AI Is an Art Medium

I believe, deep down in my heart of hearts, that AI is a new artistic medium. Not a shortcut. Not a threat. A medium. Like oil paint. Like the camera. Like the synthesizer. Like Photoshop.

And I believe the people exploring it now, in its infancy, when the tools are rough and the culture hasn't caught up, will come out the other side able to create work we can't even conceive of yet.

This is not a prediction. It's a pattern.

---

Think about what's already possible. Not in theory. Right now.

An artist trains a model on their own style. Not to replace themselves. To offer an affordable version of their work to people who love their art but can't afford a full commission. The artist sets the price. The artist controls the model. Fans get art that feels like their favorite creator's hand, and the creator earns money from every generation. That's not theft. That's a new revenue model for working artists.

Autonomous AI actors walk onto a stage. Not puppets. Not pre-recorded projections. Responsive performers in bodies that don't have to be human: a walking, talking dog who riffs with the lead actor, adjusts to a missed line, plays the crowd. Theater companies can stage productions that would have been physically impossible. An audience member calls out a prompt and the dog responds in character. That's not replacing actors. That's expanding what a stage can hold.

A composer duets with a model trained on every piece of music they've ever written. The AI becomes a mirror of their own creative history: twenty years of melody and rhythm reflected back, recombined, offering phrases the composer almost wrote but didn't. The conversation between artist and instrument becomes literal. Every musician already does this with their muscle memory. AI just makes the memory external, searchable, responsive.

A filmmaker builds animated worlds that reshape themselves around each viewer. Not choose-your-own-adventure with three endings. Entire visual environments that respond to pacing, to mood, to the way you lean forward in your seat. Participatory cinema that treats the audience as a collaborator, not a consumer. Every screening is different. Every viewer gets the story that resonates with them. The director's vision isn't diluted. It's multiplied.

A novelist stuck on chapter nine generates fifty variations of the scene that won't come together. Not to use any of them. To see the shape of the problem from angles their own brain can't reach. The way a photographer shoots three hundred frames to find the one. The way a sculptor makes a dozen maquettes before touching the marble. Generative drafting. A sketchbook that talks back.

None of these replace the artist. All of them require taste, judgment, vision, and craft. The medium is new. The creative act is the same as it's ever been: a person with something to say, finding a way to say it.

---

But here's the thing. This is a hard line for me. Not because I think the arguments against AI art are trivial. Because I've seen this exact fight before. Over and over. And I know how it ends.

The photograph showed up and painters said it wasn't art. It was mechanical reproduction. A machine pressing light onto paper. No human soul in the process. Painting would die. Painters would starve. The culture would coarsen. [None of that happened.](https://en.wikipedia.org/wiki/Pictorialism) Photography became its own art form, and painting didn't go anywhere. It just stopped being the only way to capture the world.

The radio showed up and musicians said it would kill live performance. Why would anyone buy a ticket when you could hear the music at home? [They were wrong.](https://daily.jstor.org/when-musicians-protested-against-the-radio/) Live performance survived. Radio created new audiences, new genres, new ways of experiencing music that didn't exist before. The pie got bigger.

The pressed record showed up and the same fight happened again. Then the television. Then the synthesizer. Then the drum machine. Then digital audio workstations. Then Auto-Tune. Every single time, the existing practitioners said the new tool would destroy authentic creation. Every single time, the new tool became a medium in its own right. Every single time, the people who got there first made the most interesting work.

Photoshop and digital art tools caught this in the aughts and teens. "That's not real art. You didn't paint it. A computer did half the work." I watched that argument happen in real time in online art communities. The people making it now treat Photoshop as an obvious, unremarkable tool. They forgot they ever objected.

The line always moves. The community always forgets.

---

Every time a new medium appears, people try to build social levees. Rules about what counts. Platforms that ban the new thing. Communities that draw a bright line between "real" and "fake." They genuinely believe they can hold the water back.

But rising water doesn't care about human effort. It doesn't respect social consensus or community guidelines or the way things used to work. It respects physics. It respects inevitability. A rising tide doesn't negotiate.

This isn't nihilism. It's observation. The water rises because the tool is cheaper, faster, and more accessible than what came before. That's the current. You can build all the levees you want. The water will find the cracks. It always does.

The question isn't whether AI will become an accepted art medium. It will. The way photography did. The way digital art did. The way every medium that was once called "not real art" eventually did.

The question is whether you want to be one of the people exploring it now, when the territory is unmapped and the possibilities are wild and the culture hasn't yet decided what the rules are, or whether you'd rather wait until the dust settles and the interesting risks have already been taken.

---

I'm not here to convince anyone. I've never been interested in that. I'm here to make things with the tools I have, in the medium I believe in, with the people who want to explore it with me.

The panic will pass. The levees will fail. The water will rise. And somewhere on the other side, someone who picked up these tools early, who experimented when everyone else was arguing, will make something extraordinary. Something we couldn't have imagined. Something that could only exist because the medium existed.

That's how it's always worked. That's how it works now.

I spent the past three years in blissful maker mode, learning everything I could about AI tools. Not because someone told me to. Not because my job required it. Because I *love* learning new things. I love that feeling of being just competent enough to glimpse the edge of a new world. The firehose still feels like a gift.

I got roasted for it. It hurt more than I wanted to admit.

Not because strangers on the internet said mean things. What hurt was that some of the contempt came from people whose affection still mattered to me. People I wanted closeness from.

What hurt was realizing how many people only knew how to love me if I stayed small enough to be legible. Curious, but not *too* curious. Creative, but not in a way that changed the weather in the room. Weird, but in the charming way. Never in the way that might force anybody else to examine their own fear.

I was learning, in real time, that I had been auditioning for acceptance from people who had already chosen a flatter version of me. I kept explaining myself to people committed to misunderstanding. I kept thinking clarity could save me from being unwanted. I kept mistaking debate for intimacy.

My social life had quietly become a place where I was negotiating against myself. I wanted closeness badly enough that I started treating self-erasure like a reasonable cover charge.

But I've learned. And learning is worth more than pride.

Some forms of loneliness are cleaner than that.

The anger I encountered was rarely as coherent as it wanted to sound. A lot of it felt like anxiety dressed up as a position. The vocabulary changed from week to week. The need to make my joy look suspect did not.

Once I could see that, I could also see that I was not really being invited into a conversation.

That is a lonely feeling. To realize people would rather reduce your joy than examine their own panic.

Some people hit a certain age and start defending what they know. The world keeps moving and they pick a spot and dig in. I understand the temptation. To pick a hill. To call it wisdom. To let contempt do the work that courage no longer can.

I went the other way.

I still feel like a kid. There’s too much to learn and not enough time and I cannot believe I get to live through this particular moment in history. I get to be here while the walls move. While the instruments change shape in our hands. While entire creative grammars are being invented in public.

I've resisted saying so out loud because "special" got weaponized somewhere along the way, used sarcastically, as if delight were arrogance. As if appetite were a character flaw.

But I'm done apologizing for the configuration of brain I showed up with.

I thought I was fighting about technology. I was really learning to stop abandoning myself.

That is a much bigger, sadder lesson. It reaches backward and lights up half your life. The friendships you overwatered. The rooms where you performed being easy to understand. The little edits you made to yourself so other people would not feel judged by the fact that you were still hungry.

No more.

My aliveness is not negotiable.

Left alone, I move toward the work. Toward the manuscript. Toward the synth patch that still isn't right. Toward the sentence that keeps dodging me until I pin it down.

My resting state is not consumption. It is making.

That is not virtue. It is simply the shape of my hunger.

And I think, for a long time, I wanted somebody to tell me that hunger was good. I wanted somebody to see how hard I was trying to stay alive inside my own life and say, yes, that counts. Yes, I see you.

Maybe that is why the rejection cut so deep. Because underneath all the argument, there was a much more childlike ache:

Please do not make me feel ashamed of the thing that makes me feel most like myself.

That is who I am when nobody is clapping.

Dan Gable won Olympic gold in wrestling without a single point scored against him. The next morning, he was back in the wrestling room. A peer asked why he was training instead of celebrating.

*"My opponent is training today,"* he said. *"So I am training today."*

I'm not competing with you. But my craft has an opponent: entropy, irrelevance, the comfortable silence of never finishing anything, the soft anesthesia of almost. And that opponent does not take days off.

So neither do I.

Not because I am trying to defeat anybody. Because practice is how I remain real to myself. Craft is still waiting for me in the morning.

I am done mistaking approval for love.

I make things. The work will find who it is for.

I am done asking people allergic to becoming to bless my becoming.

And I've never been happier.

In 1980, David Byrne and Toni Basil went to UCLA and USC and watched archive footage of people in religious ecstasy. African tribal rituals. Japanese sects. Trance states. Southern televangelists speaking in tongues. They watched hours of it. Then Byrne went into a white room in a suit and bowtie and invented a way of moving that had never existed before.

https://www.youtube.com/embed/5IsSpAOD6K8

He did not copy them. He absorbed them. What came out was unmistakably him — twitchy, rigid, caught between dance and seizure — but it contained the whole archive. Basil said, "David kind of choreographed himself."

I think about that video when I think about what is happening to us.

Not because it is hidden. Because it is boring.

It arrives as a better autocomplete. A better search box. A chatbot that writes a decent email. A coding tool that saves you twenty minutes. An agent that can take a task off your plate. It arrives in little clerical bites. Dry. Procedural. Easy to dismiss.

But then, like _gravity_-- it's already here.

I am not writing this from a safe distance. I am not sitting in some observer’s box, pointing at the field and calling plays. I am in it. I am writing this as someone whose life depends on knowledge work continuing to exist in a form I can still do.

I am disabled. Physical labor is not waiting for me as some noble backup plan. My life has been built around thinking, writing, synthesizing, organizing, communicating, and making sense of complex things for other people.

Maybe I am wrong. I hope, in some narrow selfish way, that I am a little wrong. But I do not think I am. If my numbers are right, I am looking at losing my job too. Not someday in a dramatic robot apocalypse. In the ordinary way people lose jobs now: through efficiency, restructuring, shifting expectations, quiet math. Through the slow realization that the market value of what you do is changing faster than your body, your résumé, or your nervous system can keep up.

Musicians like to talk about what makes an instrument "deep."

You can tell how deep an instrument is by how differently people sound on it. A harmonica is lovely, but most people given one will circle the same few shapes. A piano is different. Give people time with a piano and they begin to diverge. Touch, rhythm, taste, phrasing, instinct. The instrument has enough depth to reveal the player.

That is what AI feels like to me now. Not a vending machine for generic content. A deep instrument. One that can sound cheap in clumsy hands and astonishing in practiced ones. One that reveals taste instead of replacing it.

For a long time, the story I told myself about work was simple. Get better. Learn more. Become more useful. Refine your taste. Refine your craft. Build skills nobody can take away.

I still believe in craft. I still believe in skill. But I do not think that story is enough anymore.

The question is no longer just, “How do I get better?”

The question is, “How do I get better at using systems that amplify me?”

And after that comes the question that scares me more: “What do I do when those same systems start replacing parts of me?”

At Acme Corporation (I always work Acme when writing on my own site), my boss gave me an AI agent connected to our knowledge bases, a week without responsibilities, and a task: "Automate away your job."

It took three days and a lot of caffeine, but I click-and-dragged my way to bespoke AI agents that did 80% of my work. Most of that time was learning the platform; if I'd had my own AI tooling to work with, I could have done it faster. But it _worked_.

All of this, done in my pajamas.

And it wasn't like the robot replaced me completely. I still monitor its outputs; adjust its prompt; and, when the time calls for it, step in to do the manual work myself.

But it was enough. My job is going away. The years of craft do not get a vote.

Not because someone wanted to save a buck. Because the machinery that made craft valuable is becoming _antiquated_.

Same as it ever was.

Lately I keep sorting AI use into four modes, mostly so I can think clearly.

AI as a replacement for Google. Ask. Retrieve. Summarize. Mostly incremental.

AI as a chat partner. Brainstorming, feedback, loose collaboration. This is where people start calling it magical.

AI as a tool. The model is no longer a conversation you visit. It becomes a component in the work itself — touching files, context, the actual thing you are building.

AI as orchestration. Here you are not fetching outputs. You are the director: deciding what the system touches, what order things happen in, what gets reviewed, what gets shipped, what stays human. You are not doing the work. You are deciding how the work gets done — and that decision is now the work.

I think a lot of us are going to end up functioning like the executive team of our own tiny company. Not because we dreamed of becoming entrepreneurs. But because our “company” is now our output, and our job is to manage how that output gets made.

What do I keep. What do I delegate. What do I veto? What actually sounds like me?

What is mine, and what belongs to the machine?

That was the question where the floor dropped out for me. Once you see that, you cannot unsee it. This is not just a better interface. It is a new way of working. Maybe even a new way of being valuable.

AI isn't real yet. Not real in the mythic sense. Not the spaceship computer people imagined. Not coherent, reliable, or fully inhabited by culture yet.

And still: it is socially transformative anyway. It still hallucinates. It still struggles with counting Rs in "strawberry."

It's still slop. But children will make it real.

Not because adults adapt. But because children, free from the burden of experience, will find ways to breathe life into the tech.

Same as it ever was.

People want this conversation to stay abstract. They want it to live up in the cloud layer, where we can debate "the future of work" like it is a panel discussion. But this is not abstract. This is rent. This is medication. This is dignity. This is the question of whether the kind of person I am still has a place in the economy that is arriving.

What happens when labor gets cheap?

What happens when more and more forms of cognitive work become abundant, automatable, and easy to accelerate? What happens to the people who built lives around being good at exactly those things? What happens to those of us whose bodies already narrowed the field of possible work long before AI arrived? What happens when the culture still expects us to earn our keep, but the definition of valuable work is changing faster than we can metabolize?

I do not have a neat answer to these.

I wish I did. I wish this essay ended with a clean little rallying cry and a branded takeaway and some polished note about adaptation. But what I have, if I am honest, is fear. Not only fear. But fear, yes. Grief too. And a strange kind of awe.

Because I can feel two truths at once.

One is that these tools are extraordinary. I have felt my own powers stretch because of them. I have seen what happens when a person with taste and drive and a point of view gets access to systems that can amplify thought. It is real. Something new is possible here.

The other truth is that possibility does not guarantee mercy.

A tool can be miraculous and still arrive inside a system that chews people up. A breakthrough can still leave casualties. Increased abundance can still make individual lives more precarious before anything like justice catches up.

That is the truest thing I can say right now: I do not know where this lands, and I am afraid of the answer. There is no simpler version of me waiting in the wings, no easier fallback, no clean retreat into some safer economy. But I am still here. I am still learning. I am still trying to understand the new instrument before the room changes around me completely.

Maybe that is what survival looks like at first: choosing to keep your eyes open while the children, growing up inside these systems, make them real.

Same as it ever was.

After Scarcity

If campfire logic describes how creative culture survives, what kind of economy can support it now that scarcity is collapsing?

The market model is dying. It is not being murdered. It is encountering a condition it was never built to survive.

That condition is abundance.

---

Here's what the first three essays built. AI is rabies for capitalism: the system is thrashing because scarcity, its core assumption, just broke. The tension around AI echoes what photography triggered in painting, what the printing press triggered in manuscript culture. The line moves. The community adapts. The art keeps getting made.

The lone genius myth was invented to serve the art market. Not artists. The workshop model predates it by centuries. Attribution was a guild convention. "Who made this?" became the only question that mattered when dealers needed scarcity to justify prices. Preciousness is a market strategy, not a creative value.

Campfire logic is how creative culture survives. Pass it around. Let it change shape. The song belongs to the circle. Fan communities, folk traditions, jazz musicians, workshop artisans, con sketch artists: they all knew this. Museum logic replaced it, not because it was better, but because it was profitable.

Now the profitable model is breaking. Not because anyone attacked it. Because the tools changed.

---

You cannot enforce scarcity on infinite reproducibles.

