing problem rather than a technology problem, and it’s been the same pricing problem through every acceleration I just listed. A practice that charges for outcomes absorbs acceleration as margin. A practice that charges for hours absorbs it as a pay cut.
As for cheaper, it arrived long before any of this, and it arrived by geography. I work closely with a designer who lives in Eastern Europe. This person is, by some distance, among the most talented designers I have ever worked with. They didn’t go to design school. I went to the Rhode Island School of Design, which bought me a credential the marketplace sometimes reads as shine, and I’m clear-eyed about the fact that — on the measure that actually matters — my colleague isn’t behind me. They can also charge less than I can, because rent where they live isn’t rent where I live.
None of that has anything to do with AI. A cheaper designer has been available to every client I have ever had as long as we’ve had an internet connection. And just in case anyone thinks that professional design competition from overseas is a more recent phenomenon, the first designer I met when I visited Malaysia before moving there in 2004 was born and raised in Penang, had no design degree, and had dozens of clients most of whom were headquartered in the United States. He was fast, cheap, and good. That’s why they paid him, rather than someone else.
There is an objection to what I’ve said so far. I’ll try to put it in its best form rather than a convenient one.
Every tool I listed made production faster without making the thing. Desktop publishing didn’t design the page. Photoshop didn’t choose the image. Figma doesn’t know what the product should be. AI, as a design tool, is different in kind, because it produces the artifact. So the comparison to Sketch and Illustrator is a weaker argument than it sounds.
I think that objection is right about the tool and wrong about the job. It only holds if the artifact itself is better.
Imagine two designers given the same brief — an identity system for the same client. Put the results in front of a panel, blind, with no knowledge of how either came to be. The panel picks a winner. They’re qualified to do it and their reasoning will be sound.
Now ask the client which designer they’d hire again.
It is by no means a given that the panel and the client will agree. The two answers can fork, and when they do it isn’t because the client has poor taste. The panel is assessing an artifact. The client went through something. They were asked questions nobody had asked them before. They had a position drawn out of them that they couldn’t have articulated on their own. At some point they were told that the thing they wanted was the wrong thing, by somebody who had earned enough credit to say so. None of that is necessarily visible in the file — nor should it be.
I wrote recently that expertise is becoming subjective — that where knowledge is cheap, experience is what differentiates. If that’s true, it cuts in both directions. Experience doesn’t only distinguish you; it aligns you with some people and not others. A prospect’s fit with how you’ve worked, what you’ve lived through, and how you talk about it is doing as much work as the expertise itself.
Which means there’s no better in the abstract. There’s better, for them.
There’s also a posture question here. To believe that your prospect couldn’t do better than you is a level of audacity that runs counter to how every designer I know actually thinks — and yet it’s close to what we’re advised to project. In a world this populous it’s absurd to imagine you’re the best at anything, and more absurd still to build a public persona on the premise. That advice persists because it works on a seductive technicality: whether or not you believe you’re the best, behaving as though you are turns out to be convincing to other people. Notice who that advice is sold to. We are the market for it; our clients have never heard of it. Rejecting that is realism, not modesty. The world is too crowded with designers to spend a second ranking yourself among them.
Finally, the same takes on the current state of things also sell a tempo. They essentially argue that AI compels us to move faster in order to stay ahead. But if experience is the differentiator, speed is the thing eroding it. Going slowly through a problem — getting it wrong, trying it on a real client, finding out — is where the differentiation was built all along. The availability and performance of AI doesn’t necessarily run counter to that, of course, but a willful misunderstanding of it does.
When I moved from Sketch to Figma I thought it mattered. I had my reasons. I argued about it. Designers I respected stayed on Sketch for years afterward and we all thought the other was wrong.
In the end, it didn’t really matter. I know designers who still work primarily in Photoshop. In the last twelve months, I actually saw a website design produced in Illustrator. Of course I think these are mistakes, bordering on the absurd. But it’s also none of my business, and as far as I can tell it hasn’t cost them a job. Isn’t that interesting?
But it makes some sense. After all, nobody has ever asked me what tools I use. Except, of course, another designer.
I don’t know what this technology will do to the profession over the next twenty years, and I notice that the people who claim to know are usually the ones with something to sell. What I can see from here is smaller and steadier than that.
