Consistency is architecture, not virtue
What kept WebTranslateIt going for 18 years wasn’t discipline or willpower. It was boring systems that did the long term for me.
People sometimes ask me how I kept WebTranslateIt going for 18 years. The flattering answer would be discipline, willpower, some rare talent for the long game. The honest answer is that I’m not especially good at the long term. Nobody is. What I got good at was building systems that do the long term for me.
Humans are bad at the long term
We overvalue what happens today and underestimate what small efforts add up to over years. It’s why diets fail in March, why savings plans fade, and why most side projects die quietly after a few months. That isn’t a character flaw, it’s how we’re wired.
What actually kept WTI alive
When I look back at 18 years of running WebTranslateIt mostly alone, the things that kept it alive weren’t heroic efforts. They were boring systems.
Security checks run in CI on every commit, so I don’t have to remember to be careful. Monitoring and automatic restarts handled most night-time problems for years, and these days AI agents watch the servers too, so a 3am problem rarely depends on me waking up. And the product itself wasn’t built in one big leap. It’s the result of thousands of small improvements, each one too small to be impressive and each one making the next one easier.
None of this required me to be a better person. It required me to stop relying on being a better person.
Small steps, compounded
If I had to keep one lesson, it would be this: big changes that need a revolution usually fail. Small changes that everyone can accept, followed by another, and another, end up producing enormous change over 20 years.
This is true in a codebase, in a company, in a life.
Build the system, not the willpower
The recipe is the same whether it’s a codebase, a company or a personal habit.
Automate the right behaviour, so it happens by default instead of by effort. Measure continuously, so you know where you are before it’s too late. Move in small, predictable steps that people accept. And never go backwards, because the whole thing only works if the steps compound.
I don’t think humans will suddenly get better at the long term. I think the ones who succeed are the ones who stop trying to be heroic and start building systems that make the right thing the easy thing.
So pick one thing you keep failing at through willpower alone, and ask what system would make it happen by default. Consistency isn’t a virtue. It’s architecture.
Málaga, 9 November 2026.