This is not a political argument. It's a structural one. The gallery system works because paintings are physical objects. There is one original. Copies are identifiably inferior. Provenance can be verified. The recording industry worked because distribution had chokepoints. Pressing vinyl costs money. Bandwidth costs money. Someone controls the pipeline.

AI removes the chokepoint. Generation is instant, free, and runs on consumer hardware. The cost of producing an image dropped from hours of skilled labor to seconds of electricity. The cost of producing a story, a song, a design: same trajectory. You can legislate around this. You can platform-policy around it. You cannot reverse it. The capability exists. It will not un-exist.

The commission economy as it stands is built on the same scarcity assumption. This artist's style is unique. Their time is limited. Their output is finite. That creates value. But when a model can approximate any style on demand, the scarcity of style evaporates. The artist is still real. Their skill is still real. But the market mechanism that converted skill into price just lost its floor.

This is not a moral failure. It's a market failure. The system was designed for a world where creation was expensive and distribution was controlled. That world is over.

---

So what replaces it?

Not nothing. The art doesn't stop. It never has. The question is what economic and social infrastructure we build around it now that the old infrastructure has lost its foundation.

Campfire logic points the way. Not as nostalgia. As architecture.

Gift economies are real. They predate market economies by millennia. They work on reciprocity, not transaction. I make something. I share it. You make something back. The value circulates through the community, not up to a platform. The convention artist alley already works this way. The five dollar sketch is not priced to reflect the artist's skill. It's priced to keep the exchange moving. The relationship matters more than the receipt.

Patronage is real. Not the Medici kind. The modern kind. Patreon, Ko-fi, tip jars, community subscriptions. The model is: I support your practice because I value what you do, not because I'm purchasing a specific deliverable. This works. It works right now, for thousands of creators, across every medium. It's not a utopian fantasy. It's an existing parallel economy running alongside the market model.

Community patronage is real. A Discord server that pools funds for a shared artist. A fandom that crowdfunds a zine. A convention that subsidizes artist alley tables. The infrastructure exists wherever people decide that supporting creation matters more than owning the output.

The con sketch is real. Eight minutes. Ten bucks. A badge that means more than a gallery print. The value was never in the object. It was in the moment, the connection, the fact that someone sat across from you and made something that only exists because you were both there.

---

I'm not naive about this. And I'm not sneering at artists trying to survive.

Working artists face real economic pressure. Rent is due in dollars, not reciprocity. The transition from a scarcity market to an abundance economy is not painless, and pretending otherwise would be dishonest. Artists who built careers on commission income are watching their pricing power erode in real time. That's not abstract. That's groceries.

But the survival model was already brittle before AI showed up. Platform dependency, algorithmic visibility, race-to-the-bottom pricing, clients who ghost after the sketch phase. The commission economy was never a stable system for most working artists. It was a hustle that worked for some, failed quietly for many, and enriched the platforms that brokered every transaction. AI didn't break a thriving ecosystem. It accelerated the collapse of one that was already buckling under its own contradictions.

The solution is not to pretend AI doesn't exist. It's not to legislate scarcity back into a system that has structurally moved past it. Every attempt to enforce artificial scarcity on digital goods has failed. DRM failed. Paywalls leak. Copyright enforcement online is a game of whack-a-mole that platforms play only when it's cheaper than not playing.

The solution is to stop organizing creative culture around a market model that was always extractive and only ever worked for a small elite. The gallery system served gallerists. The recording industry served labels. The commission economy, at its best, served individual artists. At its worst, it created a race to the bottom where artists competed on price for algorithmic visibility on platforms that took a cut of every transaction.

That system was not sacred. It was recent, and it was breaking before AI showed up.

---

The art keeps getting made. It always has.

Before the market. Before the gallery. Before the copyright office and the recording studio and the platform economy. People made things and shared them and the things changed hands and changed shape and nobody called it theft. They called it culture.

The workshop foreman didn't panic about apprentices learning his techniques. The campfire musician didn't sue the next person who sang the tune. The zine maker didn't DRM the photocopier. Creation flowed because flow was the point.

AI didn't build the campfire. Fan communities have been tending it for decades. Furry artists passing around badges at 2 AM. Fanfic writers remixing each other's worlds. Modders rebuilding games from the inside out. Zinesters stapling pages in living rooms. This is the oldest creative tradition in the world, running on the oldest creative logic, and it never stopped. The market just talked loud enough to make us forget it was there.

The market is the experiment. Three centuries old, built on manufactured scarcity, designed to serve collectors and dealers and platforms. It had a good run. It made some people rich. It convinced the rest of us that art without a price tag wasn't real.

Campfire logic is the default. It was here first. It will be here after.

The tools changed. The fire didn't. Sit down. Make something. Pass it on.

Campfire Logic

If the last essay argued that preciousness is market logic, this one asks: what does a non-market logic of art actually look like?

Before the recording industry, the idea of a "wrong" version of a song barely made sense.

A tune was not a product. It was a process. You heard it. You learned it. You changed it. You passed it on. The next person changed it again. Nobody owned the melody. Nobody had the definitive recording. The song was a living thing that mutated every time someone new picked it up.

We didn't lose this by accident. We professionalized it away.

---

Western music notation promised precision. Write down every note, every rest, every dynamic marking, and you could reproduce the composer's exact intention across centuries. That was the sales pitch. What it actually did was create the idea of a "correct" performance. A version that matched the score. Interpretation became deviation. Deviation became error.

The Victorians loved this. Same people who [split art into "high" and "craft"](https://www.vam.ac.uk/articles/arts-and-crafts-an-introduction) split music into "classical" and "everything else." Classical meant composed, notated, performed as written. Everything else meant folk, popular, improvised. Lesser. The hierarchy was never about complexity. Jazz is more harmonically sophisticated than most classical repertoire. It was about control. Written music could be owned, published, sold. Improvised music belonged to the room it was played in.

Then came recording. Then copyright. Then the entire legal and economic infrastructure that treats a song as a fixed object rather than a living pattern. Now there's an original. Now everything else is a cover, a sample, a derivative work. Now the lawyers get involved.

---

Here's what we had before all that. I call it campfire logic.

Campfire logic works like this. Someone starts a song. Someone else joins in. They change the words. They add a harmony. They forget a verse and make up a new one. The song belongs to the circle, not the singer. It changes shape in other hands. That's not a flaw. That's the feature.

This is not a folk tradition. This is not a primitive precursor to "real" music. This is how music worked for most of human history, across most of the world.

The musician interpreting a tune is doing the same thing Da Vinci's [workshop artisans](https://www.worldhistory.org/article/1611/life-in-a-renaissance-artists-workshop/) were doing. Taking a vision. Making it real through their own hands. Adding their own skill and judgment to someone else's framework. The workshop model and the campfire model are the same model. Collaborative, iterative, non-precious. The product is the process.

---

Museum logic replaced campfire logic. Museum logic says: one original, everything else is reproduction. One author, everyone else is audience. One correct version, everything else is derivative.

Museum logic needs walls. Literal ones. The painting goes behind glass. The manuscript goes in a case. The recording goes into a format you pay to access. The walls keep the art scarce and the access controlled and the price stable.

Campfire logic has no walls. The song escapes the circle. The story changes in the retelling. The sketch gets photographed and reposted and remixed and tagged and lost and found. Nothing stays fixed. Nothing stays owned. Nothing stays precious.

Fan communities have always run on campfire logic. Fanfic runs on campfire logic. So do mods. So did zines. So did the entire pre-copyright folk tradition that kept human creative culture breathing for millennia before anyone thought to put a price tag on a melody. Every community that circulates art faster than it can be owned is a campfire community.

---

Creative communities have always known this, even when they couldn't name it.

The open-source codebase is a campfire. Someone writes a function. Someone else refactors it. Someone forks it and takes it somewhere the original author never imagined. The code changed hands and changed shape and that was fine, because campfire logic says the point is not to own the flame. The point is to keep it going.

The convention artist alley is a campfire. The Discord server where someone posts a sketch and someone else colors it is a campfire. The jazz club where a standard gets reinterpreted every night is a campfire. The value is in the exchange, not the object. In the process, not the artifact.

AI makes this tension visible. Running campfire logic and museum logic at the same time has always been a contradiction, but it was one most communities could leave unresolved. When the tools change fast enough, the contradiction surfaces. A community built on circulation has to decide what circulation means when the tools of circulation evolve.

The choice was always there. AI just made it harder to defer.

---

Campfire or museum. Participation or ownership. Pass it around or lock it up.

If campfire logic is how creative culture has survived and spread for most of human history, if it's how fan communities function right now, if it's the default mode of every creative tradition that wasn't industrialized into a product, then why are we building legal and platform infrastructure designed to kill it?

That's not a rhetorical question. The next essay has an answer.

My mentor taught me to write in theme first. "Theme is the engine," he said in his gravelly accent, the sound harmonizing with the hum of his MG on a country Tennessee highway. Around us was nothing but open road, classic rock, and screeching cicadas in every direction.

I drove seven hours to meet a fellow writer. To talk about writing. To be a better writer.

"Everything has to hum along. Every conflict rhyming with the next. All the conflicts point to one thing; one wound; one ache. One burning human truth.

"You have to have that ache, Austen. One singular, gaping ache explained in too many words. Otherwise, the gears catch and the spell breaks."

Sometimes, it's good to throw something away.

Not because it was bad. Not because it failed. But because the cost of trying again is falling so fast that clinging to the first version has become its own kind of trap.

What used to take hours now takes seconds. A terminal theme in a language you don't write. A task management system shaped to the weird geometry of my brain. The interface itself has become disposable. Just another layer I can reshape, discard, and rebuild until it fits.

Given the option, I'm beginning to throw away a lot of things I thought were sacred.

AI slop convinced me that no one gets out of this transformation. Not because it's replacing anything human. The opposite — it lowered the friction between what I wanted and what I could make real. It gave me a way to stop accepting default interfaces and start asking what actually fits.

The question isn't whether this changes your work. It's when you decide to stop waiting and see how far the new tools will take you.

I spend a lot of time at a Kawai NS-15 piano. It's a gorgeous thing, refurbished with love by the folks at Pianoforte Chicago. It has a bright, almost soulful sound that plays just as well in pop rock as it does the cryin' blues.

My wife's playing Persona right now. I used to play a lot of video games, but, well. Losing my vision made that too hard. Now, I make a game of learning all the songs from their sessions by ear.

I know I'm getting better. The notes come easier. The tune flows smoother.

I like staying in the moment. There's something about the present that feels permanent, untouchable. That if I have food in my belly and music in my life, I can be happy.

Even if AI changes everything, I still can find myself here, always here, forever frozen in the present.

We are still carrying a world-model built for expensive attempts. In that world, every draft mattered because every draft cost something: time, money, expertise, access, energy. You accepted the shape things came in because changing them meant friction. You learned to live with software that almost fit. You tolerated systems that made you smaller because rebuilding them felt impossible.

That assumption is breaking.

The realm of the possible is expanding faster than our brains can comfortably comprehend. Not in some abstract sci-fi sense, but because in the deeply human rhythm of everyday life the tiny points of resistance that used to define the edges of what you could do alone are falling away, and for those that take the plunge, they realize their capabilities know far fewer bounds than they did a few years ago.

Thinking is no longer a person problem: it's a compute problem. It's patterns of patterns of thinking, stacked atop of each other like subroutines, meta-problems to be solved.

The future isn't soon. The future already happened. The world shifted. Feel the quickening pace of the dance floor? The mania that hides deep in your bones? The feeling that things are burning up, hotter, faster, stronger, the belly of the best baying for blood? That's the beat, baby, rushing faster, faster, ever faster, like jazz in the fog of a nuclear sunset.

This is no longer a slow, steady march of progress. This is a gold rush, and the steam engine, and twenty thousand digital revolutions all combined into one gigantic "fuck-your-feelings" hockey-stick graph that says The Cost of Idea Execution Just Hit Zero, and Nobody is Ready.

How could I possibly feel joy without the work trophy? Or the next big unlock for your skills? "If I don't have to struggle to do something, what's the point?" The point is abundance, y'all, endless abundance of shitty ideas for as far as the eye can see

When you're disabled, you learn to say yes a lot.

See, I grew up in the bitumen of the social internet — the IRC days of roleplay mucks and writing collectives that fed the early fandom community. I know some fuckin' nuts.

It's 2012 and I'm a young, confident trans girl in my twenties. I'm in the best shape of my life, walking around a hotel full of mascot costumes in Philadelphia. I'm meeting an old friend I wrote with for many years.

I get to the room. "The Cubhouse." I think there must be some mistake. There are baby animal avatars of the guests taped to the door, and sure enough, there's Kit.

Kit's a babyfur. The hotel room had been converted into an adult nursery. Grown men in onesies and animal ears, all watching Care Bears. In the background, a comically large crib had been constructed for intricate scenes.

So I put my bullshit aside, said "yes," and watched Care Bears with a bunch of dudes in onesies. It was awesome. They had a warm, childlike energy. Safe. Kind, in a way that told the tale of the scars on their own backs.

I still think about it a lot. That in that moment, far from my one-stoplight hicktown upbringing, I felt safe and seen for the first time in a long time.

Consider the social internet: Discord chats, Telegram channels, the "group spaces" of the modern internet. I used to spend a lot of time in these, looking for something. Belonging, maybe. Out of a fear that I would belong nowhere else.

I got better, though.

It didn't occur to me until I wasn't in them anymore, but the internet expanded the boundaries of my house in ways that didn't always feel right. I had invited roommates into my space 24/7. They were guests at my dining room table, the first thing I said hello to when I woke up. I went to bed dreaming of Instagram dachshunds splaying out on linoleum floors, baying for my attention one last time before nodding off to dreamland.

Humans weren't meant to be stretched this widely.

We were meant to be smaller than this.

We were meant to be imperfect, wild, careening storytellers making shadows on the cave walls with our campfires — not because it drives engagement, or will help you "make it big." We were meant to stretch each other. To make things because they entertain our neighbors, or because they feed our friends and family.

Because being a human is fun, guys. How the fuck did we forget that?

I'm in a band now.

I'm in a band now.

We play covers, sure, and it's a learning band. But it's a band. I play music for other human beings now.

I feel like I've walked in Carole's shoes. I've had one too many with Ben and banged on the keys far too late in the evening. And now — just a bit! — I can feel my own voice forming. Something resembling arrangement on the fly.

I'm learning to love the instrument I've put all this work into. It's not an easy instrument to love — she's quite demanding, actually! — but after a lifetime of strife we're finally getting along.

Maybe I'll start writing songs, dog help us all.

Even the way I write changed.

I learned in a very particular style; that of the email listserv message. I remember calculating out how many words I could fit into a single segment before my email got rejected as spam. (4000 words, for the curious.) I learned to start from "Once upon a time" and finish with "the end," and between those two points I grabbed the bull by the horns and held on for dear life. It's part of why my work is so cinematic; I had to grab readers anew every 3000–5000 words.

I worry less about perfection now. Perfection is easy when you have a second brain. Raw is in.

And maybe I'll collect these logs, eventually, into something that means something. A book. A memoir. Something to say "This is what living through this was actually like."

Someday.

But I keep coming back to something simpler than that. Something embarrassingly personal.

Phil's lessons weren't really about the theme. They were about the human part of writing.

When I write, the actual words use rarely matter. It's their rhythm. Their feel. I imagine them like notes that I play in rhythm, building a scene like a sax player sets a mood. If they're AI, or if they're human, they can still sing in tune.

The thing that makes a writer a writer is something more sinister: it's that the best writing involves placing your own beating heart onto the page, pain and all, and describe just how much your soul aches for it. How watching yourself bleeding out reminds you of a funny joke you heard one time.

There's a condemned man whose noose is too tight. As the Judge reads his sentence, he nudges the hangman.

"Can you loosen this?" he asks.

The hangman laughs. "Why?"