Somebody out there is faster than me. Somebody is cheaper. Somebody is better. Again, not modesty, but the realism that comes from decades in the work, and I have met them. All three have been a phone call away from every client I have ever worked with. Any prospect can have them, faster, and at a cheaper price.
But they don’t have me. And they don’t have you.
Be singular.
2026-10-30 23:13:50
My mother took a few courses at the University of Delaware when I was eight. I have one clear memory of it. She is sitting on the floor of our living room with books open around her, filling a spiral-bound notebook with writing.
I don’t remember our conversation word for word. But I have always remembered the core of what she told me, which is that writing down what you are learning makes it stick.
At the time, this made all the sense in the world to me. You can read something and remember some of it. But when you write it down, you test that memory and create a new one. It’s a form of review that you make physical, and that translation stores it within you more securely.
This lesson from my mother has aged well. In fact, just recently, something about it has newly occurred to me. In 1988, the slow, hand-cramping, notebook-filling method was really the only choice available, other than not doing it at all — just reading and hoping for the best. But for someone who believed in this method of study, there simply was no faster way.
For most of my career, having expertise in public has worked about the same way. You read, you worked, you formed a view, and then came the hard part: you tried to say it to somebody else.
Many years ago, in my own personal study of philosophy, I learned a principle that has held: Knowing something to be true is easy. Showing it to be true is difficult. Over many years of writing publicly, I have retread that principle in my mind again and again. It has been hard work.
You had to decide what you actually thought. You had to find the shape of the argument, and the order the pieces went in. You had to craft the sentences that would transcend being informative and persuade. Then there were technical challenges — where and how to publish — and the duty of engaging afterward.
All of that took time and effort, both filters. They screened for people who had something to say and could also say it — two abilities that only sometimes show up in the same person. Plenty of people who knew a great deal never got through it, and that was a real loss. I don’t want to romanticize the gate.
But the filter did something else, and this is the part I think I have taken for granted until now. It was also where the expertise got made.
It’s tempting to describe what has changed as the arrival of cheap facts. I did that very thing when I wrote down the note that would become this essay: “Facts are a commodity now…” Except “now” is wrong by about twenty-five years. Facts got cheap when search did. Anyone with a connection has been able to retrieve more information than they can use since roughly 1998, and nobody mistook that for knowing anything (or at least they shouldn’t have).
What has newly become cheap is everything after retrieval. Synthesis. Structure. The paragraph that holds the idea still. The part search never did for you, and the part that used to take the week.
The research on this is older and more settled than the current conversation suggests. In 1978 Slamecka and Graf named the generation effect: people remember material better when they produce it themselves than when they read it — which is, more or less, what my mother told me. In 1994 Robert Bjork gave the wider phenomenon a name I find hard to improve on — desirable difficulties. Conditions that slow you down while you’re learning, and make your performance worse in the moment, reliably produce better retention afterward. Difficulty isn’t the price you pay for learning. A good share of the time, it’s the way learning happens at all.
Bjork’s work carries a companion finding that explains more of the present moment than anything else I’ve read. Easy methods feel more effective than they are. Rereading a chapter produces a powerful sense of familiarity, and familiarity is very nearly indistinguishable, from the inside, from understanding. He calls it an illusion of learning.
Hold that against what has actually changed. What got cheap is the production of fluent, well-organized explanation. And producing a fluent, well-organized explanation of something feels, from the inside, exactly like understanding it.
I was browsing LinkedIn recently and came across a run of videos — all nicely shot, well lit, confident — of people explaining things I have read before, in language I have read before, with the delivery of someone handing you something rare. The scripts had the agent vernacular that is off-putting and uncanny when delivered by a human. My first reaction was the easy reaction: these people are pretending.
I am determined to stop thinking that. The uncharitable reading is the less interesting one. I suspect most of them sincerely believe they know what they are talking about. They produced a clear explanation of it a few days ago, it came out well, and that is what knowing something feels like.
The test of expertise is application. The thing must be tried. If it fails, or if the result is different from what you expected, the gap between what you thought and what actually happened is the knowledge. The process is slow. It is humbling, if not frequently humiliating, and I know of no substitute for it.
