I fixed my father’s watch, it fixed me back

Nearly 30 years ago, I handed my father’s watch to a professional watchmaker for a routine service.
When I came back to collect it, he put it on the counter without explanation. And to my horror, the crystal was shattered. The hands were missing. The dial was scratched, the case bent. Irreparably damaged, he told me.
I was twenty years old.

Two years ago, remote and isolated and running out of reasons to leave my chair, I decided to fix it myself.
I had no idea what I was doing, and that was the point.

The weight of the drawer #

My mother gave that watch to my father in the mid-80ies. Then my father gave it to me in 1996, when I left for engineering school, and I understood even then that he was trusting me with something he valued.

Over the years I kept looking for someone who could undo what had been done to it, reaching out to watchmakers and the manufacturer, and never quite managing to hand it over. What had happened once was enough, so I’d put it back in the drawer and try to stop thinking about it.

Then five years of remote work stripped everything else away, and the drawer became impossible to ignore .
I stumbled into watch restoration channels on YouTube, out of curiosity more than anything, and could spend hours just memorizing how they handled things. Then I ordered a set of tools from AliExpress for less than 15 euros, and told myself that if I broke it further, I’d only have confirmed what the watchmaker told me 30 years earlier.
That felt like a manageable risk. It probably wasn’t.

Five years from bedroom to desk #

Those years were a slow subtraction.
The commute went first, then the office, then the colleagues, then the casual contact that goes with them: a coffee, a corridor conversation, someone to eat lunch with.
What was left was a home office, a screen, the particular anxiety that comes with spending five years performing presence on a rectangle while the world happens without you, and work that had started to feel like it had forgotten what it was for.

One evening I took the caseback off. The oil had dried to a film, and the movement smelled of 30 years in a drawer. I held it up to the light and I could see the scratches on the dial, the heavy marks on the case, the empty post where the cannon pinion should have been.
Now I understood exactly what had been done to it.

The somewhere else #

Looking at it under the light, I knew this wasn’t going to be a quick fix. Another side project or online course would have kept me in the same loop, and the loop was already the problem. What was missing was consequence: something that would tell me, unambiguously, whether I had succeeded or not.

What I was looking for before I knew I was looking for it was a problem that pushed back. A broken escapement doesn’t care about your seniority, your sprint velocity, or the elegance of your architecture. It either runs or it doesn’t, and it will tell you which.
You cannot ship a broken one and fix it in the next sprint, and software forgot that somewhere.

Value and the feeling of value are not the same thing. The craft solved the second problem, not the first.

A credentialed professional destroyed this watch. An amateur with YouTube and a set of cheap tools is attempting to restore it.
Physical reality, it turns out, is indifferent to credentials.

The cost of the craft #

I found the parts on eBay. The manufacturer reused the same components across models, so I bought two donor movements for around 10 euros, to strip for the hands and the cannon pinion. The crystal and strap I ordered new. I also sourced the right tools and oils. It took me more than three weeks to get everything together.

When everything arrived, I cleaned the movement with a soft brush and lighter fluid, the way the tutorials say. I oiled the jewels. I stripped the hands from the donor movements and cleaned them.

Then I installed the cannon pinion, which took longer than everything else combined. I had to dismantle the movement, keep track of every screw, remember the order to put it back, not damage anything else in the process, and then replace the pinion. Four attempts before it sat right.

At some point I stopped checking the time. The stakes were clear enough: one wrong move and I’d have confirmed what the watchmaker told me thirty years earlier. That kind of focus is hard to manufacture, but it’s easier when the thing you’re trying not to break belonged to your father.

Then the hands. Press too high and they catch the dial, too low and they drag on the crystal.

First attempt: the minute hand was off by 15 minutes. I took it off and reset it.
Second attempt: the hour and minute hands caught each other at twelve. I took it off again.
Third attempt…

I have done things that were harder than this. But I have not, in recent memory, done many things that required me to fail this many times before I was finished, and to treat each failure as just the next step.

In software, failure tends to mean a bug, a rollback, a post-mortem, endless taskforce meetings, and a document explaining what happened and what we will do differently.
The document is sincere. So is the discomfort underneath it. In watchmaking it just means you try again, and the movement doesn’t hold it against you.

The dial came off once. Briefly. I don’t want to talk about it.

The watch runs #

It ran.

I don’t have a more dignified account than that. After the cannon pinion, the hands, the resets. It ran, and I sat there looking at it for longer than I’d like to admit.

I had done it. Not the credentialed professional who broke it. Me.
Software hadn’t let me say that in years.

I called my father. He needed a moment to remember. When it came back to him, I told him I had fixed it. He said he was proud of me.
I had spent 30 years waiting for that sentence. When it arrived, I didn’t know what to do with it.

I wear the watch sometimes.

Watchmaking gave me back consequence, not purpose: a problem with a real answer, feedback you cannot argue with, a result you can hold.
I had been performing all three of those things in software for years without noticing they had gone.
The specific danger of software work is that you can simulate all three convincingly enough that even you stop noticing the difference. The meeting looks like a problem. The deployment looks like a result. The retrospective looks like feedback. Nothing tells you it isn’t real.
That is how you lose your footing without a crisis. One slow subtraction at a time, until the only thing left is the performance.

The watch ran. I hadn’t realised how much I had been pretending until it did.