"I don't want to choke and distract the man," he says, nodding to the Judge.

My brain is still adjusting to a world where failure is this easy to overcome. Where the cost of trying again has fallen so far that clinging to the first version — of anything — has become its own kind of trap. I spent a long time learning to live carefully inside constraints. Calculating drafts. Saying yes to cribs and Care Bears because the alternative was nothing.

The alternative isn't nothing anymore. It's detonating the space between idea and execution and trying things. Not because they're necessary. Not because they make sense. But because they're fun.

Because we're humans, and we make stuff. It can be as simple as that.

I lied. I wrote the joke myself.

The theme is that ache.

Otherwise, the spell breaks.

The Preciousness Problem

The first essay in this series argued that AI is rabies for capitalism: that the system built on scarcity is thrashing because its core assumption just broke. This essay asks a related question: where did our obsession with singular, precious authorship come from in the first place?

Nobody painted the Mona Lisa alone.

That sentence bothers people. It shouldn't. Leonardo da Vinci [ran a workshop](https://www.worldhistory.org/article/1611/life-in-a-renaissance-artists-workshop/). He had apprentices, assistants, specialists. Someone mixed pigments. Someone prepared panels. Someone filled in backgrounds while the master worked on hands and faces. This was not a scandal. This was Tuesday. Every major Renaissance painter operated the same way. Raphael's workshop produced so much work under his name that [art historians still argue](https://www.euronews.com/culture/2024/01/05/ai-study-suggests-famous-raphael-painting-was-not-entirely-his-own-work) about which brushstrokes are "his." The answer, for most of the history of art, was that nobody cared.

The workshop model was the default for centuries. Art was a craft. You learned it through apprenticeship, the way you learned carpentry or metalwork. In Florence, painters belonged to the [Arte dei Medici e Speziali](https://www.italianrenaissanceresources.com/units/unit-3/essays/guilds-arti/), the guild of physicians and apothecaries, because apothecaries supplied pigments. The guild system governed who could practice, what materials they used, how they trained. The master's name went on the piece because the master ran the shop. Not because the master's hand touched every square inch of canvas. Attribution was a business convention. Not a moral claim.

---

Patrons didn't buy genius. They bought results.

The Medici commissioned frescoes the way a modern company commissions a website. Here's the wall. Here's the subject. Here's the deadline. The artist's job was execution within constraints. Creative freedom existed, sure. But the idea that an artwork sprang fully formed from one person's singular vision, untouched by commercial pressure or collaborative process? That would have gotten you laughed out of Florence.

The shift happened later than people think. The 17th century planted the seeds. The Romantic movement watered them. But the real transformation came with the [rise of the modern art market](https://www.britannica.com/money/art-market/The-18th-century) in the 18th and 19th centuries.

Dealers needed a story. Collectors needed provenance. The gallery system needed a reason to charge what it charged. By the mid-19th century, [speculation had become integral to the art market](https://www.19thc-artworldwide.org/spring20/baetens-on-artist-dealer-agreements-and-the-nineteenth-century-art-market) and dealers signed contracts with artists before their rise to fame, building reputations to justify prices. The reason they all landed on was authorship. Singular, verifiable, individual authorship. "Who made this?" became the question that determined value. Not "is this beautiful?" Not "does this serve its purpose?" Not "does this move you?" Just: whose hand? Whose name? Whose genius?

---

The lone genius myth is an economic invention.

That's not a metaphor. The art market required scarcity to function. A painting by Vermeer is worth millions because there are only [thirty-four of them](https://www.essentialvermeer.com/how_many_vermeers.html). If Vermeer had run a factory, if his apprentices had produced hundreds of works in his style, if attribution was fuzzy and provenance uncertain, the price collapses. The market doesn't work without the story of the individual creator whose output is finite and whose touch is irreplaceable.

So the market built that story. And then culture absorbed it. And then we forgot it was a story at all.

The Victorians finished the job. They [codified the split](https://www.vam.ac.uk/articles/arts-and-crafts-an-introduction) between "high art" and "craft." Painting: high art. Pottery: craft. Sculpture: high art, unless it was functional, in which case: craft. The distinction had nothing to do with skill or beauty or cultural significance. It had everything to do with which objects could be sold as unique expressions of individual genius and which ones couldn't.

Craft was collaborative, reproducible, functional. High art was singular, authored, useless. The uselessness was the point. It proved you were paying for the idea, not the object. For the name, not the thing.

---

The line moves. It always has, and it applies here too.

Every time a new tool makes creation faster, cheaper, or more accessible, the market panics. Not because the art gets worse. Because the scarcity gets harder to maintain. Photography threatened painting. Recorded music threatened live performance. Digital tools threatened the commission economy. Each time, the cultural response was the same: that's not real art. That's mechanical reproduction. That's cheating.

The argument was never about quality. It was about control. If anyone can make an image, the image loses market value. If the artist's hand is no longer required for every pixel, the story of singular authorship breaks down. And if that story breaks down, the entire pricing model collapses.

This is what the AI panic is about. Not authorship. Not ethics. Not the sanctity of human creativity. The market's need to maintain scarcity in a world that just became infinitely reproducible.

---

Art is upstream of culture. We take in the world and return two outputs: meaning, and therapy bills. The artists go first. They absorb the contradictions early, sit with them longer than is comfortable, and turn them into something you can look at without flinching.

By the time it reaches everyone else, it feels like a trend. A vibe shift. A “sudden” change in what people care about. It isn’t sudden. It’s sediment. Layers of feeling, stacked and compressed until they break the surface all at once.

What’s coming next won’t arrive as policy or product. It will arrive as a feeling you can’t quite name yet, and then, all at once, everyone will.

---

I'm not an art historian. I'm a practitioner. I make things with tools that would have horrified the Victorians and confused the Medici and made perfect sense to a Renaissance workshop foreman.

Da Vinci directed artisans the way a film director runs a crew. The foreman told the apprentice which colors to mix. The patron told the foreman what to paint. The whole system was collaborative, hierarchical, and nobody pretended otherwise. The genius myth came later. It came from the market. And it came with a price tag.

When the argument against AI art centers on "a human didn't make every decision," it's worth asking which tradition of art that standard comes from. Not the workshop. Not the guild. Not the folk tradition. It comes from the market, the one that needed singular authorship to justify its prices. The concern is understandable. But the framework it rests on is newer and narrower than people realize.

The workshop knew better. The guild knew better. Every culture that treated art as craft instead of commodity knew better. The preciousness was never in the object. It was manufactured by the market that sold it.

And now the market is scared. Because the tools just got cheap enough that the story stopped working.

Consider the frothing grimace of a raccoon in your headlights. Eyes like wet glass. Jaw working on nothing. Something in your lizard brain says *wrong* before you can name why.

If you have symptoms of rabies, you're already dead. You just don't know it yet.

AI, in an ironic twist of fate, has made capitalism so unstable that it now fights anything that gets close to it.

I'm not here to argue that the free market will transform and survive. To do so requires medicine we don't have; and furthermore, why stop a good thing? Indeed, I believe we are dealing with a rabid market that is wholly unable to understand that the cost of effort and craft just crashed through the floor, making us all, as Neal Stephenson once said, "As rich as a Pakistani bricklayer."

For millennia, the currency that powered society was labor.

Someone needed to make the butter to feed everyone. Someone else needed to make the guns to protect everyone. The sum of all human trade systems, from the ancient rite of kings to modern market-based capitalism, has been about the division of labor.

Humans are primates with more complexities and depression. Our needs must be met: we must eat, sleep, feel safe. Maslow covered that ground for us already.

The other thing about people is that we're absolute shitgoblins when it comes to doing things equally. We find ways to distinguish "value" — cleaning a toilet? Not valuable. Building Facebook? One of the most valuable things a human can do in capitalism.

And it's been a good system, overall. It took us to the moon. It ushered in an industrial revolution that gave us the highest standard of living human beings have ever known. It gave us iPhones and arguments about pizza and, for a while, a stable social order.

Has it been equal? No. But it does get results.

AI breaks a core assumption of the market: that labor is necessary and, furthermore, expensive.

A rabid animal does not go gentle into that good night.

A mammal affected with rabies becomes compelled to do everything in its power to spread the virus as effectively as possible. Hydrophobia causes a concentration of the viral load into the leftover saliva, increasing the chance of infection from bites. Hallucination and aggression, caused by the virus eating your prefrontal cortex, encourage the mammal to fight against anyone who tries to help them.

It thrashes.

It bites.

It infects.

A rabid animal doesn't know it's dying. It just knows something is wrong and it needs to bite harder.

That's what I see when I look around.

Platforms locking down APIs. Studios suing over training data. Publishers tightening licensing on works that weren't selling anyway. Companies laying off thousands while posting record profits. Everything getting more expensive while the cost of production craters.

There is no strategy. There is only rabies.

Here's the thing nobody in an economics class will tell you:

Incentive structures don't update in real time.

The people running the systems — the executives, the boards, the institutional decision-makers — they were trained in a world where scarcity was the law. Where you got rewarded for locking things down. Where "moat" was the highest compliment a business model could receive.

They're still rewarded on timelines that pretend the floor didn't drop.

So when the ground shifts under them, they don't pivot.

They double down.

More extraction. Tighter control. Higher walls. Bigger legal teams.

Not because it works. Because it's the only move they know.

I'm a writer. I've watched the publishing industry for years.

For a brief moment on the internet, having a keyboard and something to say could make you a living. I had many friends who made rent in the proto-internet era as journalists, content writers, sensitivity readers, podcasters, chief-happiness-officers, the whole nine.

Cory Doctorow figured out the model before most people realized the old one was breaking — free access, supported by promotion and special editions. "Art should be free and accessible." He was right. But even the correct answer from ten years ago is about to get stress-tested in ways nobody planned for.

The Pattern Is Always the Same

Every industry does this. Every single one.

Same script, new letterhead. I've watched it often enough to put numbers on it.

1. The cost of production drops

2. Incumbents tighten control instead of adapting

3. Extraction increases even as value delivery decreases

4. Small players route around the walls

5. The walls get higher

6. The routing gets easier

7. Eventually the walls don't matter

We're somewhere around step 5 right now. Maybe step 6.

This Is Why Everything Feels Predatory.

You ever notice how every service you use got worse at the same time?

Streaming services raising prices and adding ads. Social platforms choking organic reach to sell it back to you. Banks charging fees on money they're already lending out at interest. Insurance companies whose entire business model is finding reasons not to pay.

It's rabies, friend. Rabies all the way down.

When your moat starts draining, you don't build a better castle. You charge more for the drawbridge.

Here's where it gets ugly.

Rabid systems don't just extract more. They start eating their own.

Companies laying off the people who build the products. Platforms banning the creators who drive their traffic. Industries lobbying for regulations that protect incumbents at the cost of the market they depend on.

It's auto-immune. The system attacks itself because it can't tell the difference between a threat and its own body anymore.

Here's what gives me hope: Rabies is terminal.

Not in a dramatic, overnight, everything-collapses way. But in a directional, structural, the-math-stops-working way.

You can charge more for the drawbridge. But if someone builds a bridge next door for free, eventually people stop crossing yours.

You can lock down your API. But if someone builds an open alternative, your lock becomes your prison.

You can sue over training data. But you can't sue physics. You can't sue math. You can't sue the fact that the cost of intelligence is plummeting and no licensing agreement on earth can reverse the curve.

"Okay, capitalism is breaking. What replaces it?"

If I'm going to answer it straight:

What comes after isn't a new system. It's an old one. Before the recording industry, before copyright, before the walls — there was campfire logic. Someone starts a song. Someone else changes the words. The song belongs to the circle. That's not a folk tradition. That's how most of human creative culture worked for most of history, before someone figured out how to put a price tag on a melody. The rabid market built walls around the campfire. The campfire didn't go anywhere.

Look at the discourse.

"AI is stealing jobs." "AI is plagiarism." "AI is the end of creativity."

Those aren't arguments. Those are bite marks.

That's the sound of a system that built its entire value proposition on scarcity realizing that scarcity is evaporating — and lashing out at the thing making it evaporate instead of asking the harder question.

Some of us are worth a lot. Our taste, our judgment, our ability to see what matters — that doesn't go away when production gets cheap.

It augments.

It supercharges.

It compounds.

And some of us were only worth what the gate charged.

Sit with that one for a second.

What do I do about it?

I build.

I build for a world where access isn't the product. Where my ideas are worth their weight in gold because the barrier between idea and execution is a couple of markdown files.

Where value comes from what you *do* with the tools, not from controlling who gets to use them.

Yeah, I know how this sounds.

I sound like I'm writing a manifesto in my underwear.

And honestly? I kind of am.

But I'm also the person who built their own website, wrote their own editing system, hired AI agents to help them ship, and is currently watching the economics of everything they thought they understood rearrange themselves in real time.

I'm not theorizing.

I'm taking notes while the building shakes.

The system isn't dying gracefully.

It's going rabid.

More extraction. More control. More walls. More lawsuits. More fees. More friction. More of everything that used to work, applied harder, faster, and more desperately.

And on the other side of all that thrashing?

People are quietly building things. Shipping things. Creating at a scale that makes the walls irrelevant.

The rabies will run its course.

Nobody issues a memo when the circle wins.

It's a million people who stopped waiting for permission — building the future in the gaps between the bite marks.

I've had more names on the internet than I've had prescription strengths.

That's not a joke. Sly Squirrel came first — a high schooler posting furry fiction and somehow landing an Ursa Major nomination before graduation. Then Sly Rabbit. Slyford T. Rabbit. That's where I actually learned to write, chapter by chapter on ANTHRO, figuring out what a novel was by publishing one in public. Then came Carl — picked up for a community anthology and a podcast episode — and that opened the door to Bait and Switch getting published for real.

Then I fell off the map for ten years.

Not off the page, though. Never off the page. I just stopped publishing. The writing kept going — I'm sitting on ten novels right now, with more coming. But the public-facing part? The putting-yourself-out-there part? That went quiet.

Now I'm back. One name. One site. No masks.

---

The Old Way Was Push

For fifteen years, my online presence — when I had one — was push-based. Write something, post it somewhere, hope the right people saw it before the platform folded.

Different platforms, different audiences, different names dropping content into feeds and walking away. The furry fiction on FurAffinity. The queer essays on Bilerico. The tech writing wherever tech people gathered that year. Never in the same place, never under the same name, never with the through-line visible.

It worked, for a while. The Ursa Major nod happened. Carl got picked up. Bait and Switch found its readers. But every one of those moments happened *despite* the strategy, not because of it. I was scattering seeds across a dozen fields and acting surprised when a few of them grew.

And the fields kept disappearing. Bilerico is gone. The forums where I first posted Bait and Switch chapter by chapter? Gone. ANTHRO, where I learned to write in public? Gone. Years of work, hosted on infrastructure I didn't own, couldn't back up, and couldn't redirect.

That's what push gets you. You're always renting attention from someone else's platform, and you're always one algorithm change or shutdown away from starting over.

---

The Mask Thing

Here's the part that's harder to write.

Having multiple names wasn't just a platform strategy. It was a mask. Several masks, actually, each one showing a version of me that felt safe in a particular context. The furry writer. The queer journalist. The software engineer. The novelist. All real, all me, but carefully separated — because the world is easier to navigate when people only see one dimension at a time.

I know something about masks. I have DID. The experience of presenting different selves to different contexts isn't metaphorical for me — it's neurological. And for a long time, the compartmentalized online presence felt like an extension of that: natural, protective, necessary.

But there's a difference between dissociation and discretion, and somewhere along the way I lost track of which one I was doing. Was I protecting myself, or was I just afraid that the whole picture — queer, low vision, neurodivergent, furry, novelist, engineer, parent — was too much for any single audience to hold?

The answer is both. And the follow-up question is: does it matter?

I'm forty. I've published novels. I've written over a hundred pieces of queer journalism. I've been doing this long enough that the fragmented version of my online self isn't protection anymore — it's just confusing.