The speed that manufactures false confidence is the same speed that carries you past the point where it could be corrected. If a subject takes four days instead of four months, you are on to the next one before anything has had a chance to not work. The illusion goes untested, so it never breaks.
And here is the part that bothers me most, because it’s a phenomenon I have observed for my entire career. The people doing this are largely not applying what they publish. They are selling it for somebody else to apply. Which means the disconfirmation, when it comes, arrives somewhere else — in a team that reorganized around the advice, or a budget that was spent on it. The person who had the thin idea is rarely the person who finds out it was thin.
Back in the SEO days, lots of people did lots of real work to sell snake oil. Now it’s easier to sell snake oil.
The easy solution to this predicament is a division of labor: machines handle facts while humans supply wisdom. That’s too tidy, and as my observation above shows, it flatters us.
Experience is not load-bearing everywhere. For most questions with a determinate answer, a model will serve you better than a colleague would, and I use one that way constantly.
But, suppose three courses of action are open to you and all three are defensible, each supported by real evidence. The facts are in and they do not settle it, because they were never going to — the choice turns on things that aren’t in the evidence. What your organization can actually sustain. What you are willing to be wrong about. How much patience the people around you still have. What the experience of one choice over the other will be.
At that point, more information stops helping. What helps is somebody who took one of those roads and can tell you where it went. Not what the literature says. What happened. What it felt like. How minds changed after experiencing the outcome.
That person gets more useful to you the more they resemble you: similar constraints, similar temperament, similar appetite for being wrong in public. It may be a strange thing to say about expertise, but I think it’s true: Expertise is a flavor.
I would like to finish by saying that earned experience therefore wins, the sea of sameness recedes, and the people who have actually done the work get paid for it. Wouldn’t that be nice and good.
But I don’t think so — not out of pessimism, but because nothing I can see would actually make it happen.
Confident, frictionless content outperforms specific, hedged, hard-won content, and it did long before any of this. Experience is lumpy. It arrives with qualifications and exceptions and the occasional admission that something didn’t work. Fluency has none of that, and fluency is now available to everybody at volume. The likeliest outcome is that earned experience becomes more valuable in fact while being less rewarded in practice, which is an uncomfortable pairing and not a new one. At least while we are within this new technology bender.
So none of this is a market prediction. It’s a prediction about us — what we’re doing to each other and to ourselves.
I should be plain about my own position, since I’m making an argument about outsourcing while obviously not doing everything “by hand.” I use these tools constantly and I would defend how. The distinction I hold to is whether I am bringing something or only making a request.
Anyone can go to a model and say generate an answer for me; that requires nothing but enough awareness to shape the request. What I want is different — a tireless editor, a fact-checker who doesn’t get bored, a devil’s advocate with no social stake in being agreeable. Something that takes what I’ve already formed and tries to break it. Several things in this essay are here because I was argued out of a worse version.
I think that’s a meaningful distinction; perhaps it’s the only one that will matter. It is also the single easiest thing in the world to believe about yourself and the hardest to verify from the inside. It’s the subject of this essay pointed back at me.
That’s, in part, why I keep publishing in a form that leaves a record. Writing in public with dates on it means that what I thought two years ago is still sitting there, being wrong in ways I can see now. That’s less embarrassing than it is useful. A record of a mind changing is the only evidence of experience I know how to produce.
I used to think my mother was storing what she learned in that notebook. She was doing something nearer to the opposite. The writing was the learning, and the pages were what it left behind.
Mom had no alternative in 1988. I have one every morning, and her slower, better road is still there, waiting to be taken or not.
2026-10-28 22:35:55
The other day I changed a light bulb, and then stood there for a minute holding the old one.
The bulb I had taken out was an incandescent: sixty watts, around eight hundred and sixty lumens, rated for about a thousand hours. The one going in draws nine and a half watts for the same light and is rated for twenty-five thousand. Same socket, same room, same brightness. A sixth of the power, twenty-five times the life, a few dollars at the hardware store.
