When someone behaves in a way that makes no sense to you, you don't get hurt or offended the way other people might. You do something colder and faster: you classify them. Within seconds you've slotted the person into a category — irrational, careless, low-competence, not worth modelling further — and moved on. It doesn't feel like judgment to you. It feels like an efficient read. You took the data, ran it, and reached a verdict.
But that verdict is a closed model built on almost no information, and closing a model early is the one error your entire mind is organised to avoid everywhere else. You'd never sign off on a strategy after seeing one variable. Yet with people, you do it constantly — you watch a single illogical move, conclude you've understood the whole person, and file the case. The judgment gives you a clean, finished picture. The problem is that the picture is usually wrong, and being confidently wrong about people is precisely how a strategist gets blindsided.
Judgment is a model that stopped updating
For most types, snap judgment is about feeling superior. For you it's subtler and more dangerous, because it's disguised as analysis. When you decide someone is an idiot, you're not being emotional — you're declaring the model complete. There's nothing left to learn here; the case is closed. That sense of closure is what makes it so seductive. An open question costs you attention; a verdict frees you to stop thinking about a person you've decided isn't worth the compute.
But notice what you've actually done. You've taken a system you barely understand — another human being, with motives and constraints you can't see — and frozen your model of it after the first data point. You'd recognise that instantly as bad reasoning in any other domain. A behaviour you can't explain isn't proof the person is stupid. It's proof your model of them is incomplete, which is information that the case is not closed at all.
Curiosity is intelligence-gathering, not warmth
You can let go of the idea that curiosity is a soft, relational virtue — being nicer, more generous, more forgiving. That framing will never move you, and it shouldn't. Reframe it as what it actually is: superior data collection. The judgmental read gives you a low-resolution model built on one observation. The curious read — 'what would have to be true for this behaviour to make sense to them?' — keeps the model open long enough to gather the variables you're currently missing.
And the missing variables are almost always there. The colleague who seems careless is protecting a constraint you can't see. The person acting irrationally is optimising for a goal you haven't identified. Their behaviour is rational inside their information set; you just don't have their information set yet. Curiosity is the only instrument that retrieves it. A snap verdict guarantees you never will, which means you keep operating on a model you already know is incomplete.
A wrong read on people makes you easy to fool
Here's the part that should actually bother you. The whole value of being a strategist is prediction — modelling how things and people will behave, then positioning ahead of it. A judgment that closes early doesn't just make you a little colder to be around. It corrupts the input to every prediction you'll make about that person. You're now forecasting from a model you built lazily and never updated, and you don't even know it's wrong because the verdict felt so clean.
That is exactly how sharp people get played. The person you filed as 'incompetent' runs circles around you because you stopped watching them. The motive you dismissed as 'irrational' was the real driver of their next move, and you didn't see it coming. Cheap superiority is expensive: it trades the satisfaction of a quick verdict for a permanent blind spot. Curiosity costs you the comfort of closure now and buys you a model that can't be exploited later.
Hold the file open for one more question
The fix is small and fits how your mind works, because it's a protocol, not a personality transplant. The instant you catch yourself reaching the verdict — 'this person is an idiot,' 'that was careless,' 'they're being irrational' — treat it as a flag, not a conclusion. The speed of the judgment is the tell that you're working from too little data, not that the case is obvious. Then ask one question before you file: what would make this behaviour the sensible move from where they're standing?
You don't have to like the answer or agree with it. You're not granting approval; you're refusing to lock in a low-information model. Sometimes the answer confirms your read — and now it's confirmed instead of assumed, which is stronger. More often it surfaces a constraint, a goal, or a piece of context that rewrites the whole picture. Either way you end with a better model than the snap verdict gave you, and a better model is the only thing you were ever actually after.
You don't judge people out of feeling — you do it out of a craving for a finished model, and closing the file early is the one error your mind exists to prevent. Treat the snap verdict as a flag that you're short on data, not proof you've understood. Curiosity isn't you going soft; it's you refusing to forecast a person from a model you already know is incomplete — which is exactly what makes you harder to fool and, almost by accident, far easier to be around.