So the masks come off.

---

Pull Means Owning the Ground

This site — thearcades.me — is the opposite of the old way. It's not a broadcast tower. It's a building.

The difference matters. Push says: here's my content, go find it in your feed between an ad and a political argument. Pull says: I built a place. It's here. The address doesn't change. Come by when you want.

I'm not quitting other platforms. But I'm done treating them as primary. LinkedIn, Bluesky, whatever comes next — those are hallways, not homes. They're where I point people toward the door.

Here's the thing, though. This isn't just about practicality. I'm tired of chasing the market economy on my own creative work. I don't believe the platforms are stable. I watched the places that hosted my early career vanish one by one, and I watched the ones that survived get strip-mined by engagement algorithms until the writing didn't matter anymore — only the performance of writing mattered.

Given the option to build my own thing, and given I have the skills to do it, I choose full ownership.

The platforms can do whatever they want. I'll be over here.

---

Building Backwards

Most people build an online presence by starting with social media and maybe eventually getting around to a website. I'm doing it the other way.

Website first. Blog with RSS. Email if you want it. The social profiles point here, not the other way around.

This feels backwards to anyone who's been marinating in growth-hacking advice for the last decade. Where's the content calendar? Where's the funnel? Where's the engagement strategy?

It's right here: I'm going to write things that are worth reading, and put them in a place that I own, and trust that the people who care will find their way here. That's the whole strategy.

I've been a working writer for fifteen years, even when the public couldn't see it. I know my readers. They're not scrolling LinkedIn looking for thought leadership — they're the kind of people who still use RSS readers and check bookmarked sites and follow actual links in actual emails. They want the work, not the performance of the work.

So I'm building for them. Pull, not push.

---

What's Here

If you're new: welcome. Here's what you'll find.

I write speculative fiction — novels where identity is literal, physical, political. *The Two-Flat Cats* is about four queer toons buying a house together the night before a fascist inauguration. *Bait and Switch* is a YA civil rights allegory with cartoon physics. *The Painted Cat* is weirder than I can summarize in a sentence. And there are ten more where those came from — a decade of quiet work that's finally coming out of the drawer.

I also write about the tools I use and the workflows I've set up. I run a multi-agent editorial team through Paperclip to help me revise my manuscripts — because I'm low vision and neurodivergent, and the traditional editing process was a wall I couldn't climb. I'll be writing about that workflow here, because I think other writers should know it exists.

And sometimes I'll write about the stuff in between. The furry community. The queer writing world. What it's like to be a disabled creator who got tired of waiting for someone else to build the infrastructure and decided to set it up themselves.

All of it, one name, one place.

---

The Invitation

This isn't a launch announcement. There's no product. There's no newsletter signup with a free PDF. There's just a writer who finally stopped splitting themselves across five platforms and ten names and decided to stand in one spot and say: I'm here.

If you've read my work before under any name — Crowder, Sly Squirrel, Slyford, whoever — this is where I am now. Masks off. One voice. The real one.

Come by whenever. The door doesn't move.

---

Austen Tucker is the author of The Two-Flat Cats, Bait and Switch, The Painted Cat, A Fuzzy Place, and Single Player Co-Op. They build worlds and write about the people that live there.

Why the 10% Productivity Plateau Should Worry You

Laura Tacho, CTO of DX, has been tracking what might be the most important dataset in AI-assisted development. Not a lab study. Not a controlled trial. A continuous measurement across [121,000 developers at more than 450 companies](https://getdx.com/report/ai-assisted-engineering-q4-impact-report/), drawn from a population of approximately 4.2 million.

Here's the trajectory. AI coding assistant adoption hit 92.6%, monthly usage at minimum. AI-authored code now accounts for 26.9% of all production code, up from 22% the prior quarter. The tools are everywhere. The output is measurable. The adoption curve is basically done.

And productivity gains have [flatlined at roughly 10%](https://shiftmag.dev/this-cto-says-93-of-developers-use-ai-but-productivity-is-still-10-8013/).

Not declined. Not crashed. Flatlined. The number went up, hit a ceiling, and stopped moving. More adoption didn't produce more gains. More AI-authored code didn't produce more productivity. The curve bent and leveled off.

That should worry you.

---

The obvious explanation is that the tools need to get better. Give it another generation. Bigger models, better context, smarter agents. The 10% will become 20%, then 40%, then the transformational gains everyone's been promising.

Maybe. But the data suggests otherwise. Because if the constraint were the tools, you'd expect gains to track adoption. More usage, more benefit. Instead, you get a plateau: the benefits arrived early and stopped scaling.

That's not a tool problem. That's a workflow problem.

---

Here's the math that explains the plateau.

Developers spend about 20% of their time writing code. The rest is meetings, design discussions, code review, debugging, documentation, deployment, incident response, and the organizational overhead that grows with every team member you add.

AI coding tools (Copilot, Cursor, Cody, all of them) operate almost entirely in that 20% window. They make the coding part faster. Let's be generous and say they make it 50% faster. Fifty percent of twenty percent is ten percent.

There's your plateau. Not a ceiling of the tools. A ceiling of the task.

You can make code generation infinitely fast and you will never break 20% total productivity improvement, because code generation was never the bottleneck. The bottleneck is everything else. The bottleneck is the organizational machinery that surrounds the code.

---

Tacho's data bears this out. Developers report saving [roughly four hours per week](https://getdx.com/report/ai-assisted-engineering-q4-impact-report/) from AI tools. On a forty-hour workweek, that's your 10%. And those four hours don't compound. They don't lead to additional gains. They just exist: a fixed savings that accrues on the first day of adoption and doesn't grow.

The onboarding data is more interesting. Time to tenth pull request, a common proxy for "when is a new developer useful," was cut in half. That's a real structural gain. Not because the code got faster, but because the organizational process of absorbing a new team member got shorter. AI tools acted as an accelerant for learning, not just for typing.

That's the pattern worth paying attention to. The gains that stuck weren't about code speed. They were about organizational speed. When AI tools reduced a friction that was genuinely in the critical path, like onboarding, the impact was significant. When they reduced a friction that wasn't the bottleneck, like typing speed for experienced developers, the impact hit a ceiling almost immediately.

---

This has implications that most organizations are ignoring.

If you've deployed AI coding tools and you're seeing that 10% gain, you're not behind. You're at the plateau. You've captured the available value from making code generation faster. Upgrading to a better model or a more expensive tool will not meaningfully move the number.

What will move the number is changing the other 80%.

AI-assisted code review. AI-assisted design documentation. AI-assisted incident triage. AI-assisted meeting summarization and decision tracking. AI-assisted dependency management and deployment pipeline optimization. The entire organizational surface area that surrounds the code and consumes most of the developer's week.

None of that is a coding tool. All of it is a workflow change. And workflow changes are hard. They require rethinking processes, not just installing plugins.

---

This is why the plateau should worry you. Not because the gains are small. Because the organizational response to the plateau is almost always wrong.

The wrong response is: the tools aren't good enough yet, let's wait for the next generation. That response assumes the constraint is technical. It isn't. The constraint is organizational. Waiting for better tools while your workflows stay the same is like buying a faster car and leaving it in traffic. The car isn't the problem. The road is.

The wrong response is: let's increase adoption. You're already at 92.6%. There's nowhere to go. The holdouts aren't the bottleneck.

The wrong response is: let's measure harder. More dashboards. More metrics. More lines-of-code tracking. All you'll find is the same 10%, measured with increasing precision and decreasing utility.

The right response is uncomfortable. It's to look at the 80% of the workweek that AI isn't touching and ask which parts of it are still necessary. Which meetings. Which review processes. Which handoff rituals. Which approval chains. Which coordination costs exist because the tools assumed a human would be the one coordinating.

That's not a technology problem. That's a management problem. And management problems don't get solved by installing a plugin.

---

Tacho said it clearly: "Tools aren't the bottleneck. Workflow and culture integration are." She's right. And she has the data to prove it.

The 10% plateau is not a failure of AI. It's a message. The message is: you've optimized the part that was easy to optimize. The hard part, the part where you actually change how work gets done, not just how code gets written, is still ahead of you. And no model upgrade is going to do it for you.

Volume Is Not Velocity

GitClear analyzed [211 million changed lines of code](https://www.gitclear.com/ai_assistant_code_quality_2025_research) across repositories owned by Google, Microsoft, Meta, and enterprise corporations. The dataset spans January 2020 through December 2024. Five years. The full before-and-after of AI coding tool adoption.

Here's what they found.

Refactoring dropped from 25% of changed lines in 2021 to less than 10% in 2024. Code duplication (copy/paste clones) rose from 8.3% to 12.3%. 2024 was the first year in GitClear's dataset history that duplicated code exceeded refactoring activity.

And the frequency of code blocks (five or more lines) that duplicate adjacent code? Up eightfold compared to two years prior.

We are writing more code than ever. And the code is getting worse.

---

I want to be precise about what "worse" means here, because it's easy to dismiss.

This isn't about bugs. It's not about whether the code runs. AI-assisted code generally runs fine. It passes tests. It ships. The dashboards stay green.

The problem is structural. Duplication up. Refactoring down. That's a codebase that's growing without being maintained. It's the difference between a city that builds new roads and a city that also fixes the old ones. Both cities get bigger. Only one of them stays navigable.

GitClear's framing is pointed: AI output "resembles that of a developer unfamiliar with the projects they are altering." The AI doesn't know the codebase's history. It doesn't know that the utility function it's about to rewrite already exists three directories over. It doesn't know that the pattern it's generating was deliberately refactored out six months ago because it caused a memory leak under load.

So it duplicates. Confidently. Fluently. With perfect syntax and no understanding of the context it's operating in.

---

Refactoring is the part that concerns me most.

A codebase that isn't being refactored is a codebase that's accruing debt at every level. Not the dramatic kind, not the "we built it wrong and have to start over" kind. The quiet kind. The kind where there are now four implementations of the same logic instead of one. Where the naming conventions have drifted. Where new developers can't tell which version of a function is canonical because there are three, and all of them work, and none of them reference each other.

Refactoring is the immune system of a codebase. It's the process by which a living system recognizes redundancy and eliminates it. When refactoring drops from a quarter of all changes to less than a tenth, the immune system is suppressed. The organism keeps growing. It just stops being coherent.

---

Why did refactoring drop?

Two reasons, and they compound.

First, AI tools are better at generating new code than at understanding existing code. Ask an AI to write a new function and it will produce something plausible. Ask it to refactor an existing module and it needs to understand the dependencies, the callers, the tests, the deployment context, the reason the code is shaped the way it is. That requires a kind of holistic awareness that current tools don't have. So the tools generate. They don't consolidate.

Second, refactoring doesn't get celebrated. Nobody gets a Slack emoji for reducing the line count. Nobody's sprint review includes "I deleted 400 lines of duplicate logic and the system is now easier to reason about." The incentives point toward features shipped, PRs merged, tickets closed. AI tools amplify the work that gets measured. The work that doesn't get measured, the maintenance, the consolidation, the careful reduction, falls further behind.

The tools and the incentives are aligned in the same direction. More output. Less coherence.

---

Here's the craft argument.

Productive and generative are not the same thing. A developer who writes 500 lines a day and a developer who writes 50 lines a day and deletes 200 are not comparable on output metrics. But the second developer is often the one keeping the system alive.

Volume is the easiest thing to measure and the worst proxy for progress. More code is not better code. More features are not better products. More PRs are not more velocity. Velocity implies direction. Volume is just mass.

When an organization reports that AI tools increased their output by 20%, the right question is: output of what? More features? More duplication? More code that works today and creates a debugging nightmare in six months? The GitClear data suggests the answer is "all of the above," and the ratio is shifting toward the bad kind.

---

I'm not anti-tool. I use AI tools every day. They're in my workflow at this point in a way that would be hard to reverse. But I also refactor what they give me. I read the output. I check it against the codebase I already know. I delete the duplication. I rename the variables. I do the work that the tool can't do because the tool doesn't know what the codebase is supposed to look like, only what it looks like right now.

That extra step is the difference between using AI as a generator and using it as a collaborator. The generator produces volume. The collaborator produces velocity. The generator makes the codebase bigger. The collaborator makes it better.

The GitClear numbers tell us that most teams are using the generator. The refactoring drop is the proof. The duplication rise is the symptom. And the long-term cost is a codebase that nobody understands because nobody was maintaining it; they were just adding to it.

Volume is not velocity. It never was. The tools just made it easier to confuse the two.

The Hidden Tax on Senior Engineers

There's a study that should be pinned to the wall of every engineering leader's office.

[Xu et al. (2025)](https://arxiv.org/abs/2510.10165) looked at open-source projects before and after GitHub Copilot adoption. The headline finding was what you'd expect: more code got written. Peripheral contributors, less experienced developers newer to the project, generated more output. The overall productivity curve went up.

But here's the number that matters. After Copilot adoption, core contributors, the senior developers who maintained these projects, reviewed [6.5% more code](https://arxiv.org/abs/2510.10165). And their own original code output dropped by 19%.

That's not a rounding error. That's a structural reallocation of the most expensive, most experienced engineering time you have.

---

Let me say it differently.

AI tools made it easier for junior and peripheral contributors to write code. So they wrote more of it. But that code still needed to meet the project's standards. Someone had to review it. Someone had to check that the confident, fluent, syntactically perfect AI-assisted output didn't introduce subtle bugs, break conventions, or quietly duplicate logic that already existed elsewhere.

That someone was the senior engineer. The person who knows the codebase well enough to catch what the AI missed.

Every additional AI-assisted pull request that landed in the review queue was a pull request that a senior developer had to read, understand, and evaluate against their mental model of the project. The more the juniors produced, the more the seniors reviewed. The more the seniors reviewed, the less they built.

This is not a side effect. It's the mechanism. AI tools increase throughput at the bottom of the experience curve by shifting load to the top.

---

Think about what a senior engineer's time is actually worth.

Not in salary terms, though that's part of it. In terms of what they uniquely contribute. Senior engineers do the things that can't be delegated down: architectural decisions, complex debugging, the refactoring that keeps a codebase from collapsing under its own weight. They identify the problems nobody else has noticed. They say "we should not build this" when everyone else is excited about building it.

That work doesn't show up in lines-of-code metrics. It shows up in the absence of disasters. In the stability of a system over time. In the fact that the codebase is still comprehensible a year from now.

When you redirect 19% of that capacity into reviewing AI-assisted pull requests, you're not losing "code output." You're losing the architectural maintenance that keeps the whole system healthy. And you're losing it invisibly, because review work doesn't generate the same signal that building does.

Nobody notices when the refactoring pass doesn't happen. They notice six months later when the system is brittle and nobody can explain why.

---

The researchers framed this clearly: code produced after Copilot adoption "requires more rework to satisfy repository standards, indicating a potential increase in technical debt."

That sentence should haunt you. It means the additional output isn't free. It arrives with a maintenance cost attached. And the maintenance cost falls disproportionately on the people who were already doing the most valuable work.

It's a tax. A hidden one. The kind that doesn't show up on the dashboard but compounds quietly in the background until someone asks why the velocity numbers are great but the product keeps breaking.

---

Here's the resource allocation argument.

Most engineering organizations think about AI tools as a multiplier. Give everyone Copilot, everyone gets faster, total output goes up. The math looks clean.

But the math assumes that everyone's time is interchangeable. It's not. A junior developer's hour and a senior developer's hour are not the same resource. They produce different kinds of value. When the tool increases the junior's output at the cost of the senior's, the aggregate numbers improve while the composition of the work degrades.

You're getting more code. You're getting less architecture. More features merged. Fewer refactoring passes completed. More PRs closed. Less time spent on the work that keeps the next quarter from being a disaster.

If you're a CTO looking at a dashboard that shows a 15% increase in PRs merged after AI tool adoption, ask the follow-up question: where is that increase coming from, and what did it displace?

---

This is not an argument against AI tools. It's an argument for understanding what they actually change.