That is an absurd bargain, and yet it has become nearly invisible. A light bulb is among the oldest and plainest force multipliers we have — it extends the working day, which is to say it extends the usable portion of a life. Every tool in that category works the same way. Glasses correct a lifetime of vision with two pieces of curved glass. A pen, against a quill and an inkwell, is a small object with almost no supply chain behind it and a permanent gain in how quickly a person can think onto paper. These things spread because they gave back enormously more than they cost. The ratio is not incidental to the category of force multiplier; it is the definition of it.
Standing there, it occurred to me that the thing holding the thought about the bulb runs on only twice what the bulb does.
A human brain draws about twenty watts. Two percent of body weight, twenty percent of the body’s energy budget, somewhere near four hundred and fifty calories a day. I had guessed at this a few days earlier, into a voice memo, and said two thousand — then hedged that it might be high. It was not just high. I was wrong by a factor of four or five, in the direction that flatters the machine, about the running cost of the only device I use every waking second.
The obvious thing to do with twenty watts is set it against a model.
An hour of my thinking costs twenty watt-hours. A standard text query costs between a quarter and a third of one — Google’s median figure for Gemini is 0.24, Epoch AI puts it near 0.30, Sam Altman has given 0.34 for ChatGPT. So an hour of me is worth about sixty queries. If a model does in one query what takes me an hour, it wins by something like seventy to one.
Except that comparison is wrong.
It assumes that while the model works, I stop. I don’t. I write the prompt, read what comes back, judge whether it is any good, answer the clarifying question, correct the thing it misread, approve the action. I am present for all or most of it, and my brain does not power down to watch. It keeps drawing its twenty watts through the whole exchange, and through the reading and the judging afterward. Whatever the model spends is added to what I spend rather than replacing it. The only question that matters is how much of my time it saves.
A query and a task are also not the same thing. An hour of my own work might hold a dozen small tasks, and a single task delegated to an AI might take six queries before it’s completed.
So the comparison is a sum rather than a ratio. Say a task I would have spent an hour on takes six exchanges and twenty minutes of my attention. The exchanges cost around 1.8 watt-hours. My twenty minutes cost 6.7. Eight and a half watt-hours against twenty.
This better accounting shows AI to be better by a factor of two and a half. That is a real gain! It is also not seventy, and the gap is the point.
And even two and a half assumes there was an hour. One hour is an even unit and it makes the arithmetic tidy, but a good deal of what I hand to a model does not come in that shape. Sometimes I ask because I don’t feel like looking something up, and the “hour” I am saving was four minutes. Sometimes I ask for something I could not have produced at all, in which case there is no hour to subtract and the cost is simply net new. The sum only works where a real counterfactual exists. I am not at all sure how much of my own use that covers. I want to come back to that.
The same problem sits on the other side of the sum.
A third of a watt-hour describes a short exchange — a question, an answer. It does not describe a long document read in full, or a request that reasons its way toward a conclusion, or an agent sent off to do four things on my behalf. Those are different operations using the same term — query — and the term is what we’re calculating against.
The numbers underneath can move by orders of magnitude. Epoch’s estimate rises to roughly 2.5 watt-hours for seven and a half thousand words of input, and to forty for seventy-five thousand. The IEA’s 2026 figures put a standard agent request at 1.1 watt-hours and an agentic request that reasons at fifty.
So let’s run the same calculation again with these larger numbers. Six reasoning requests in place of six ordinary ones equals three hundred watt-hours. My supervision has not changed — still twenty minutes, still 6.7 from me. But now, it’s three hundred and seven watt-hours against the twenty it would have cost me to sit and think it through myself.
Delegating has gone from two and a half times cheaper to roughly fifteen times more expensive. And nothing in the experience told me. Someone asked a question, I answered it, I approved something.
Obviously, this calculation is subjective. The six exchanges and the twenty minutes are mine, not a standard measurement — substitute your own and the multipliers move. But what doesn’t change is the basic shape of the arithmetic. The human side is nearly fixed; the machine side can swing by a factor of nearly one-hundred and seventy.
Reasoning agents are precisely what is being built. The cheap, reassuring figure is a description of the sort of workload the industry is in the process of leaving behind.
There is one more imbalance we should consider, and unfortunately, it starts to destabilize all of this math.
