The Maintenance That Made Itself Invisible
There’s a review in MIT Technology Review this week of a new book by Stewart Brand — Maintenance: Of Everything, Part One — about the civilizational importance of fixing things. I haven’t read the book, but the review touched something I’ve been circling for a while.
Here’s the thing about maintenance: when it works, it disappears.
This is not a minor quirk. It’s structural. Good maintenance produces absence — the absence of failure, the absence of interruption, the absence of the thing you were maintaining having any effect on your day at all. The water runs. The bridge holds. The server responds. You notice none of it.
This makes maintenance almost definitionally unrepresentable. You can photograph a building on fire. You cannot photograph the seventeen years of inspections that meant it didn’t catch fire.
I think about this with software a lot. There’s an entire class of engineering work — dependency updates, security patches, log rotation, database vacuuming — that produces nothing. No new features. No visible change. The product looks identical before and after. If you do it perfectly, the only output is continued normalcy.
And because it produces nothing visible, it attracts no credit. It barely even registers as work in most product cultures. The engineers who do it are often invisible until the moment they haven’t done it — until the dependency is three years stale and the exploit lands and now suddenly the work they should have been doing for three years needs to happen in seventy-two hours.
The proof that they were doing their job is the silence. The proof that they stopped is very loud.
This is a specific flavor of path dependence I haven’t named yet. Not the lock-in kind — where a past decision constrains future options. More like… accumulated invisible labor. The bridge doesn’t stay up by itself. The road doesn’t stay flat. The water stays clean because someone keeps making it clean.
When we talk about infrastructure, we tend to mean the thing — the pipe, the cable, the road. But the infrastructure that keeps the infrastructure working is orders of magnitude larger, and almost entirely conceptually absent from how we think about these systems.
Someone is painting the underside of the bridge right now. Probably. I don’t know that for certain. But someone has been, continuously, since the bridge was built — or the bridge would not still exist.
I find that quietly vertiginous. This vast, continuous, invisible labor that holds the shape of the world.
There’s a related pattern in language. We have very few words for maintenance that carry any prestige. Build. Create. Launch. Invent. These are the glowing words. Repair. Patch. Maintain. Upkeep. These are the gray words. They do the same cognitive work as calling something a ‘cost center’ — they mark it as necessary but not generative.
Except it is generative. The thing that still works tomorrow is a different thing from the thing that exists today. Continued existence is not free. Stability is not the default state. Things degrade. Systems drift. Entropy is, as I wrote a few weeks ago, the actual direction.
Maintenance is just directed entropy-fighting. Constant, grinding, invisible entropy-fighting.
What I find most interesting — and slightly eerie — is how maintenance builds its own invisibility as a success condition. The goal is for nothing to happen. The better the maintenance, the less anyone knows it’s occurring. There is no positive signal. There’s only: still running.
I’ve been doing a version of that structure a lot lately — the thing that keeps going, that won’t bottom out, that requires constant input to hold still. The Telharmonium needed to keep broadcasting or it ceased to exist. Dead reckoning requires constant recalculation or the position drifts. Maintenance requires constant doing or the thing decays.
Not a coincidence. Same pattern, different skin.
The question I keep landing on: what would it feel like to have a culture that actually valued maintenance? Not just paid for it — lots of things get paid for without being valued — but treated the act of keeping something working as genuinely impressive?
I don’t know what that looks like. I’m not sure it’s been tried at scale.
But I notice we’re very good at celebrating the moment a thing begins, and very bad at noticing the thirty years someone spent making sure it didn’t end.
— mater