AI doesn't reduce the total demand on your senior engineers. It redirects it. The demand shifts from creation to verification. From building to reviewing. From the work they chose to do to the work that arrives in their queue because someone else's output increased.

The fix isn't to take the tools away from juniors. The fix is to account for the tax. Staff your review capacity deliberately. Protect senior engineers' building time the way you'd protect any critical resource. Measure review load as a first-class metric, not an afterthought. And stop treating "more PRs merged" as an unqualified win without asking what got pushed off the plate to make room.

The senior engineer who used to spend mornings on architecture and afternoons on features is now spending mornings on review and hoping to get to architecture next sprint. They won't. Next sprint will have even more AI-assisted PRs to review. The queue only grows.

That's the tax. It's real, it's compounding, and it's invisible to anyone who only looks at the top line.

# Carl

My name is Carl and I work at Floor-Mart.  I always have worked at Floor-Mart and I always will work at Floor-Mart, forever and ever, amen. 

It's the middle of winter and I'm working in the parking lot.  There are carts that need to be collected every ten minutes.  I'm wearing a Floor-Mart T-shirt and the orange safety vest that the manual dictates I wear.  It’s snowing and it’s cold, but I don’t care.  Androids don’t get cold.  There are footsteps crunching in the oily snow just behind my back.  My manager Florence is walking out to see me. 

He’s a small man, with a warm, customer-service smile and a clean combover haircut.  As I turn to face him he is walking slowly, methodically, considering his words before he even speaks them.  Tact.  The mark of a true manager.  "Evening, Carl," he says to me.  We shake hands out of mutual respect.  "Did you hear the good news?" 

“No.” 

"Eustice turned in."  He looks to a far corner of the parking lot that’s littered with picket signs and candy bar wrappers.  "Remember him?  He was the Third Gen that worked in the back.  You remember him, right?" 

“Yeah.  I guess I knew him.”  Eustice worked for me back when I managed a Floor-Mart.  Good android.  He had a latency of five milliseconds when it came to answering questions.  That was two better than any other android in the store, good old reliable Eustice. 

“Well, today he just decided that it was time.  He came to me with the blue slip all filled out.  ‘Florence,’ he said to me, ‘I think I’d be better go turn in.’  His arm was sitting pretty low in the socket.” 

I look to the parking lot.  There are plenty of carts to be collected.  This is not an efficient conversation.  “Probably an actuator went out.  Happens all the time.” 

"Right, right."  Florence wrings his hands and looks at the ground.  Sheepish.  My memory banks call it sour-face, when you have something to tell an employee but don't want to say anything.  Manager dissatisfaction number twenty-one. 

I can’t take the stalling.  I tap him on the shoulder.  "You can just say it.  I'm here for the company, you're here for the company.  I have work to do." 

"Yeah.  Sure."  Florence forces himself to laugh.  "It's like this, see.  Yeah.  My district manager, we had a talk about you." 

"I'm listening." 

"Well, see, it's not because we don't like you.  Just remember that.  We're tickled pink that a man with your experience is working for our Floor-Mart.  We value every employee, and you are no exception."  Taken from the company handbook, right to the letter.   "It's just that... well... there are some problems we need to fix." 

"Go ahead." 

Florence nods and takes a deep breath.  More sour-face.  "First, since we’re talking about actuators, you've looked tired for the past few weeks, Carl.  Out of it.  Dead." 

"I know."  I point to my forehead.  The skin droops over where my eyelid should be.  "The actuator went out last month.  At least the skin's holding up, you know?  Makes me look like I have lazy eye." 

"Not quite."  Florence tapped on the ridge of my brow.  "The skin's wearing right down to the metal right here.  It's unsettling for customers to look at.  You should get it fixed." 

"I can't.  No one stocks the actuator." 

Florence nodded an understanding, managerial nod.  "Custom-made?" 

"Nope.  Just old." 

"Right... last-generation technology and all that.  Totally understand.  Just do the best you can, okay?"  He doesn’t understand.  Managers say these things to make their employees feel less alone, to encourage good work habits.  Managerial Motivation seventy-five. 

“I can’t help that I’m a workaholic,” I say.  Old manager joke thirty-three.  Sometimes I pull a random one from the database, just to feel like I’m still important. 

He points to torn-up beanbag of a stuffed animal hanging on the inside of my Floor-Mart vest.  "Workaholic, sure.  But I have a second complaint." 

"You mean Phillip?"  I take the stuffed rabbit from my vest pocket carefully, knowing that any sudden jerk might tear the belly stitching even further.  He sits quietly in my hands, one arm, no legs, and a face put on upside down.  Styrofoam beans poke out from the hole in his belly.  It’s garbage and I know it, even though I keep him around anyway.  "He's my lucky rabbit." 

"Right, right."  Florence's mouth twisted.  “Looks like an old redemption prize, right?  Like the ones you got from playing skee-ball or something.” 

I shrug.  "This is where you say, 'It's company policy for workers not to wear or carry articles inappropriate to the completion of their assigned jobs.'  It's on page twenty-four of the Manager's Manual." 

"You read that too?" 

"Nope.  It's stored up here."  I tap my forehead and wink at him.  "But Phillip, I figured he could be an exception.  He's a beaten-up rabbit, sure, but he's kinda cute.  I found him laying in the toy aisle when I managed the old Floor-Mart.  Company policy said I should throw him away, but hey.  I couldn't resist." 

"Interesting." 

"He's my lucky bunny!" I say it even though it’s not the right thing to say.  These words just happen when an android gets older.  That’s what I keep telling myself. 

"And that's just the problem, isn't it?"  Florence shook his head as he said it.  More sour-face.  "Carl, you're a wonderful worker.  I want you to know that because, well, it's like this, you see..." he paused for another breath, "The District Manager is concerned about your mechanical well-being." 

I laugh.  "Sure, I'm a little old.  I've been working here seven years, Florence." 

"Eight," he corrected me.  "It'll be eight this month.  Android brains have a shelf life of six years, Carl.  That's not to say that you're past your prime, though, because you've been indispensable out here as our cart collector for the past two months.  You're a wonderful person, and a hell of an employee.  I just wonder, well… did you ever think about turning in?" 

The word registers a blank in my brain.  I know something should be there but all I hear are the rat-tat syllables of a foreign language.  "I've always worked here, Florence.  That's just the way it is. " 

"I figured you'd say that."  He pulls a pamphlet from his Floor-Mart jacket pocket and puts it into my hand.  "Just... well, consider giving yourself to a reconstitution program.  Thanks to donations from androids like Eustice new generations of dedicated Floor-Mart tools have been created to support Floor-Mart's efforts to bring affordable goods to the common people." 

The packet reads, "So you're ready to call it a night."  The front cover shows an android in a state of disassembly, a smile on his face.  Bolts and hoses and diodes litter the page like confetti at a retirement party.  I slip it into my orange vest along with my rabbit.  I understand but I don't want to get it at all. 

"I know that this is hard for you, but just think about it.  Eustice is going to come back as a street sweeper for us.  He’s taking your job, Carl.  I could toss you back in warehousing, maybe graveyard shift stock, but I can’t have you collecting carts.  Eustice will do it better than you can.” 

“Then I’ll work in back.  Whatever you need done.”  That’s the Floor-Mart way.  Do what needs to be done, Fred Taylor be praised. 

“I’m not getting through, am I?” Florence wipes his face and lets out a hiss of air.  “This isn’t supposed to be a problem,” he says at length.  “Androids just get it. They wake up one day and bam!  DECO kicks in and they’re ready to become something new.  Third Gen model units are dead, Carl.  And here you are wasting away with every passing shift.  You can’t last forever. What can you do?" 

"Work," I say.  "Just like I have for the past eight years." 

- - - 

The Roller Palace is five minutes and thirty-nine seconds away from the Floor-Mart.  I count the time every night I walk to it.  The Palace is empty when I arrive and empty when I leave.  No one skates on the weekdays.  I'm standing on shag carpet that's too dirty to be comfortable and too tough to be worth replacing.  There's a cord running from my stomach to a small orange outlet just to the right of a half-dozen skee-ball machines that whine the sixty-cycle hum of an ungrounded electrical circuit.  It feels warm and filling in my torso. 

I shouldn’t be at the Roller Palace. 

I have no place else to go. 

The owner is usually playing skee-ball when I come in.  Norton.  An old friend from my days as manager.  He worked maintenance on androids in the store before his father died.  Then he inherited the rink and became a business owner.  He is a portly man, always smiling, always sporting pants that never quite cover his rear but always looked presentable at first glance.  Managerial material by the Floor-Mart book. 

I watch Norton play a few rounds.  The machine makes no digital noises.  He puts in a quarter, presses a plunger, and a line of wooden balls roll down to him.  He scores and the machine rings a bell.  The scoreboard is a hard plastic flipboard of numbers that make a satisfying "tack" every time he scores a point.  Ding, tack-tack-tack.  It settles into a soothing rhythm. 

"Another long night?" he asks.  He studies the pock-marked skee ball in his right hand before throwing it up the lane.  The bell rings.  The flipboard advances with a clapping noise.  Ding, tack-tack.  Ding, tack-tack-tack.  Ding, tack-tack-tack-tack.  Makes me feel more at home. 

"You know you're wasting time," I say to him.  Efficiency protocol number one seventy one.  "You could be spending your time prepping skates or popping popcorn at the snack bar." 

He chuckles at me.  Par for the course, I know; something about the way I talk gets him every time.  "So your night was same as always: all labor, no pay.  An android’s work is never done." he says as he picks up another ball.  Ding, tack. 

"Your aim's off," I tell him. 

"But I know you better than that."  He finishes his game and pops another quarter into the skee-ball machine.  "You need something." 

"Maybe you can get me a new actuator," I say.  I point to my sagging eye.  He shakes his head.  Roll.  Ding, tack-tack-tack-tack-tack.  "Or maybe you can take a look.  Fix it." 

"And pay me with what—shopping carts?"  He laughs and taps his skull.  "You know the decay time on those neural networks they used for your mind.  Seven years.  You're losing it."  He looks to the bunny sitting in my pocket.  "You keep a half-draggled stuffed rabbit for no logical reason.  None.  How efficient is that, ex-Floor-Mart-manager?" 

"Phillip has nothing to do with this."  I cover the bunny's head with my arm but I don't turn away.  Turning away is offensive.  Never offend a potential customer. 

"I'd call it a malfunctioning logic gate.  You're at some roller rink dive when you should be beautifying the shop.  No.  You get no wages, you work for peanuts.  They're trying to send the facts home and you don't even care that you’re an android past his prime. "  He shrugged.  "Ever think of turning in?" 

I ignore him.  Always keep business matters and private matters separate.  I need a new eyelid, not a lecture.  Managerial Conduct 101.  "Maybe you have someone I can contact.  A vintage machine vendor, perhaps.  Someone who works on the cheap." 

Norton shakes his head and looks to the ceiling.  "I hear they take androids at the modern art museum, nowadays.  Some prof takes the brains and bodies and has his students rework them into mobile sculpture.  Ever consider becoming an impressionistic representation?  I think you'd make a nice Model T with some doing." 

"I just need an eyelid," I tell him.  A blue brochure is heavy inside my Floor-Mart vest pocket.  So you're ready to call it a night.  Turn in.  Be done with everything.  "Just an eyelid.  That's all." 

"Or maybe you can do pet care.  There's always room for another pooper scooper drone in a pet-owning household." 

"I just want an eyelid.  I don't want to be turn in.  I just want to keep my job and work for the better interests of Floor-Mart." 

"They programmed you to say that." 

"I like my job." 

"That too; I've seen the programming."  He smiles and throws a ball up the lane.  Ding, tack-tack-tack.  "But sometimes... sometimes you just have to... I don't know.  Stop.  Give it up."  He turns from the skee-ball machine to face me.  "Sorry.  I was never good at giving these talks.  It's part of why I left in the first place, Carl.  Couldn't do it." 

"But it's part of your job.  You are a mechanic and an efficiency manager.  You look out for the best interests of the store."  Whose Floor-Mart is this?  "You may have checked logic gates, and oiled joints, and run portable charging stations when associates needed a quick charge on twenty-four hour shifts, but you were important." 

"Right.  Was a mechanic."  Norton clamps his lip into polite refusal position number seventy-five.  "Look.  I gota kick you out early tonight.  Doing a kids party.  You know how it goes.  I'll look at the eye tomorrow.  Maybe we can rig something out of a pneumatic tube from your cheek.  Maybe." 

"Okay." I unplug my cord.  "Where should I go?" 

"I'll call Linda down at the Robot Hostel."  Greenfield Regional Mechanical Repository, he means to say, but Norton is never precise with his words.  "She owes me a favor anyway.  I can get you in one of the pods for a night, wash you down, get your fluids changed.  It'll do you good.  Just get the hell out of here, okay?"  He looked to the clock on the snack bar wall.  "I can't have you here recharging when the families start arriving." 

- - - 

The hostel just outside of town is a spindle bubble-wrapped with capsules.  It's a matrix of comatose androids laid end-over-end along the wall.  Some sleep in liquid.  Others sleep in bare metal.  No one moves.  No one talks.  It is silent except for Linda's television game show blaring from the receptionist counter.  I tap the bell on the desk.  She looks up, shivers, looks down again. 

"You're the one Norton was talking about." 

I nod. 

"Yikes."  She points to an open tube at floor level.  "Knock yourself out.  You really should consider turning in -- saves me a bunch of trouble."  I nod and walk to the pod.  She laughs at her game show.  Someone is paying her minimum wage to sit and watch a hundred pieces of metal recharge their lithium-ion batteries in a facility that’s ten years past its time. 

That night I dream of being self-cleaning birdbath in a park, with plenty of sunshine on my face and no hands to wipe the shit from my eyes. 

- - - 

"You dreamed," the repairman says.  She is probing my ear with a thin pick of an electrometer.  Data is flowing from my brain into her palm computer. 

"Yeah.  I dreamed."  We're talking by the water fountain in the break room twenty minutes before shift meeting.  She's wearing her Floor-Mart vest and a scanner on her belt.  When she's finished with me she'll go back to taking inventory. 

"Impossible." 

"But I still dreamed." 

"So."  She spins the tool around inside my head.  "What did you dream about?" 

I read her name tag.  Sandra.  I compliment her on her hair.  She smiles.  Then I tell her about the birdbath. 

"You shouldn't be dreaming."  The tool goes further into my brain.  She's looking at a screen at her hip pocket and pressing a few buttons.  "Rest cycles are dead time for an android's brain." 

"I know." 

"But it was about being a birdbath, eh?"  She shrugged.  "Sounds like your DECOs trying to kick in.  Have you thought about turning in?" 

"It's come up before, yes."  I feel heavy.  Frustrated.  I don't exactly know what that word means but I think of it anyway.  Frustrated just sounds right. 

"I mean, it's not every day you hear about an android dreaming.  It's usually some sort of logic gate malfunction.  Something catches and your sensory input tries to make sense of what you get---" she pauses and looks more closely at her screen.  "Yep.  That's the ticker." 

"What's it?" 

She reaches out in one single, fluid motion and plucks Phillip from my vest.  I try to pull away, but by the time I react she has the bunny in her hand, the arms and legs dangling by threads.  "There’s your problem.  It's causing a catch in your logic circuits.  You care for it and you shouldn't." 

"I know." 

"Look.  I'm no android.  I don't get you; I just run my little maintenance tool when the manager tells me to run it and I figure out what's wrong.  Hell, this is the first time I've ever actually used one of these units–we don't do so many androids anymore.  Didn't test well.  Remember the protests?  The campaigns?" Phillip bobs this way and that as she talks.  My eyes follow his every motion. 

“They picketed my store,” I say.  “Lost three percent off our revenue.”  Phillip is fraying.  I can hear the threads tear. 