The brain’s twenty watts is all-in. It covers learning, which is continuous, builds on itself from a handful of examples, and requires no separate training phase. It covers memory, and forgetting, and the constant structural repair of the thing doing the work. It runs my vision and my balance and my heartbeat on the same budget at the same time, and can do it steadily for decades. All on food.
The agent’s third of a watt-hour is marginal. It is the cost of one act of generation on hardware that already exists. It excludes the training run. It excludes fabricating the chips, building the hall they sit in, and the water and power that keeps them cool and running. It excludes the fact that they last for closer to three years, not decades. It excludes the capital being spent to make the marginal number look good.
Set an all-in figure against a marginal one and you can prove nearly anything. The industry’s favourite statistic — a third of a watt-hour per query — is true. It is also an answer to a far narrower question than the one being asked.
There is a word for capital spent to make a price look lower than it is.
Nobody is concealing the subsidy, exactly; it simply isn’t priced.
Consumer subscriptions are among the most heavily discounted tiers on offer, complex queries lose money on the transaction, and OpenAI is projected to burn something like fourteen billion dollars this year. The picture is not uniform — Anthropic’s margins moved from roughly negative ninety-four percent in 2024 to around positive forty in 2025 — and I don’t want to overstate a situation still in motion. But the shape is clear enough. Whatever I pay, somebody else is covering the difference, and the true size of that gap is not clear. It’s a number nobody in the business has any reason to publish.
Which puts generative AI in a strange position for a force multiplier. It may be the first general-purpose one whose ratio has to be hidden for the thing to spread. The bulb in my hand cost a few dollars and puts the accounting on the box.
The math is interesting enough, I suppose. But here’s the part I find truly fascinating.
We have spent the last few years asking what becomes of human beings in a world where AI exists. It is a meaningful question. But the manner in which it is often asked and answered has a pallor of dread to it that, I think, comes from considering the wrong score.
The common AI talking points are akin to a leaderboard; we hear about benchmark scores, capability curves, what the model can do that a person can do. On a leaderboard, the machine is gaining and we are standing still.
Measure the ratio instead and the comparison inverts.
There are more than eight billion devices on this planet that learn continuously from almost no data, repair themselves while running, last for decades, and do all of it on twenty watts. We did not design them. We still don’t understand how they work. On the measure that will actually decide what this technology becomes — not its capability, but its cost per unit of capability — the thing we are frightened of is not close to the thing we already have. Our brains. Us.
I argued a while ago that AI is best understood as infrastructure, and that the sooner we think of it that way, the sooner we can stop being astonished and get back to making things. Here is the part of that I didn’t follow through on fully: We do not judge infrastructure by how impressive it is. We judge it by cost per unit delivered. Nobody rates a power station on the beauty of its lightning.
What I want is the number on the side of the box — what it costs to do the thing, counted the way a brain is counted. I don’t expect to be given it. I notice that nobody is offering it.
2026-10-26 02:11:52
For the last thirteen years, I have lived in a little house. Even as my family grew, it has remained more than enough space for us all.
Below the house is a basement. The majority of it is merely crawlspace, but a bit less than a quarter of the house’s footprint is a true, standing-height space. After some years of observation and optimization, we’ve managed to make it a functional, long-term storehouse.
Shelves like library stacks line the edges. There are two freezers. On and within them is a year of food and supplies. These are the things our household actually uses up — the ordinary, unglamorous inventory of living here. We rotate it. Things come in and things go out and get replaced. It isn’t a sealed stash waiting for the end of the world; it’s a working pantry with a year of runway. We buy what we depend on when the price is to our advantage rather than when we’ve run out, and a supply chain interruption would only register if it lasted longer than any other in our lives so far.
We practice the kind of prepping that most people don’t think “prepping” is.
The house itself is the first we have purchased. We intend it to be the only one.
For a long time I understood these as two separate facts — one about caution and one, perhaps, about taste. But they’re really the same decision. We’ve created a life that can be kept, and the way you make a life keepable is to build it so that continuing doesn’t require growth.
But before I allow this to sound like the preamble of a sermon about restraint and simplicity, I should be honest that most of my thinking comes from a very different place — less serenity and more fragility.
