The Quiet Benefit of Doing Things Yourself

There is a peculiar kind of knowledge that cannot be Googled or Chatgpt'ed, summarised, or outsourced, and I say this as someone who has, at various points in his life, tried to do all three (usually in that order, usually at midnight, usually while something was already broken).

The knowledge I mean is the kind you get from doing something yourself. Not expert knowledge — the plumber who comes to fix your pipes has forgotten more about plumbing than you will ever learn. Not complete knowledge, either. But something more useful than both: a working model of how a thing actually functions, acquired firsthand and lodged in a place that reading about it simply cannot reach.

I've noticed this most with software, which is the obvious example and also the honest one. People who've written even a small amount of code — a script, a function, one terrible afternoon with a tutorial — have fundamentally different instincts about software products. They know, without being told, which features are probably held together with string and optimism. They understand why that button is grayed out, why that loading spinner has been sitting there for four seconds, why "we're working on it" sometimes means three weeks and sometimes means never. They are better users of software not because they are smarter, but because they have a model of what is going on underneath. The people who built the product can feel it when they walk in.

The same logic applies everywhere. Making a budget, even once, changes how you read your own spending; not because budgets are accurate (they are famously not) but because the act of constructing one forces you to reckon, in precise numerical terms, with where the money actually goes (the answer is always slightly embarrassing). Cooking your own food changes how you evaluate a restaurant — suddenly the pricing makes sense, the speed makes sense, the occasionally exhausted quality of a Thursday night service makes complete sense. Writing something yourself — anything, an email, a document, an explanation of a thing you half-understand — changes how you read other people's writing, because you know, from recent personal experience, how difficult it is to say a thing clearly and have it land.

Here is the counterintuitive part, and I think it gets undersold: the floor of understanding you acquire from doing something once is higher than the ceiling of understanding you acquire from merely reading about it. Reading gives you vocabulary. Doing gives you intuition. These are not the same thing, and when you need to make a decision — about what to build, what to fix, what to delegate, what to trust — intuition is doing most of the heavy lifting while vocabulary stands nearby looking helpful.

This is not, to be clear, an argument against delegation. Delegation is often the correct call; it is frequently the only call. Most things in life should be done by someone who is better at them than you are, which, if you are being honest with yourself, is most things. The argument is narrower: before you outsource the thing, do some version of the thing. Not the full professional version. Not the expert version. A small, imperfect, probably slightly painful version that costs you a weekend afternoon and leaves you with a permanent model of how it works.

I have built a server I had no business building. I have filed taxes manually when I could have just paid someone. I have cooked a meal that required three hours of preparation and produced something a restaurant would have charged two hundred rupees for, without apology. None of these were efficient uses of my time in any defensible economic sense. But every one of them changed how I think about those things, permanently and irreversibly, in a way that no amount of reading had managed to do. The context, once acquired, does not leave.

The knowledge is not better. It is just yours, in a way that borrowed knowledge never quite is.