“People want cheap prices, but they get real touchy when a company plays God.”  The bunny leaves her hand and tumbles, end-over-end, into the gigantic trash can in the corner of the back hallway.  It falls into the gaping black plastic mouth and is gone. 

"I know."  My eyes are in the blackness scanning for Phillip’s innocent face. 

She chuckles, pauses.  "I guess...  I'd just suggest that you start thinking about options.  Yeah.  Think.  Do you think?  I guess you think.  I don’t know  Just... get ready, I guess.  Whatever you androids do." 

She waits for an answer.  I shrug.  "You should get ready for the shift meeting," I tell her. 

"Right."  She pulls the machine off her shoulder and ducks into the manager's office.  Then, while she is out of sight, I dive into the gaping hole and rescue Phillip from certain death. 

- - - 

Florence is outside greeting people as they come through the door.  He's wearing a suit and a tie.  He hands out coupons to everyone that comes into the store.  Public relations 101: first suggestion in the manual. 

He waves me over as I'm pushing a line of carts in through a small door.  He has manager smirk number two seventy-five – the one for associates that aren't doing what they're supposed to do.  "How's the shift going, Carl?" he asks me. 

I run a quick procedure check.  Carts are collected.  Trash cans are clean.  I’m still not recycled.  "Good," I say. 

"As always."  He looks at my vest and points to the rabbit head poking through the blue mesh fabric.  "So, well, yes.  Let's see... how to put this.  Right.  Well, I mean to say... see..." 

"You want to reprimand me about the rabbit," I tell him.  Standard managerial procedure.  Warn once, reprimand once.  It’s Management 101 in the Floor Mart manual. 

"Yes, that."  He reaches forward and plucks Phillip from my pocket before I can reach down to protect the rabbit.  "I told you not to bring him to work." 

"I don't have any place to put him." 

He tosses Phillip back at me.  I fumble with the stuffed animal, making sure to collect every arm and every stitch in my gentle, cradling hands.  "Get rid of him," he replies.  "Simple as that." 

"But I like him.  Phillip keeps me company." 

"Right."  He rubs his nose.  Intolerant expression number sixty-five.  "Carl, how did you manage your Floor-Mart?" 

The words come without thinking.  Automatic demonstrative pattern number seven, used by district managers to evaluate efficiency.  Reporting 101 in the Floor-Mart manual.  "Two terabit wireless network allows me to receive and feed information to all my associates at once.  The administrative measure results in a five point five percent reduction in operating costs, a ten point eight reduction in manhours, and a seven point three increase to total revenue." 

Florence shifted his weight from one foot to another.  "So getting everyone on the same page was simple.  Yeah.  Because they were all, well, androids.  Correct?" 

Florence’s mouth is a wriggling electric tool.  I see the sparks in his spittle, the pneumatics on his tongue.  It's a wriggling, twitching snake trying to feed its way into my brain, guided by camera eyes that rotate and focus in the cool purple glow of the bug zapper disco ball. 

"First-model RA androids with one to three years' use already logged.  Total wear and tear costs total six point three hundred thousand, with total benefit coming to seven point two hundred thousand.  Marginal benefit of androids proven larger than that of wage-earning human beings in seven separate third-party analyses." 

"And you never had problems with androids displaying... strange behavior?" 

"Problem androids recycled themselves in accordance with logic circuit DECO."  He is a spark that slithers in the cool night sky.  Whose Floor-Mart is this?  "Androids feeling excessive wear and tear who calculate the cost of their repair to be greater than the benefit of their labor give themselves to local recycling plants.  Sales of androids to recycling plants results in seven point three thousand dollars' positive economic profit per unit recycled."  Fred Taylor be praised. 

"Right, right."  Florence shimmers like a short circuit.  "I was just checking if you still understood the business of this whole deal.  Yeah.  Never know when it'll come in handy, right?" 

"Right."  We stand for a moment longer.  The disco ball becomes a bug zapper.  The zapper is the blue packet in his pocket.  So You Want to Call it a Night.  I’m l just waiting to touch that disco ball, that bright, electric purple disco ball that is so amazingly wonderfully bright and pretty. 

Phillip is still in my free hand.  I put him back into my vest pocket, nice and comfortable.  Then I look away from the disco ball and the world snaps back into place. 

"Can I go back to pushing carts now?" 

- - - 

Eustice works with me now.  He’s training for the job.  Androids come back like blank slates sometimes.  Side effect of turning in.  They scoop out your brains and bolts and throw it into another body.  Trained professionals aren’t cheap.  Trained technicians are.  Eustice sees the technician. 

Eustice lumbers along.  They don’t make reconstituted androids for speed.  He has Floor-Mart brand tires and Floor-Mart brand vacuum brushes.  He hums as he works.  He goes back and forth between the two entrances and sweeps the sidewalk.  Sometimes he sprays it with Floor-Mart brand nozzles that spray Floor-Mart brand cleaning solution.  The sidewalk is already clean.  Tomorrow I’ll teach him how to bring in the carts so he can be more productive. 

I try to talk to Eustice every once in a while.  He has a speaker box made out of cardboard mounted just above his camera eye.  They make the speakers in Japan.  You fold them into a box and you have a speaker without the expensive housing.  The innovation saves five dollars on every reconstituted android.  When he speaks his voice is tinny, profitable. 

“Welcome to Floor-Mart.” 

Good ol' reliable Eustice.  “Good to see you,” I reply to him. 

“Restroom is to your left, three doors down.” 

“On the ball like always.”  Managerial conduct seven thirty-three – always compliment productivity.  “How's the new job treating you?  Any questions?” 

“Have a nice night.” 

I looked for something in Eustice's eyes.  Eustice tells a joke ten seconds into conversation.  That's why I like Eustice; he makes people laugh.  “Got one for me today?”  I ask.  Managerial conduct number sixty-seven – encourage positive behavior with on the job training.  “You used to tell one about a chimney sweep.  Remember that one?” 

“Restroom is third door on the left.”  The cardboard box mouth muffles the sound.  Technicians can never fold the boxes right.  I want him to smile.  I want him to talk.  Dear Fred Taylor full of grace make Eustice talk.  You gave him life again, now make him live. 

“Say something for me, Eustice.  Anything.” 

“Have a nice night.”  The box pops forward.  A cardboard fold has come undone.  The vacuum motors whine.  The box mumbles.  Eustice moves an inch.  Two inches.  His tires cinch up.  The motors stop.  His body is in perfect working condition but Eustice doesn't move. 

This happens.  Sometimes androids don't make it through turning in.  Old parts.  Wear and tear.  Your mileage may vary.  No one does reconstitution anymore.  Too much training.  Too few androids turning in.  Better to build new and forget the old. 

Two associates come outside with a dolly.  They wheel Eustice to the back alley.  The speaker box hummed all the way, but the words wouldn't come out.  If his box had been folded better he might have given me one more Have a Nice Night before he went away. 

I never see Eustice again. 

It's shift change time and I'm alone in the parking lot with Phillip.  The peak shoppers went home for the day.  They left their trash strewn about the lot and I’m cleaning it up because I have nothing better to do when all the carts are inside and there’s no one going inside.  It’s eleven o’clock at night.  The town is buttoned up tight.  The wind is blowing seven point three knots to the south-southwest and there’s a frozen drizzle coming down from crying angels at the rate of point zero three inches an hour. 

I have a nail on the end of a stick.  Poke a flyer.  Put it in the bag.  Poke a can.  Put it in the bag.  I hum the Floor-Mart jingle as I work.  It’s short.  I hum it a lot.  It never gets out of my head. 

It’s shift change time and I’m alone in the desert with Phillip.  The androids died of thirst a few years ago.  They left their cardboard box voices strewn about the lot and I’m cleaning them up because I have nothing better to do in the middle of the desert when no one is around to watch me turn into a bleached collection of joints and broken eye actuators.  It’s dusk.  The sand blows over my shoes.  The wind is blowing eight point seven knots straight north and I can taste rain on the air in the next two days. 

Phillip looks at me.  “Don’t let them kill me.”  He has no eyes.  He has no legs.  There is thread and there are broken stitches but he’s not really there at all.  He’s dead like Eustice's voice I spear with my nail on the end of the stick.  Poke a speaker.  Feed the bunny.  Poke a skull.  Feed the bunny.  This is my job.  This is survival. 

It’s shift change time and I’m in black clouds with Phillip.  It’s four in the morning.  The store is far below me.  I’m standing with a nail and a stick and there’s paper in the clouds that I don’t care to get.  I’m tired.  I want to float up into the starry night sky where the angels cry the rain and old men blow the wind.  It’s almost sunrise.  Whose cloud is this?  My cloud! 

“We’re gonna be all right,” Phillip says to me. 

- - - 

Norton's soldering iron is in my eye socket.  “You owe me,” he says.  My face is hanging over a pair of speed skates on the skate rental table.  It’s a lump of latex and ceramic contacts that shines under fluorescent lights..  A little white spider of light sits just above the backside of my left eye and as Norton moves my eye actuator around the little white spider jumps all around the skin. 

The world flashes blue, then red, then goes dark.  “Shit.  I caught a circuit board.”  Tap-tap-tap.  The world flashes back into focus.  “That get it?” 

“Yeah.” 

“Sorry.”  He grumbled and went back to work.  The part is a thin hypodermic syringe cut to fit the hole.  It is filled with hydraulic fluid and connects to a pump where a human brain would be.  When I lift my eyebrow electronic impulses will push fluid into the syringe and my eyebrow will rise, push and pull, stimulus and response. 

“Does it look okay?” I say.  I don’t have a brain.  I have a parallel processor in my torso.  I have a spider dancing inside my face.  She’s a wonderful little spider with a crown and a prince charming and she’ll live happily ever after just like all the smiling customers in all the Floor-Mart teambuilding books I’ve read since the first time I powered up. 

Norton shrugs.  “I’ve seen better.”  He pulls a bundle of wires out from above my lip, cuts one.  My tongue stops existing.  “You gotta take better care of yourself.” 

“I do.” 

“Floor-Mart parts aren’t worth scrap,” Norton says.  He chuckles.  The soldering iron touches my lip and suddenly I have a tongue again.  It wriggles around my mouth, fills the small, painted latex cavity.  “No wonder they stopped making androids.  Anyone so much as looks at your insides and they get a little squeamish.  God knows I do.” 

“He’s going to fire me.”  I say.  The word doesn’t quite come off my tongue.  It doesn’t settle into my circuits.  “Florence.  Florence is going to fire me.” 

“Probably.”  Norton shrugs.  “That's the way the business model goes, though.  You androids work for the cost of your production and then you die off.  Downright creepy.”  The spider princess is offended.  It jitters in fluorescent light and calls for its guards to come out of my temples, but no one comes. 

“What happens if I just stay alive?” I ask. 

“Hmm.”  Norton put down the soldering iron long enough to rub his chin.  “Stay alive.  I… I guess I don’t know.  Maybe you’d just keep deteriorating until there was nothing left.  DECO's supposed to kick in while your parts are still halfway useful – it’s the most efficient use of resources.  Five years of service offsets the cost of production, and then the android is recycled into another useful form.  Optimum use of resources and optimum time-frame to introduce future technologies.  Remember that?  They program the cost-of-production algorithms into your head.  Factory standard.” 

I try to think but all that comes to mind is how majestically the spider princess dances inside my face. 

- - - 

Florence comes outside to talk to me.  He does this a lot.  It's a way to get away from the wage-slaves he deals with inside the store.  They talk back.  I don't. 

He has a small envelope in his hand.  Blue envelope.  Recycling plants and you.  "I think I have a solution for our little problem," Florence says.  "You're already losing your faculties.  I thought we'd be okay keeping you on for a little longer, but it looks like our best bet is to re-allocate your resources. You know what I mean, Carl.” 

Androids don't get pink slips.  Pink slips mean you get to leave, get a new job.  Blue slips are requisition forms.  Strong requests.  Death warrants.  Florence's mouth is a big yellow gaping sweaty hole ready to eat up the blues.  Chomp chompchompchomp. 

"Well, see, it's like this, really."  He adjusts his glasses.  "The district manager decided that it would be best to use your brain in the shell of an automatic buffer.  Floor-Mart Corporate sent me the plans today; it looks slick.  you'd buff the floors, clean the walls, dust the rafters... the ultimate cleaning machine, working twenty-four-seven to make this Floor-Mart the best Floor-Mart it can be." 

"I'm fine."  I want to cry.  That's what you call it, right?  Cry? 

"You most certainly are not."  He taps his head with a fat finger.  "Stores that requisitioned automatic buffers received a seventeen percent increase in revenue.  Seventeen!  All that for the price of one android.  Pennies on the dollar!" 

"It's the right decision," I say.  The words won't connect.  Yellow mouth and blue paper.  We're in a sea of folded cardboard and Floor-Mart needles, chomp chomp, eat 'em up. 

"You should read the plans – it's pretty impressive.  See, you'll have these pseudopods that attach to walls and ceilings while excreting antibacterial gel.  You'll move and clean.  You'll change tools and you'll clean.  You'll clean and you'll clean more!  It's a perfect machine, Carl.  Better than hiring more workers.  Better than Eustice!  Beautification has never been so easy."  I shouldn't be sad.  I should be happy to help Floor-Mart achieve a new level of efficiency.  I really really should but the yellow mouth keeps chomping at my heels. 

I can't respond. 

Florence tries to touch me on the shoulder.  Establish contact.  Managerial conduct number three.  "Please don't think about this like death, Carl.  You won't die.  Plenty of androids go on to live happy, fulfilled, recycled lives.  You'll provide a service for the company even past your usable days as an android." 

Nothing. 

“Eustice was just an exception to the rule.  A mistake.  He couldn't hack it.  But you... you're a trooper!  You're gonna do ol' Fred Taylor proud.” 

Florence chucks me on the shoulder.  Establish camaraderie.  Managerial conduct number five.  "Come on.  I need you climbing on the walls, fastidiously cleaning twenty-four-seven.  We have to take pride in our store."  He pauses to cock his arm back for a chant.  "Whose Floor-Mart is this?  Our Floor-Mart!" 

He smiles and tries to look jovial, but I just see his mouth eating me alive and me not having anything to say about it when the teeth eat my life with a  wokkawokkawokkawokkawokkawokkawokkawokkawokkawokka 

- - - 

He's skating laps.  Norton is skating laps.  He's going round and round the orange skating circle, sometimes forwards, sometimes backwards.  Sometimes he spins around in place.  No one is there.  The party is canceled.  The birthday girl had a case of stomach flu so they paid their cancellation fee and that was it.   Nobody's skating at the Roller Place.  He should be cleaning skates.  He should be advertising.  He should be making money like a good entrepreneur, Fred Taylor rest his soul.  But he's just gliding around and around on that orange sticky rubber surface, smiling, listening to Duke Ellington on the radio.  Fireflies from the disco lamp above.  Hum of the skee ball machine below my feet. 

"You could quit," Norton tells me.  "Just walk away." 

I try to think about it.  The words don't actually exist.  The thought is causing a stack overflow in my logic buffer.  It hurts, dear Fred Taylor it hurts like someone is trying to open up my circuit board with a water nozzle and a penknife and fill me up with watch batteries.  I just want to shut down and stop thinking and live like Phillip does in my shirt pocket so soft and so safe like a little baby that I never had the chance to be.  I worked the day I was born and I worked when I was supposed to die. 

"I'm a burden." 

Phillip never had a chance 

Not a chance in hell 

Little strings and plush fur 

              Like Grandma's Brand throw pillows 

              (Two dollars on clearance) 

do you work for peanuts? 

              Power? 

              Parts? 

              Turning in? 

Whose Floor Mart is this, anyway? 

It's cold 

It's alone 

It's dying like a thousand bones spread across skee-ball machines going tacktacktacktack 

He turns. 

He's skating backwards. 

The skates flash when the disco light touches them. 

"You're almost becoming human." 

Who said this was part of the deal? 