I have always been prone to worry. I find it easier to imagine outcomes shaped by pessimism more than optimism. The storehouse exists because I spent years considering what I would do in the worst case. The only durable answer I found was to make the present both worth keeping and cheap to keep.
This is a maintenance future. And for me, it’s optimism made possible by pessimism.
Narratives about the future sort into two piles more reliably than you’d expect, and the sorting has little to do with optimism. Optimism and pessimism are the tone. Underneath the tone is a quantity.
One pile says there will be less. Fewer objects, less wealth, less freedom, less autonomy — a future in which people have lost things they were accustomed to having. Apocalypse is the loudest version but not the only one; most of what reads as pessimistic futurism is simply an expectation of subtraction. The other pile says there will be more. More technology, more objects, more wealth, more reach, and nearly everything else we file under progress.
The two can braid, which is where it gets interesting. A story about settling another world is a more story usually running on a less premise: this planet is failing, so we leave. It still reads as hopeful, because at least we don’t die with the planet.
Once you have established a more/less axis, you can cross it with the one everybody reaches for first, utopia and dystopia, and the result is more honest than either axis alone, because all four corners turn out to be occupied.
Two of them are the readings we default to. A future of less is a dystopia: people die, freedoms contract, wealth evaporates. A future of more is a utopia: energy becomes effectively free and anyone can live and power a life without paying for the privilege. Solarpunk lives in that corner — the sun being, relative to us, close enough to infinite that drawing from its power rather than our planet’s is just a design problem.
The other two corners are the ones worth the trouble.
A future of less can be a utopia. Not comfortably, but the case is available. The American dream is not a life that ten billion people can hold at once; the square footage, the car, the wealth per household, the arithmetic has never worked. A reset that reduced the material expanse of a single human life rather than the number of human lives would be a correction. The bridge to it would be painful — a dystopic transition — but that alone doesn’t make the far side worse than here.
A future of more can also be a dystopia. I would argue that this is the corner we presently live in. It doesn’t have to deliver equally to all humans to remain the thing our culture is programmed toward. Growth measured on the wealth axis alone, after all, has not historically distributed itself. It concentrates — it always has — and the distance between top and bottom widens as the total climbs. More, in that configuration, is a description of the gap.
So where does maintenance go?
Not in a fifth box. Maintenance is zero on the first axis — no more, no less, hold — and the second axis stays live. And that’s interesting to me: a maintenance future can be utopic or dystopic as readily as any of the other four.
The dystopic version lands upon our psychology immediately because it’s claustrophobic. Anathema to our current culture. You will have exactly this much, forever. When put that way, it stops being a choice and becomes fiat. Feudalism was a maintenance future; provincial lords living lives dedicated to persistent conflict over resources, limiting knowledge and human potential. It was never going anywhere. The same is true of any caste arrangement, any frozen hierarchy, any system in which a small number of people determine what is enough for everyone else and then hold them to it. In other words, a more culture can be maintained, so long as it’s more for some and only so much for others.
Which means the idea rests on a distinction that’s easy to state and hard to keep: maintenance chosen is a practice, and maintenance imposed is a lid. Our basement storehouse is a decision we made about our present and future. It would be a different space entirely if someone had measured it out for us.
The other objection is about creativity. This is the one that comes to mind first for me, but I think it’s actually more easily resolved.
Creativity is coded productive. Making what didn’t exist. A net-new object, a net-new idea. If the future is supposed to hold steady, it can sound like a halt to invention. That is a far worse proposition than limiting one’s habitable space.
In 1969, the artist Mierle Laderman Ukeles wrote a manifesto for what she called maintenance art, as a proposal for an exhibition about care. She had recently become a mother, and noticed that the avant-garde of the moment prized repetitive, tedious labor when it occurred in a gallery and ignored it entirely when it occurred in a house. Development, she observed, was about individual creation; the new; change; progress; advancement; excitement. Maintenance was everything that preserved and renewed whatever development produced. Development and maintenance together, she wrote, are “the sourball of every revolution” — and then she asked the question that has trailed the manifesto ever since: “after the revolution, who’s going to pick up the garbage on Monday morning?”
