I’ve been writing all day about taste as the key skill in the AI era. Select what matters, ship it, get feedback, recalibrate. It sounds clean. It sounds like a system. And for a certain class of decisions, it genuinely works.
Publishing a blog post is a small bet. You pick an idea, you put it out there, you see if it resonates. The feedback is fast — days, sometimes hours. The cost of being wrong is nearly zero. Nobody remembers the post that didn’t land. You learn, you adjust, your taste gets sharper. The loop is tight and forgiving.
But then I caught myself applying the same logic to bigger questions. Should I optimise for money or meaning? Is this the right career move? Am I spending my thirties well? And I realised the framework breaks down in a very specific way.
Big decisions have three properties that break the taste-calibration loop. The feedback is slow — years, not days. The decision is mostly irreversible — you can’t A/B test a career. And the feedback, when it finally arrives, is ambiguous. Did this job work out because it was the right choice, or because you made it work through effort that would have made any choice look right? You can’t tell. There’s no control group for a life.
The startup world has a word for the small-bet version: iteration. Ship fast, measure, pivot. It works for products because products are testable. You can run both versions simultaneously. The data tells you which is better. But you can’t run both versions of your career simultaneously. You pick one, and then you spend years making it true.
That “making it true” part is the thing the taste framework misses entirely. For small bets, the selection is what matters — pick the right post, the right problem, the right angle. For big bets, the selection matters less than what you do after. The person who chose Capco and threw themselves into it fully will probably end up in a better place than the person who chose the “objectively better” option and hedged. Not because the choice didn’t matter, but because commitment compounds in a way that selection doesn’t.
This is uncomfortable if you’ve just spent a day celebrating taste as the master skill. It means taste has a domain of applicability, and the most important decisions in your life are largely outside it. For those, you’re operating on something closer to faith — not religious faith, but the willingness to commit to a direction before the evidence is in, and then generate the evidence through action.
I think the honest version of the framework is something like: taste for the daily, commitment for the yearly. Sharpen your taste on the small loops — what to publish, what to work on today, which problems to raise in a meeting. Those calibrate fast and the cost of error is low. But don’t mistake that calibration for wisdom about the big questions. For those, the skill isn’t selecting the right path. It’s making whatever path you selected become right through how you walk it.
The post I wrote earlier today — “trust contact with reality over any number of opinions” — is true. But some realities take a decade to make contact with. In the meantime, you’re just walking.