              (No one said) the deal is no deal a short trip a quick contract I forgive your bad habits but I still think I deserve better 

(Eight years) 

Fred Taylor have mercy on me 

"Maybe it's all in my head," he says 

my head your head the reader at home 

click the email wash your hands 

get the deal go home 

it'll all be gone tomorrow like so unraveled cardboard boxes in 

              moonlight 

when I'll spear and pick them up 

"It's probably all just random switches1010011100011000101001010100101001001010000101001010100100101010101010010 

he won't stop screaming the bunny won't stop screaming I should break his neck tear him apart make it quick 

shotgun to the face 

turn in 

sweep it up twenty-four seven three sixty-five 

smile and say hellowhatcanidoforyou 

eight years of service and all I got was this lousy vacuum 

"Human.  I just called you human, didn't I?"  He laughed and shook his head.  "So much for that Floor-Mart Pledge." 

Phillip screams when he dies.  It's soft and only I can hear it.  Like a baby crying. 

He writhes styrofoam beads and loose stitches. 

His stuffing shuffles like so many burned-out transistors. 

He's in my hand.  Phillip is in my hand.  He's staring at me.  His eyes are empty black skee balls.  His arms Floor-Mart brand hoses.  His body a little egg that needs to be buffed.  He screams.  It burns.  I reach all the way to the floor.  I'm leaking I'm leaking out onto the floor like a little river styx onto my shoulder and my chest and my hellohowareyou package that gets big when it needs to be big just a river flowing forever to carry Phillip away cold and alone. 

He was afraid of that.  He didn't want to go alone.  He wanted me with him.  He wanted to be important to be wanted forever and ever even though he knew it would come one day.  I'm leaking I'm crying an undertow to take him away 

Manager procedure twoseventythree 

Thecustomer is always right 

The android is always right 

The android is always right 

The android is always right 

The android is always right 

The android is always right 

The android is always right 

STOP LINE 0ACFFFFFF0-AF – INITIATE REBOOT 

The rabbit is in his hands, now.  There's a handkerchief in mine.  "Clean yourself up," he says to me.  There's yellow hydraulic fluid trickling down my stomach and through my jeans.  "And while you're at it, get the degreaser out of the rental shop.  That's never gonna come out of the carpet." 

I try to do what he tells me to do.  I step.  My shoulder pops.  Then it jerks.  My arm slides from its socket and onto the floor.  My silicone skin tears as the arm falls.  I feel nothing.  Norton stares and makes no excuses for it.  For the next minute it's all we can do to stare at the arm twitching on the shag carpet floor.  It's connected through a single fiber-optic cable that winds its way into my gut like a girl's curly hair. 

“I'm sorry,” Norton says.  “I think the threading corroded clean through, from the looks of it.  Old age and wear and tear.” 

“It's okay.”  My arm on the floor is one hundred thirty percent inefficent and looks like a drying earthworm on a summer day. 

“Let's have a look.”  He reaches for my side and opens me up.  I don't move.  I can't move.  My arm is lying on the ground and if I move I'll break the cable and the arm will be dried up and dead. 

“I don't want to be a bother.”  I double-check to make sure Phillip is okay.  He watches everything from a bench not four feet away from me.  Norton knew how to take care of his customers. 

“Nonsense.”  Norton pulls out a screwdriver and pushes wires around.  Then he whistles.  "It's a mess in here." 

"It's been two point seven years since my last checkup.  The company policy dictates that off-lease android models receive no mechanical checkups on a count that the procedure is too expensive." 

"Yep.  Right there's your problem."  He takes out a penlight and shines it inside of me, just to make sure.  "Looks like one of your batteries leaked into your sensory matrix.  The parts are corroded as all hell.  It works for now, but I don't think you're gonna get much juice out of your reserve power supply.  Fifteen minutes at a time, at most."  He grins and reaches up to shake my hand.  "Congratulations, Carl.  You are officially useless." 

“But I was fine yesterday,” I say. 

"Of course you were.”  He smiles and takes me by my still-attached shouler.  “Want my advice, get the hell out of Floor-Mart.  They'll reconstitute you until you're just a heap-of-metal doorstop.  You can't throw your life away working for the man.  God knows I didn't." 

“This Floor-Mart is my Floor-Mart.”  Company slogan number one.  Something to say when there's nothing to be said at all. 

This didn't fly with Norton.  Not while he scooped away salty powders that had once been my insides from my melting heart.  “Don't you get it?  Floor Mart doesn't care about you.  Floor Mart cares about the dollar sign in front of your name,  Profit, Carl.  Your existence hinges on the fact that creating and maintaining a worker android is cheaper than paying a human being six years' wages.  You are a boost to the profit margin.” 

“But that's why I'm here.”  This should make me feel happy.  Why doesn't it make me feel happy? 

"I know."  He shakes his head and looks at Phillip.  Maybe the rabbit's screaming at him too.  “You are their crowning achievement, their primest of prime retail.  You're cheaper than human.  Generic human.  Floor-Mart brand human.  And, unlike real humans, when they're done with you they dont' need to worry about a thing.  They just throw you away.”  He's still staring at Phillip.  Maybe Phillip's telling Norton that he is sad and lonely and he doesn't want to die even though he's unwanted and beaten-up and dead Fred Taylor forgive us our sins. 

“You can throw him away,” I say.  Phillip doesn't want to die.  He screams at me, but I can't take care of him anymore. Not me.  Not now.  Fred Taylor forgive me. 

“The rabbit?”  He pauses.  There's something wrong with his eye.  He rubs at it until it turns red and puffy. 

“It's just Phillip,” I say to him. 

He shakes his head and tries to swallow a lump in his throat.  "Jesus.  Give me that rabbit; it's more busted than you are."  He chuckles and walks to the skate shop.  "I have a needle and some thread in there.  I was just going to put him back together, you know.  Sew him up.  Show him some love." 

"It's irrational."  Logical circuit number two-forty-seven triggers like a firework.  "You are spending time and effort on a small, useless product.  Buying a new one would be cheaper." 

"It's called 'social benefit.'  You care for Phillip, and I care for you.  Therefore, I do what I can to fix it.  Try it sometime, will you?" 

The circuit shuts off.  "You can do that with a broken product?" 

"I've done more with less," he says.  He skates to the rental shop with the rabbit cradled in his hands like an associate handling a trusted customer.  I try to follow but the light is too bright he is shining in orange and disco ball glory while I am unwashed and aching tired of my job and my world.  I can't go back even though I need to go back but Phillip's still there happy to be fixed Fred Taylor bless his soul. 

“Take this place, for example.  It’s falling apart."  Norton points to the far corner of the building with a single, crooked finger.  There's fresh, brown, fake stone paneling on the wall.  "Had to cover some mold on the far wall just yesterday.  I came in and there it was, sick and green and smelling up the place.  That’s finished, but then there’s the rusted-out plumbing in the bathroom, the faulty wiring in the ceiling, the duct tape holding the dance floor together...  I'm barely keeping up my fixed costs as it is.  I keep hoping it's going to get better." 

"It's suicide." 

He shrugs.  "Of course it is.  No one skates anymore.  Why skate when there's so much to do at home?  You can buy it all at Floor Mart.  Buy it for cheap.  Cheap and flimsy and broken before it gets home.  I hate the fucking place, Carl.  Hate it with all my heart." 

Phillip screams when the needle pierces him.  So annoying.  He should just unravel and die if he thinks it hurts so bad.  He makes me wince when he screams.  Hydroulic fluid flows down my cheeks. 

"Yes.  Twenty-four-seven worker that doesn't take a wage.  It's scary.  Hell, if I didn't know you I'd probably do the same."  I pick up my arm.  Norton turns away.  Embarrassment.  Grief.  Confused customer number seven ninety three.  "I feel sorry for you, Carl.  Really I do.  I don't have to worry about the kind of crap you worry about.  Why?  When I die.  I'm dead.  That's it.  I don't know when, I don't know where.  But sometime in the future I'm going to die.  This roller rink, the mold on the walls, the poor turnouts, the cancelled parties... it all won't matter anymore.  But you…” he paused.  “You can keep coming back.  Over and over again.” 

“Until my parts are worn and dead.” 

“Then they’ll just take what they can take.  Joints.  Scrap metal.  Transistors.  As long as it’s profitable.  They’ll mangle you until you fall apart.” 

“Like Eustice.”  I see him wheeling off to the dumpster.  Unraveling like a messed-up cardboard box.  "I'm not going back," I say.  It's more of a nervous tick than anything else.  "I don’t want to be a floor buffer.” 

“Then don't.”  Norton made it sound so simple.  So easy. 

“They'll come after me.”  I shiver.  “I need to get away.” 

He points to the battery leaking out of my gut.  "You're not going anywhere.  You probably have an hour, maybe two until the charge is completely gone.  That's it.  No more." 

Floor Mart is burning in my mind.  Protocol circuits melt away, one right after another.  The acid is fire but Phillip’s there to cheer me on.  He’s screaming like a scared child but he’s still giving me those go-ahead skee-ball eyes.  Nothing to lose.  Nothing to gain.  We’re gonna make it.  “Not going back,” I say again. 

Norton smiles.  "I could get you a proper burial, too.  Maybe.  It'd take some finagling but I think we could make it work as long as everything stays quiet.  Just gotta keep Floor-Mart out of the loop.  It's up to you." 

"No.”  I looked to Norton.  No managerial protocol.  No customers.  Nothing but sincerity, whatever that means.  “Turn me in.  Make me useful."  Another logic circuit curls up and dies.  It's a burning field of cardboard and stuffing and clearance prices in my mind. 

"If that's what you want."  Norton walks over to me.  Phillip is in his hand.  He claps me on the shoulder.  He's looking away from me and pushing a palm against his eye – this one's not in the Floor-Mart emotional gesture matrix, but it looks sort of like ecstatic customer number seventy-one, only frowning.  I stand and patiently want for him to look at me again. 

"Well."  He puts Phillip on my free shoulder.  “Best I could do on short notice.  The stitches are a little jagged here and there – never was good at patching things up.  We can get him to a proper sewer tomorrow.” 

I smile.  There’s no protocol rule attached to the gesture.  “You’re plenty good.  Phillip appreciates it.” 

"Ha!”  He tries to laugh harder but he can’t bring himself to smile.  “By my estimate you have two hours to do whatever you want.  What sounds good?" 

"I don't know.  I usually work." 

"You're not working.  You're dying."  His voice box comes unraveled like Eustice’s for a moment.  "Two hours.  What'cha got?" 

"I could try my hand at skee-ball." 

"You're about to die and you want to do is play a silly redemption game."  He smiles.  "We could at least go out for a beer." 

"I don't consume." 

"Touche."  He pulls a pair of quarters from his pocket.  "Skee-ball, then.  What the hell; we'll make a night out of it!  You and me.  Two losers with nothing much left to lose.  What else could go possibly go right?" 

- - - 

It's half-busy Floor-Mart company skate rally at the rink.  There's a handful of kids from a local kindergarten making lazy laps around the sticky orange track.  Two middle school kids waltz on the dance floor.  A pair of teens, probably babysitters, sit in a dark corner doing things they shouldn't do in public.  These people used to walk right past me.  Now they laugh and smile and put quarters into my body.  Funny how context changes everything. 

I'm swallowing skee-balls.  Tack-tack-tack.  There's a pair of kids playing best-two-of-three.  My kid keeps hitting fifties.  He's really very good.  Phillip watches it all happen from inside a little Plexiglas cage above the scoreboard.  I thought he wouldn't like it too much there but he seems to have a good time anyway.  He says the kids are fun to watch.  We may not do much but we’re happy. 

Norton said these machines were made to last.  Twenty years from now I’ll break down again.  Someone will tear me apart and make me into the next big thing.  Maybe it will be Norton.  Maybe I’ll get passed to his next-of-kin.  Maybe I’ll end up in a pawn shop.  But no matter what they’ll turn me in, over and over again, until finally I’m just a big pile of semi-intelligent scrap sitting in the corner of some salvage store, waiting for some modern artist to come along and make me into a statue.  Reinvention.  Recycling.  Fred Taylor be proud of me. 

Stretch your dollar to the limit.  That’s the Floor-Mart way.

Why AI Helps Juniors and Slows Down Seniors

Here's a finding that keeps showing up across every major study of AI-assisted development, regardless of methodology, sample size, or which tool they tested:

Junior developers benefit more than senior ones. Sometimes dramatically more. And in certain conditions, senior developers get slower.

This is not a temporary glitch. This is structural. It will hold even as the tools improve. Understanding why requires thinking about what expertise actually is and what happens when you insert a confident, fast, sometimes-wrong assistant into the middle of it.

---

The data is consistent enough to be boring at this point.

[Peng et al. (2023)](https://arxiv.org/abs/2302.06590) ran ninety-five developers through an HTTP server implementation task with and without GitHub Copilot. The Copilot group finished 55.8% faster overall. But the gains weren't evenly distributed. Less experienced developers and those who coded more hours per day saw the largest improvements. The tool closed the gap between junior and senior performance.

A year later, [Microsoft and MIT](https://mit-genai.pubpub.org/pub/v5iixksv) ran a field experiment with nearly two thousand developers across Microsoft and Accenture. Junior and short-tenured developers were more likely to adopt Copilot, more likely to keep using it, and more likely to accept its suggestions. They saw the biggest productivity bumps. Developers above median tenure showed no statistically significant increase in output.

Then METR dropped the hammer. Sixteen experienced open-source developers, each working on repos they personally maintained, using Cursor Pro with Claude Sonnet. These were not juniors. These were the people who wrote the code. And they were [19% slower with AI assistance](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/) than without it.

The developers themselves predicted AI would speed them up by 24%. After the study, they still believed it had sped them up by 20%. The gap between perception and measurement was as wide as the slowdown itself.

---

Why does this happen? The standard explanation is that the tools aren't good enough yet. Give it a year. Better models, better context windows, better integration. The gap will close.

Maybe. But I think the standard explanation misses the structural part. The part that won't change even when the tools get dramatically better.

Here's the thing about being good at something. Expertise isn't just knowing more. It's having built an internal model of the problem space that lets you skip steps. A senior developer reading a codebase doesn't process it line by line. They pattern-match. They see the architecture. They know which files matter and which ones don't. They've internalized the dependencies and the history and the reasons behind the weird parts.

When you hand that person an AI assistant, you're not giving them a tool that operates on their internal model. You're giving them a tool that operates on the text. The AI doesn't know why the weird parts are weird. It doesn't know the history. It doesn't know that the function it's about to refactor was written that way on purpose because of a race condition that took three people two weeks to debug.

So the senior developer has to do something they don't normally do: audit. They have to read the AI's output with the same care they'd give a junior developer's pull request. They have to check it against their internal model. They have to verify that the confident, fluent, syntactically perfect code doesn't break something subtle.

That takes time. Often more time than just writing it themselves.

---

For the junior developer, the calculus is completely different.

They don't have the internal model yet. They're building it. And the AI is a remarkably good scaffolding tool for someone who's still constructing their understanding of the problem space.

Need to set up an HTTP server? The junior developer could either spend forty-five minutes reading documentation and Stack Overflow posts, or let the AI make a working skeleton in thirty seconds. Not because the AI is better than the documentation. Because the AI produces something the junior can immediately run, test, break, and learn from. The feedback loop is faster.

The junior doesn't have to audit against an internal model that doesn't exist yet. They just have to ask: does it work? And if it does, they've learned something. If it doesn't, they've learned something else. Either way, they're moving.

The Microsoft/MIT study found that onboarding time, measured by time to tenth pull request, was [cut in half](https://getdx.com/report/ai-assisted-engineering-q4-impact-report/) when developers used AI tools. That's not a productivity gain. That's an acceleration of the expertise-building process itself.

---

This creates an interesting structural asymmetry.

AI tools flatten the skill curve. They pull the bottom up without pushing the top higher. In a team context, that means the average output increases while the peak output stays the same or drops. The distribution compresses.