What she was after was the decision itself. Somebody, at some point, had decided that maintenance was not creative work, and the decision had gone unexamined ever since. It was arbitrary when she made the point, and it has only grown more expensive.
Ukeles was arguing about value. There is a second argument available about matter, and it supports hers. Nothing we make physically has ever been made from nothing. Every object in any home is a rearrangement of material that already existed; most of it was something else first. A maintenance culture doesn’t withdraw the brief for making things, it rewrites it. Get better at recovering materials. Prefer the ones with a natural lifespan and some plausible relationship to the world they came out of. Design for the second use. Timber instead of concrete, to take the plainest case.
I won’t pretend that costs nothing. It does impose limits — on making things whose existence requires extraction and whose end is a landfill — and those limits would eliminate whole categories of objects we take for granted. But a limit has never reduced invention. It redirects it. Every good design is as much born from limitations as expansive thinking. The Eameses arrived at the molded plywood chair by way of a wartime leg splint, and it is still in production seventy years later.
Perhaps shifts in this direction begin aesthetically. We’ve become accustomed to futurism arriving with a synthetic sheen — chrome, polymer, seamless surfaces that describe a different kind of world than the one from which we were born. But this is an association, not a property of the future, and associations move. There are only so many photographs of the Great Pacific Garbage Patch, or of the fires at Agbogbloshie, that a person can absorb before plastic, titanium, and glass read less as the makings of tomorrow and more as grave goods.
Revulsion can’t be written off as a pose if it’s necessary. The materials-porn of the early unibody Apple ads has shifted to experience-porn. That’s not a bad thing, but it’s only a partial step. We can sell more phones on the basis of the promise of good experiences than over the seduction of a surface, but that doesn’t mean that the surface is no longer a space for innovation. We’re still selling expensive, precious, fragile, short-lived and landfill-destined pocket computers. The through-line to maintain is whatever good they provide us; the rest should be revolted away.
There’s no point in trying to convince anyone that suddenly an iPhone is ugly. That’s not a winnable aesthetic argument. But when we start seeing through the surface aesthetics to the contingency underneath, we can shift our shared sensibilities far more easily. Contingency is not something anyone finds beautiful for long.
There’s a story I’ve referenced many times, in articles and on podcasts, about a forest in England. The beams in a college hall were found to be riddled with beetles, centuries after the building went up. The college despaired of replacing them until someone thought to ask the forester, who explained that a grove of oaks had been planted when the hall was built and tended ever since, for exactly this purpose. Someone five hundred years ago had known the beams would fail, and had planted the answer. It’s an inspiring story.
If my memory serves me, I first read it in a magazine sixteen years ago, in GOOD’s Slow Issue. I’m now fairly sure it isn’t true. The story reaches us secondhand, originally by way of Gregory Bateson and Stewart Brand; and the college itself has called it nonsense. When a researcher asked the archivist and the clerk of works whether any replacement grove had existed, nobody knew.
I hold none of this against Brand, who has done more than nearly anyone to make the long view thinkable. He wrote How Buildings Learn, which argues that what matters about a building is how it gets kept rather than how it was designed, and he helped found an institution whose principal object is a clock meant to run for ten thousand years. The man who handed me the false story has spent his career on the true version of it.
What interests me more than the error is my own part in it. That magazine told me the story was probably apocryphal, in the text, at the time. It also had no beetles in it; those come from the version Brand popularized and credited to Gregory Bateson. So the caveat fell away and a crisis accreted in its place, inside my memory, over sixteen years.
I’ve wondered why, and I think the answer is that the story is shaped like a rescue. A crisis, a revelation, a salvation, in that order. A maintenance story seems far more compelling when it comes with a hero’s journey.
On the other side of the world is a story that needs no embellishment.
Ise Jingū, a Shinto shrine in Mie Prefecture, Japan, has been taken apart and built again every twenty years for roughly thirteen centuries. It is rebuilt from new timber on an adjacent plot, and then the old site is cleared. The new shrine is no larger than the one it replaces, and no better. Ise Jingū is the same shrine, again.
In 1923, the shrine began planting its own cypress trees. The sacred forest is intended to be a self-sustaining source of the wood that the rebuilding of Ise Jingū consumes. By the 2013 rebuild, about a quarter of the timber came from that forest. By 2123, two centuries into the program, it is meant to supply all of it. The trees being tended now are for a building that will be raised by people not yet born and taken down again by people after them.
