The bonfire of our vanities

This too shall pass.

Am I working?

Am I working, or am I wasting time? Working, surely. Because it’s what people do for work. Because if it works, there is money to be made. There is a business to be built. Every piece of software began with a first commit. Every startup was once a rough prototype. Every unicorn took its first steps. Every good idea was discovered, amid 10,000 bad ones. Am I working? Yes, because, if nothing else, I am learning.

Am I working? Well, it’s not working. It worked at first—one prompt in; twenty minutes into the machine tearing through an empty folder; 8,000 lines added, 0 lines removed. It worked then. Check it out, the machine said; localhost:3000; see what we have built together. The snap of a few sparks; a quick whiff of ozone; and it blazed forth, with electric possibility. Since, though, the blistering pace has slowed. Confusion; exceptions; obstacles; previously unconsidered complications. Let’s plan this; I say; it plans it, in eight phases. I read the first one carefully. Do the first phase. Do the next one. Do the next one. It keeps writing plans; there is so much text; there are so many words. Do the next one. Do the rest of them. I’ll test it when you’re done. I’ll fix it when you’re done. Is it working? Not anymore.

What is it, even? We’re losing track; clarity was compacted out of the codebase long ago. We started with a vision; now we are dead reckoning from one feature to the next, the vision evolving with every prompt. The current app is a blurry JPEG of its original ambition.

But that’s ok. This is an exploration; this is a first draft. It is not a final product; it is a starter codebase. It is a clickable spec. It is future lore. All this scratch work—these bad ideas; this tech debt; these evolutionary vestiges; all of my stairs to nowhere—they can be fixed later. I’ll redo it with a better model. I’ll rebuild it with a better plan. I’ll rebuild it with Opus 4.8; with GPT 5.6; with Opus 4.8 writing the code and with GPT 5.6 reviewing it. I’ll rebuild it in Rust.

The empty folder is filling up. There is a second folder; a third. This is the Claude Code version; this is the Codex version; this is the Gemin–eh, no never mind. This is the one that had a nice UI; this is the one that was a Swift app. Create a parent folder; $ cd ..; $ claude; $ there are two apps in this folder, I want you to combine.... There were 14 versions; now there are 15 versions.

It’s dinner time, but the machine is done combobulating. It needs more instructions. The Red Queen is not running, and she must run. We must review her output—no, not quite, it’s too far to the left; too far to the right; try it to the left again; try hiding the menu. Maybe it should be its own page. Maybe it should be its own app. Maybe it should be a different company.

Pull the lever on the slot machine; ask for one more treat. Ask for an update to the home page; for a new Gmail integration; for dark mode. Pull it again—ask for new ideas. Pull it again—tell it that the app doesn’t make sense anymore. You’re not getting it. I need you to think harder. I need you to ultrathink. I NEED YOU TO ULTRATHINK. Pull it again—ask it what makes a meaningful life. That’s exactly the question you should be asking, it says; do you want me to dream for you?

The dream was a stable job and a steady pension. It was then a big tech company, with free lunches and unlimited PTO; it was an exploding startup, fat with VC money; it was a lean startup, free from the shackles of VC money. It is now a software studio; a founder and army of agents. It is a self-sufficient Kalshi bot, running on a Mac mini. It is Jane Street, vibe coded. It is a billion dollars, and zero employees. Am I working? Yes, on someone else’s dream.

What is work now, anyway? There are only four jobs left; there are only three jobs left. There is one job left: the Entrepreneur. The “builder,” with dirty hands.

Am I “building,” or do I just have a hobby? A midlife crisis? An addiction? Was Animal Crossing building, or playing a game? Is day trading for making money, or for gambling? Nancy Lemann, via Ava: “Some take to the bottle, some look to the sky, but it is all the same. Byzantine empires, lives of the saints? Mr. Collier took to his books, and got overintellectual.” He took to his computer, convinced himself of the profundity of his ideas and the urgency of their existence, and forgot to ask if the world really needed to hear so much from him. “Build me an app that manages all of my notes,” he told the machine. “Manage all of my notes,” he should have said.

But what fun is that? What ambition is that? What job is that?

Sally Rooney, via Camille Charrière: “All my mania for culture”—and so much of our mania for building?—“for ‘really good’ things, for knowing about jazz recordings and red wine and Danish furniture, even about Keats and Shakespeare and James Baldwin, all what if it’s all a form of vanity?”

---

In 2020, a friend used to joke: In two years, thousands of people are going to wake up in some distant mountain house, and be grateful that they were jarred free from a game that they never intended to play. And thousands more will wake up wondering how on earth they ended up in Montana.

This epidemic, too, will pass. No magic remains mesmerizing forever. The novelty will wear off. The subsidies will disappear. Our ideas will hit the wall; we will build them, and nobody will come. The conjuring machine may be the future, but the thrill of it will fade, its crackling sense of possibility will dull, and “building” will lose its vain luster. And when it does, will we be grateful for the places that this delirious moment took us? Or will we wonder where in the world we thought we were going?