Someone Still Has to Fix the Machines

Every utopia I have ever read is described in the present perfect. The gardens have been grown. The scarcity has been solved. The old conflicts have been laid down. You arrive as a reader at the moment after the work, into a world holding its finished shape, and the book invites you to admire it.

What almost none of them describe is Tuesday.

Not the founding. Not the crisis. Tuesday — the ordinary middle of an ordinary week, when nothing is being built and nothing is collapsing, and somebody still has to go check on the machine that keeps the other machines running. When I started building the world of the Utopia trilogy, that was the question that kept ambushing me, and I could not find a satisfying answer in any of the blueprints I admired. Who is on call? Who carries the pager in paradise?

I have come to think that the silence around that question is not an artistic choice. It is the design flaw itself.

A utopia is a claim about maintenance

Strip the poetry away and every vision of a perfected society is making one of two claims. Either the systems that sustain it require no upkeep, or the upkeep has been arranged in a way the author would prefer you not examine closely.

The first claim is a fantasy, and an unusually revealing one. Nothing that does anything is maintenance-free. Gardens go feral. Code rots against a world that keeps moving. Institutions drift from the purpose that created them within about a generation, and drift faster if nobody is watching. Even the elegant part — especially the elegant part — degrades, because elegance is usually a set of assumptions written down at one moment in time, and the moment does not hold still.

The second claim is where the trouble actually lives. Somebody is doing the upkeep. If a society is described as effortless, the effort has not vanished. It has been moved somewhere out of frame: onto a machine that is itself assumed to be perfect, onto a class of people the narrative does not follow home, or onto a future generation that has not yet been asked. Every utopia that reads as frictionless is, on close inspection, a utopia with a rendering problem. The load-bearing wall is there. It has just been painted the same color as the room.

What I learned building companies, and then had to un-learn writing fiction

I have spent forty years building things that had to stay up. Payments systems, mostly, and then other things that moved money or held trust, which is the same discipline wearing different clothes. And the single most consistent structural feature across every one of them was this: there is a visible layer that gets the credit, and an invisible layer that keeps the visible layer standing.

The visible layer ships features. It demos well. It appears in the fundraising deck and the press release and the org chart. The invisible layer handles the reconciliation that did not balance, the vendor that changed an API without telling anyone, the batch job that has silently failed for three days, the migration that has to happen at the hour when the fewest people are awake to be hurt by it.

Here is the cruelty built into that arrangement: reliability is invisible by definition. When the maintenance layer is doing its job, the observable output is nothing happening. No outage, no incident, no story. Which means the people responsible for nothing happening are structurally underweighted in every conversation about who matters — right up until the single week when they are the only thing that matters, and by then it is far too late to have been paying attention.

I got this wrong for years, in the ordinary way founders get it wrong. I over-rewarded the launch and under-rewarded the long quiet stretch afterward, because the launch was legible and the quiet stretch was not. The correction was not a policy. It was learning to ask a different question in reviews: not what did we ship, but what did we keep from breaking, and who did it.

The generation that inherits the maintenance

The version of this that I find genuinely hard, and that fiction is unusually good at holding, is the handoff.

The people who build a system understand it in their bones. They know which part is load-bearing and which part is decoration, which rule exists for a reason and which one is a scar from an argument in year two. None of that is written down, because to the builders it is not knowledge — it is just how things obviously are.

Then those people age out. The next generation inherits a working system and a set of rules with the reasons removed. They maintain it faithfully for a while, and then, reasonably, they begin to prune what looks decorative. Sometimes they are right. Sometimes the decorative thing was the only reason the whole arrangement had held for thirty years, and nobody alive can say why.

That is not a plot device. That is how real institutions fail, and it is much sadder than sabotage. No villain, no collapse — just a slow forgetting, carried out competently by people doing their best with what they were handed.

The test I now apply to any perfect world

When someone shows me a design for how things ought to work — a society, a platform, an organization, a product — I have stopped asking whether it would be good if it worked. That question is too easy; the answer is almost always yes, which is why so many bad designs survive first contact with a room.

The questions I ask instead are all maintenance questions. Who is on call when it fails outside office hours? What happens on the ordinary Tuesday when the one person who understood the old subsystem is on vacation? How does this thing get repaired by people who did not build it and do not love it? What does it cost to keep, not to launch — and is anyone honestly budgeting for that cost, or is it being quietly assigned to somebody who has not been consulted?

A world is not what it can produce on its best day. It is what it can sustain on its most ordinary one. Every utopia is a promise about the best day. The ones worth taking seriously are the ones willing to describe Tuesday.