I built a system that scans job listings every day at noon, clusters them by relevance, and surfaces anything worth looking at. I built another that pulls from 145 AI news sources overnight, runs them through an analysis pipeline, and leaves a brief on my desk by morning. Calendar, reminders, deadlines — all funnelled into a single command I can run any time: what should I be doing right now?
And this morning, instead of reading the brief the system produced, I spent an hour refining how the brief gets delivered.
Oliver Burkeman would recognise this immediately. In Four Thousand Weeks, he argues that the entire productivity project is a sophisticated anxiety displacement — we build systems to manage our finite time, and the system-building itself consumes the time we were trying to protect. The to-do list becomes the work. The meta-work eats the work.
He’s partly right. There’s an anxiety that no system can solve: the fundamental unease of having more possibilities than hours. “Is there something I should be doing?” is not a question with an answer. It’s the sound of a finite mind confronting infinite options. No amount of automation makes that feeling go away.
But Burkeman underplays a distinction I think matters: cognitive load is not the same thing as existential anxiety.
When I automated the job scan, I didn’t solve career uncertainty. I still don’t know if the right role exists, or if I’d recognise it, or if the timing will ever feel right. What I eliminated was the friction — the twenty minutes of scrolling LinkedIn, the nagging “I should check,” the context switch that breaks whatever I’m actually doing. The anxiety persists. The busywork doesn’t. That’s a real win, not a Burkeman shell game.
The honest test is simple: am I using the system, or am I building the system?
If the job scanner runs and I review its output in two minutes while walking to the office — that’s friction eliminated. If I spend the walk redesigning how the scanner formats its output — that’s anxiety displaced. Same tool, different relationship to it.
The pattern I’ve landed on is this: a system is done when it removes the friction. Not when it’s elegant, not when it handles every edge case, not when it sparks joy. When the thing that used to take twenty minutes now takes two, stop. The remaining unease isn’t a bug the system can fix. It’s just what it feels like to be a person with limited time and unlimited wants.
Burkeman is right that you can’t optimise your way out of finitude. But you can stop wasting your finite hours on tasks a machine handles better. The trick is knowing which feeling you’re responding to — the friction, or the fear — and having the honesty to stop building when only the fear remains.
I closed the laptop on the delivery pipeline this morning and did the quiz that was actually due. The system worked. The temptation to keep perfecting it was the part that didn’t.