Also in 中文. Three-part series: The pattern · Why it happens · How to judge.
There is a kind of work that feels exactly like progress and is, in fact, the most expensive way to avoid it.
You have probably seen it, maybe in yourself. The engineer who keeps refactoring the codebase and never quite gets around to facing a real user. The person whose note-taking system is immaculate — everything tagged, cross-linked, filed — who has not written the first sentence. The founder who builds a clean deployment pipeline for a product that has no users yet.
They are all busy. They all advance every day. And every day they drift a little further from the thing that actually matters.
Why we flee toward building
Because building is comfortable, and the thing that actually matters usually is not.
Building is controllable, legible, and quick to reward: you write code and it runs, the structure visibly improves, today is tidier than yesterday — a certain, immediate payoff. The work that actually moves the goal closes its loop in weeks, if at all. You put something out and hear nothing. You wait on a signal that refuses to arrive. The effort won’t reveal whether it worked until much later, and sometimes the answer is that it didn’t.
Anyone who can build will instinctively flee toward the first thing. Not because it matters more, but because it hurts less. The feeling of having your hands full is exactly what covers the line you would rather not say out loud: I am not ready to do the real thing.
The kind that looks like work
We picture procrastination as lying on the couch. That kind is easy to quit; you know you are slacking.
The kind that is genuinely hard to quit is the kind that looks like work. It ships. It has commits. It has a progress bar, and it lets you say, with a straight face, that you are “laying the foundation.” You feel productive every day, you move forward every day, and the direction merely happens to route around the thing you were supposed to do. It has KPIs, an architecture diagram, a Kanban board — the most respectable kind of procrastination there is, and the most expensive, because you never notice you are paying.
A few plain rules
Quitting this is not a matter of willpower but of a few rules, simple to the point of sounding dumb, that nonetheless hold.
First, ends and means. The outcome you actually want is the end; the code, the tools, the process you are polishing are all means. A means has no “is it good,” only “is it moving me toward the end.” The moment a means starts asking to be finished or made elegant, it has turned from servant into master. The test is simple: make it more perfect — are you any closer to what you wanted? If not, stop.
Second, attention is the only thing truly scarce. Money can be re-earned, materials re-bought, code left alone does not rot — those recover, or can wait. Today’s attention cannot: the afternoon you spend on A can never be spent on B. It is the only truly zero-sum quantity in the whole system, and the only one worth being stingy with. Yet we keep spending it to optimize the things we already have plenty of, and call it hard work.
Third, only the bottleneck counts. Everything has one link that constrains it. If the goal is stuck on nobody knows, nobody trusts, it isn’t built yet, then effort poured into a link that isn’t binding moves the output by nothing — a big number times zero is still zero. Find the binding link first; ask what it is stuck on right now, not what you happen to be good at.
Last, making is one-time, keeping is forever. People think AI dropped the cost of building to the floor, so they should build more. Half right: it cut the cost of making, not of keeping — the maintenance, the midnight debugging, the small permanent tax on your attention. None of that fell.
Anything you can generate in a night, you can spend a lifetime maintaining.
So treat making as free and count only what keeping will cost you in attention over time. Then one more question: would fifteen manual minutes, or the smallest stopgap, get you eighty percent of the way? If yes, don’t build it.
But isn’t this just sharpening the axe?
By now you probably want to push back: some preparation really does matter. Sharpening the axe doesn’t waste the woodcutter’s time. Laying groundwork, clearing obstacles before the real work — that is simply the right thing to do. And telling someone to never prepare and just charge in is its own kind of stupidity; it has a name, recklessness, and it ends at the same wall.
True. So the real question was never build or don’t build. It is whether the thing in front of you is real preparation or avoidance wearing preparation’s clothes. From the outside the two are identical — both hands-on, both producing something, both keeping you busy. The difference is not in the activity but in its relationship to the real work. Four questions separate them:
- Is there a line called “enough”? Real preparation is bounded — I need this to take the next step, it takes two days, then I take that step. Avoidance is unbounded: there is always one more thing to perfect, and “done” keeps receding. Before you start, can you say where you’ll stop? If not, it’s avoidance.
- Is a real action waiting right behind it? Behind real preparation sits a specific, committed next move; behind avoidance sits more preparation. When this is finished, do you walk into contact with something real, or merely unlock the next round of setup?
- Does it clear the link that’s binding now, or an imagined future one? Good prep removes the bottleneck you have today; bad building solves problems you don’t have yet — “what if it has to scale.”
- Does it get you to reality sooner, or later? The thing any early effort most needs to know is whether it works, whether anyone wants it — and that can only be answered by contact: shipping, showing, asking. Good preparation is the minimum needed to make that contact happen; anything beyond that minimum is mostly postponing the answer.
The proverb hides two conditions: sharpening the axe only saves time if the axe is actually blunt and you are actually about to chop. Sharpen an already-sharp axe a tenth time, or with no wood in front of you, and the sharpening is the waste. So the honest version is not “don’t prepare.” Over-preparing and under-preparing are opposite diseases, and both are fatal; the skill is calibrating preparation to the minimum that gets you to the real test.
A question to carry
Compress all of it into one line and keep it on you:
This hour — can it move the outcome I actually care about forward, this week? If yes, do it. If it only makes the tools around the outcome nicer, set it down.
That one line guards against both failures at once: working where nothing is constrained, and mistaking the means for the end.
Freeze, don’t delete
One more, for anyone who has already built too much: when you’re not sure, don’t delete it and don’t keep building it — freeze it. Deleting is too final; building on is too expensive in attention; freezing is the cheap option in the middle. Leave it where it is, stop letting it tax you, and keep the option to revive it the day it is genuinely needed. Let time and new information decide, instead of this moment’s mood.
The hard part
The hard part was never the building. Building is easy — easy enough to hide inside. The hard part is not building: doing the slow, illegible, rejectable work instead, and then being patient with it.
If you are the kind of person who likes to make things, the next time the thought “let me get the setup perfect first” arrives, pause for three seconds and ask the one question above. More often than not, you are just trying to retreat somewhere comfortable.