Subtraction Was the Job

The first product my studio shipped no longer exists. Neither do the next several — a comparison-shopping site, a couple of command-line tools, a document generator, a bot. Over a year I started more things than I can now name from memory, and the record of the studio is, more than anything it built, a record of what it removed: whole products retired, shared libraries folded back into the apps that used them and deleted, two front-end frameworks dropped for things the browser had quietly learned to do on its own, an entire layer of internal tooling taken out in the weeks after it stopped earning its place. Close to a fifth of every change I made was a deletion. I did not plan it that way. It is simply what the work turned out to be.

Everything you build, you carry

This runs against the instinct you begin with, which is that a studio is the sum of what it builds. You ship a product, then another, then a library to share what the two have in common, then a tool to manage the libraries — and each addition feels like progress, because each one is, locally, a thing that now exists where nothing did. The trouble is that everything you add also has to be carried. A shared package is a standing promise to keep two apps agreeing with each other forever. A framework is a bet that its way of doing things will outlast the platform underneath it. A clever abstraction is a structure that someone — usually a later, more tired version of you — has to hold in their head before they can change anything near it. The bill for all of it comes due not on the day you write it but every day after, quietly, as the set of things you can no longer move without first moving six other things.

A studio is mostly a graveyard

The products were the same lesson at a larger scale. A few earned their keep; most did not, and the discipline that mattered was never the building. Building was easy, and over the year it got easier, until it was close to free. The hard part was looking at something I had made and concluding that its continued existence was a cost I was paying out of habit. A studio, run honestly for long enough, is mostly a graveyard. The interesting question is not why so much of it died, but why the instinct to keep things alive is strong enough that killing them feels like failure, when most of the time it is just accounting.

When adding is free

What makes this sharper now than it would have been a few years ago is that the cost of adding has collapsed. When a capable model will produce a working version of almost anything you can describe, in an afternoon, the constraint that used to discipline a builder — that building was slow and expensive, so you only built what you were fairly sure of — is gone. Adding is nearly free. And when adding is free, the only scarce thing left is the judgment about what not to add, and what, having added, to take back out. The bottleneck moved. It used to sit at construction; it now sits at subtraction, and almost none of the tooling or the instinct of the field has moved with it. The whole conversation is still about what to build next.

An older name for it

There is an older name for the thing I keep relearning a package at a time. The contemplative traditions called it the via negativa — approaching a thing by removing what it is not, on the suspicion that the additions obscure more than they reveal. The Japanese sense of 間, the interval, treats the empty part of a composition as the part doing the work. Both are pointing at what an operator discovers the dull way. Whatever you accumulate, you slowly lose the ability to question — because each accumulation quietly turns into a premise, and premises are invisible. Subtraction is how you keep the premises few enough to still see them. I don't have a tidy method for it. What I have is a year of evidence that the studio got better roughly in proportion to how much of it I was willing to remove — and a working suspicion that this, and not the next thing to build, is the actual skill now.