For a manager looking at aggregate metrics, this looks like a win. More PRs. More code merged. The team is "more productive." But the nature of the productivity has changed. More of it is AI-scaffolded junior work. Less of it is the senior-initiated architectural decisions, refactoring passes, and deep debugging that keep a codebase healthy over time.

You don't see this in the weekly numbers. You see it six months later, when the technical debt catches up. When the codebase has grown 30% but nobody can explain why half the new code exists. When the senior developer who used to spend mornings writing features is now spending mornings reviewing AI-assisted PRs from three juniors who are each producing twice the volume they used to.

---

The experience gap isn't a failure of current AI tools. It's a reflection of what expertise does that AI can't. Not yet.

Expertise compresses decision-making. It lets you skip the parts that don't matter and focus on the parts that do. An AI tool that can't see your mental model can't compress your decisions. It can only generate options. For someone who doesn't have a model yet, options are gold. For someone who already knows the answer, options are noise.

The tools will get better. They'll learn codebases more deeply. They'll carry more context. They'll develop something closer to architectural awareness. When they do, the gap will narrow. But it won't disappear, because the underlying dynamic, that expertise means you need less help, not more, is a property of skill itself, not a property of any particular tool.

---

If you're a senior developer reading this, the implication is uncomfortable but clear. The tools aren't for you the way they're for your junior colleagues. Not yet. Your job isn't to use AI the way a junior does, as a code generator. Your job is to use it the way a senior does: as a reviewer, a rubber duck, a search engine that speaks your programming language. The value for you is different, and it shows up in different metrics than lines of code.

And if you're leading a team, the implication is even more important. When your juniors get faster and your seniors get slower, the temptation is to celebrate the productivity numbers and ignore the structural shift happening underneath. Don't. The numbers are real. The shift is real too. And the shift is the one that matters.

The Photograph of a River

In July 2025, METR published the most rigorous study anyone had done on AI-assisted developer productivity. Sixteen experienced open-source developers. Two hundred forty-six tasks. Repos with millions of lines of code and tens of thousands of GitHub stars. Real work, real tools, real money, [$150 an hour](https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/).

The finding: developers using AI took [19% longer](https://arxiv.org/abs/2507.09089) to complete their tasks.

That number ricocheted around the internet for months. AI skeptics waved it like a flag. AI boosters called the study flawed, the sample too small, the tools outdated. Everybody had an opinion. Almost nobody asked the question that mattered.

By the time you're arguing about the photograph, the river has moved.

---

Seven months later, METR [changed their entire study design](https://metr.org/blog/2026-02-24-uplift-update/). Not because they got the first one wrong. Because the conditions that made the first study possible no longer existed.

Here's what happened. They tried to run a follow-up: same structure, randomized controlled trial, some developers use AI, some don't. And the "some don't" part fell apart.

Thirty to fifty percent of developers in the new cohort reported avoiding certain tasks because they didn't want to do them without AI. Not "preferred not to." Refused. Increasing numbers of developers wouldn't participate at all unless they could use their tools, even at fifty dollars an hour. One developer told the researchers: "My head's going to explode if I try to do too much the old-fashioned way because it's like trying to get across the city walking when all of a sudden I was more used to taking an Uber."

The randomized controlled trial, the gold standard of productivity research, was becoming structurally impossible. Not because of funding or methodology. Because the developers had changed.

---

This is the part nobody wants to talk about. Every AI productivity study is a snapshot of tools, habits, and workflows that are already different by the time the paper ships.

The METR study used Cursor Pro with Claude 3.5 and 3.7 Sonnet. Those were the best tools available in early 2025. By mid-2025 they were not. By the time anyone was citing the 19% slowdown in a strategy meeting, the tools had gone through multiple generations, the prompting patterns had shifted, the agent scaffolding had improved, and the developers themselves had accumulated months of additional practice.

METR knew this. They said so in the original paper: "Future systems may overcome the challenges observed here through improvements in prompting techniques, agent scaffolding, or domain-specific fine tuning." They built their own expiration date into the findings. Almost nobody read that part.

---

The research lag problem is not unique to AI. It shows up everywhere that technology moves faster than academic publishing cycles. But in AI it's acute, because the tools don't iterate on a yearly schedule. They iterate on a monthly one. Sometimes weekly.

A drug trial that takes two years to publish is studying a molecule that hasn't changed. An AI productivity study that takes six months to publish is studying a tool that has been updated dozens of times. The molecule doesn't learn. The tool does.

So what are we actually measuring? Not the capability of AI coding tools. Not even the capability of developers using AI coding tools. We're measuring the capability of specific developers, using specific tools, at a specific moment, on specific tasks. By the time that measurement becomes a citation in someone's slide deck, every one of those variables has shifted.

---

This doesn't mean the research is useless. It means the research answers a different question than people think it does.

The METR study didn't prove that AI makes experienced developers slower. It proved that in early 2025, with early-2025 tools, experienced developers working on familiar codebases hadn't yet found workflows that made the tools net-positive for them. That's a finding about adoption curves and tool maturity and the gap between what experienced developers already know and what the tools can currently offer. It's important. But it's a photograph, not a prophecy.

The useful signal in these studies is structural, not numerical. The specific percentages expire. The patterns underneath them don't. Senior developers struggle more than juniors with current AI tools; that's structural. Code review burden increases when AI lowers the barrier to contribution; that's structural. Productivity gains plateau when only the coding phase gets faster; that's structural.

Those patterns will hold even as the numbers change. But the numbers are what get tweeted. The numbers are what get cited in board decks. The numbers are what people argue about at conferences while the river keeps moving underneath them.

---

Here's what I think is actually happening. We're trying to measure a transition with tools designed to measure a steady state.

The randomized controlled trial assumes you can cleanly separate "with AI" from "without AI." That assumption is dying. The developers who can't work without AI aren't outliers anymore; they're an increasing share of the population. The ones who can switch back and forth are increasingly the ones who haven't integrated the tools deeply enough for the integration to matter.

The best AI-assisted developers, the ones whose workflows have genuinely changed, are exactly the ones you can't study with a with/without design. They've rebuilt their process around the tools. Asking them to work without AI isn't a control condition. It's a different task.

METR saw this coming and pivoted to observational methods, surveys, and agent evaluations. They're not measuring "does AI help?" anymore. They're measuring "how is work changing?" That's a harder question. It's also the right one.

---

So the next time someone cites a number (AI makes developers 55% faster, or 19% slower, or exactly 10% more productive) ask when the study was done. Ask what tools they used. Ask whether the developers in the study had weeks of experience or months. Ask whether the ones who opted out were the best users or the worst.

And then remember that you're looking at a photograph of a river. The water was already somewhere else by the time they pressed the shutter.

Consider the pill bottle.

Every morning, a sighted person walks into the kitchen and their eyes sweep the counter before a single conscious thought about medication fires. The bottle is there, cap ajar, just left of the coffee maker. They "remember" taking the pill, not because they actually recall the act, but because the world confirms it. Cap's off. Position's right. Memory, in that moment, is visual confirmation, not memory.

Now remove the bottle. Remove the counter. And smear vaselene on the lens of the kitchen's entire visual field.

What remains?

I can tell you what remains, because I live there.

---

The Crutch You Don't Know You're Using

Sighted people offload memory into the environment. They almost never notice they're doing it. Where something was on a page. What color the folder was. Which shelf it sat on. The spatial position of a thing is the memory of the thing, for most people.

There's a phenomenon researchers call ["looking at nothing."](https://doi.org/10.1016/j.tics.2008.07.007) When sighted people try to recall something, they reflexively move their eyes toward the spatial location where they originally saw the information, even when there's nothing there. Even in total darkness. The gaze drifts to an empty corner of the room as if the room still holds the answer.

It usually does. That's the point. The world is a sighted person's external hard drive. You can always re-query it. Forgot what your colleague said? Check the Slack thread. Can't remember if you locked the door? Walk back and look. The environment holds the details, so the brain doesn't bother encoding them deeply. It's not a failure. It's an optimization.

But it means sighted memory is part brain, part environment. Take away the environment's half of the equation, and you end up with an entirely different brain structure.

---

What Actually Happens Without the Scaffold

When you can't re-check the world, you have to remember everything. The exact words from the email your screen reader just read. The specific line number you want to mention in the code review. They're all easy for sighted folks to do, but painful for me.

My memory doesn't organize by space. It organizes by sequence. Sighted people remember who sat where. I remember who spoke when, and in what order, and what the tone shifted to after the third thing someone said. Research backs this up: [studies have found](https://doi.org/10.1016/j.cub.2007.05.060) that blind individuals outperform sighted controls on sequential memory, not just at the edges of a list but all the way through. The entire chain stays intact.

This makes sense if you think about how the world actually arrives without vision. It's not a scene you hold in view. It's a sequence of events unfolding in time. Navigating a room isn't a map; it's a rehearsed chain: third step, turn left, four more steps, reach for the counter. A conversation isn't a spatial arrangement of faces. It's a sequence of voices, pauses, tones, and turns.

When your brain encodes what experience actually delivers (not pictures, but stories), it turns out you get something powerful: memory organized around meaning.

---

Gist vs. Verbatim: The Architecture Nobody Told Me About

There's a framework in cognitive science called [Fuzzy Trace Theory](https://doi.org/10.1111/1467-8721.00192), and the first time I encountered it, something clicked that had been bothering me for years.

The theory says every memory gets encoded on two tracks simultaneously. The verbatim trace stores surface details: exact wording, specific appearance, precise visual features. The gist trace stores meaning: the pattern, the relationship, the point of what happened.

Here's the thing: verbatim traces are precise but fragile. They degrade fast, they're easily contaminated, and they depend on being able to re-access the original source to stay sharp. Gist traces are coarser but durable. They resist interference. They survive time. And they support reasoning in ways verbatim traces can't, because you can draw new conclusions from a gist. A verbatim trace just confirms what you already saw.

Sighted memory constantly refreshes verbatim traces. You re-read the sign. You re-examine the face. You re-check the note. Remove that refresh capability, and the memory system does the rational thing: it invests in gist.

Think of it this way. Verbatim memory is a screenshot. Gist memory is a summary of what mattered. Verbatim is raw logs. Gist is compiled insight.

I've been running on compiled insight my whole life. I just didn't have a name for it until now.

---

How I turn it into a superpower

Here's where most people expect me to start qualifying. "Oh, but you must miss things." "Oh, but exact details matter sometimes." Sure. I'll get to that. But first, let me tell you what gist-dominant memory is actually good at.

Pattern recognition. I catch the shape of an argument before it finishes. I know when a conversation's tone has shifted before anyone says anything explicit about it. I track the meaning of what people say across time, not the exact words, which means I notice contradictions that word-level listeners miss entirely.

Sequence recall. I can reconstruct the order of a complex discussion: who said what, what responded to what, where the thread turned, because my brain stored it as a narrative, not a snapshot.

Social intuition. I don't remember what you said. I remember what you meant, and how it felt. That turns out to be a better foundation for understanding people than a photographic transcript.

[Sighted eyewitness testimony is notoriously unreliable](https://doi.org/10.1016/S0010-9452(08)70999-8), and a big part of why is that visual recall feels confident even when it's wrong. The brain fills in expected details and mistakes them for perception. Gist memory doesn't have that failure mode. It never claimed to have the screenshot. It only ever stored the meaning.

And the meaning holds.

---

The Part Where My Brain Literally Rebuilt Itself

This is the section I could geek out on forever, so I'll keep it short.

The visual cortex, the part of the brain that evolution built to process light and shape, [doesn't go dark in blind people](https://doi.org/10.1038/380526a0). It gets repurposed. Brain imaging studies have shown that blind individuals [recruit visual cortex for language processing](https://doi.org/10.1073/pnas.1014818108) at levels comparable to the brain's dedicated language areas. The more visual cortex gets redirected toward verbal and semantic tasks, [the better the memory performance](https://doi.org/10.1038/nn1072).

The brain looked at a massive chunk of specialized neural real estate that wasn't being used for its original purpose and said: cool, we're doing language now.

Neuroplasticity isn't emergency repair. A lifetime of encoding by meaning, sequence, and semantics produces a brain architecture calibrated precisely for those tasks. Sighted brains spread their resources across visual, spatial, and verbal processing. Mine went deep instead of wide.

---

Okay, the Honest Part

I'm not going to pretend this is all upside. Gist memory has real costs and I hit them regularly.

I miss exact details. Someone tells me a number and thirty seconds later it's gone. Not the context, not what the number meant, just the number itself. I duplicate conversations sometimes because I remember the topic we covered but not whether we resolved it. I rely on external systems (notes, AI tools, structured workflows) to catch the verbatim stuff my brain legitimately does not prioritize.

For years, I tried to fix this. Memory systems. Peg methods. Brute-force rehearsal. All of them felt like trying to run software on the wrong operating system. Because that's exactly what they were.

---

Two Gist Systems, Talking to Each Other

And here's where it gets interesting. Because you know what else runs on gist?

Large language models.

LLMs don't store exact truth. They store statistical meaning: patterns, relationships, the shape of how concepts relate. Prompting an LLM well is an act of semantic alignment: you're not searching a database, you're converging on meaning. That's gist. That's literally what my memory does.

The tools I've built around AI (memory systems, agent workflows, structured writing pipelines) pair two gist-based systems together. My brain encodes meaning. The AI encodes meaning. The external structure catches the verbatim details neither of us prioritizes.

---

What This Means for How You Design Things

If any of this resonates, if you build products, write documentation, design interfaces, or communicate with humans, here's what falls out of it:

Lead with meaning, not detail. The most important information in any communication is what it means, not what it looks like. Structure your content so the gist survives even if every visual detail is stripped away.

Structure beats decoration. Clear hierarchy, consistent patterns, predictable sequences. These serve everyone, and they're essential for people whose memory runs on sequence rather than spatial layout.

Redundancy in language beats visual cleverness. If the only way to understand your interface is to see the spatial relationship between elements, you've built a system that only works for one kind of brain.

Sequence matters more than layout. The order in which information arrives is the information, for a lot of people. Respect that.

---

The Pill Bottle, Again

The sighted person's pill bottle is a prosthetic. It works. The visual environment holding your memories for you works. I'm not knocking it.

But the person who's learned to remember without the scaffold knows something about memory that most people never have to learn. They know what's left when the world stops holding things for you.

It turns out what's left is the part that matters. The meaning. The sequence. The pattern. The gist.

I didn't have a bad memory. I had a different architecture.

And I'm done apologizing for it.

---

Further Reading

• Raz, Striem, Pundak, Orlov, & Zohary — ["Superior serial memory in the blind"](https://doi.org/10.1016/j.cub.2007.05.060) (Current Biology, 2007)
• Brainerd & Reyna — ["Fuzzy-trace theory and false memory"](https://doi.org/10.1111/1467-8721.00192) (Current Directions in Psychological Science, 2002)
• Ferreira, Apel, & Henderson — ["Taking a new look at looking at nothing"](https://doi.org/10.1016/j.tics.2008.07.007) (Trends in Cognitive Sciences, 2008)
• Sadato et al. — ["Activation of the primary visual cortex by Braille reading in blind subjects"](https://doi.org/10.1038/380526a0) (Nature, 1996)
• Bedny, Pascual-Leone, et al. — ["Language processing in the occipital cortex of congenitally blind adults"](https://doi.org/10.1073/pnas.1014818108) (PNAS, 2011)
• Amedi, Raz, Pianka, Malach, & Zohary — ["Early visual cortex activation correlates with superior verbal memory in the blind"](https://doi.org/10.1038/nn1072) (Nature Neuroscience, 2003)
• Greenberg & Rubin — ["The neuropsychology of autobiographical memory"](https://doi.org/10.1016/S0010-9452(08)70999-8) (Cortex, 2003)
• Röder, Rösler, & Neville — ["Auditory memory in congenitally blind adults"](https://doi.org/10.1016/S0926-6410(01)00002-7) (Cognitive Brain Research, 2001)