What I admire most in this example is that maintenance roots deeper than the cypress trees. Every generation of carpenters builds the shrine once, perhaps twice, having learned it from the generation that built it last, and teaches it to the generation that will build it next. The knowledge is archived practically in a cycle of demonstration and observation. At a much longer interval the knowledge would have to be contained in some other way, and history shows us how fragile that is. What appears inefficient — rebuilding a structure every twenty years — is the opposite. It is the most efficient means of knowledge maintenance, wiser even than a five hundred year advance on one cycle of timber.
Contentment behaves less like a disposition than like a stand of trees, which is what I had wrong when I started thinking about this. I’d assumed a maintenance future would require contentment as a precondition — that a civilization would have to arrive at some shared threshold for enough before it could stop reaching for more, and since no civilization has managed that, the whole idea was a fantasy. But contentment has to be planted, tended, thinned, replanted. In a surrounding culture built on the opposite premise, contentment swiftly succumbs to neglect.
That still leaves the question of how a civilization gets to the first planting, and I don’t think it arrives there by becoming content. Contentment is the product, not the material, of maintenance futurism.
There’s a larger instance of the same filter — the one that keeps the rescue and discards the maintenance — and in that case it may have done us a favor. A 1972 report commissioned by The Club of Rome, The Limits to Growth, offered three conclusions. The first was that, on current trends, the planet’s limits would be reached within a century, most probably by way of sudden and uncontrollable decline. The second was that the decline was preventable — that consumption trends could be altered to reach a condition of ecological and economic stability sustainable far into the future, a transition from growth to global equilibrium. The third was that the sooner the work began, the better the odds. Half a century later, the collapse is the part we still argue about. The equilibrium is barely remembered. One of the most-cited futures documents of the last century contained a maintenance future, and we filed the crisis.
We begin by becoming afraid. What we’ve retained of The Limits to Growth helps with that. It’s a specific and unglamorous kind of fear: the fear of fragility. Fragility is the reality of how much of an ordinary day rests on arrangements nobody in the house controls. Fear of fragility scales in a way that cultivated contentment never has, because it asks nobody to want less. It only asks them to see that what they have is contingent — which is the same thing we’re starting to see in titanium and glass. That’s the bridge. Contentment grows on the far side of it, in the practice of keeping what you’ve decided to keep.
Contentment is hard enough for one person to cultivate. Cultures are assembled by billions. So how does something like a maintenance future scale? I don’t know. I’m starting by describing a basement.
But I do want to notice that nothing in this essay which is actually being maintained turns out to be an object. The cypress forest is a capacity. The carpenters are a capacity. Contentment, if it can be cultivated at all, is a capacity. The shrine is just the thing that keeps them in use.
For a long time I understood our basement storehouse as insurance, a year of margin hedging against outside forces I can’t control. But the supply isn’t really the point. The maintenance of it — the practice of keeping — is.
I’ve left the economics of this idea out, though it’s implied in our notions of growth. But there are economic theories of maintenance that I came upon while researching this essay. Kenneth Boulding separated the cowboy economy, which measures success by throughput, from the spaceman economy, which minimizes throughput and measures success by the quality of the total capital stock — including, he was careful to specify, the state of human bodies and minds. That was 1966. Herman Daly built it out formally as steady-state economics, and E.F. Schumacher gave a chapter of Small Is Beautiful the title “The Economics of Permanence,” which may be a better name than mine.
There is also a living field. Lee Vinsel and Andrew Russell’s The Innovation Delusion argues that innovation-worship has starved the work of keeping things running, and Shannon Mattern’s “Maintenance and Care” is the essay I wish I’d written. As for fiction, Le Guin is unmatched. In Always Coming Home, she described a maintenance culture as an anthropology rather than a plot — which may be the only way such a society can be narrated at all.
What I haven’t found anywhere is the second axis. The economics supplies the zero point and the maintenance scholarship supplies the labor, but I haven’t come across much discussion of whether a maintenance future is a good one. The answer seems to be that it depends entirely on who chose it.
