The clinical eye and the code review

Pharmacy school trains a specific kind of seeing: the clinical eye. You look at a prescription and within seconds the trained part of your mind asks a stack of quiet questions — is this dose right for this weight, does this interact with the other drug, is the frequency plausible, does the prescriber know something I don't? Nobody consciously runs that checklist anymore. It's been drilled until it runs itself.

Years later I caught myself doing the same thing to a pull request, and I realised the clinical eye is exactly what a great code reviewer has.

How the eye is trained

From whole to parts. First the gestalt — does this prescription look right? Then the parts — each drug, each number. Code review works identically: does this diff feel sound, then line by line. Beginners start with parts and never develop the gestalt; training runs both, deliberately.

Pattern library from exposure. The eye knows what wrong looks like because it has seen wrong a thousand times — in tutorials, in error drills, in real charts. Reviewers are the same: they've seen the injection bug, the race condition, the swallowed error so many times the shapes trigger recognition before reasoning.

The index of suspicion. Pharmacists are taught calibrated suspicion — higher for certain drugs, certain combinations, certain patients. Code has the same tiers: auth changes, money paths, migrations get the raised eyebrow every time, regardless of how clean they look.

The compounding part

The clinical eye isn't talent; it's accumulated exposure, structured until it becomes instinct. That's learnable in code too — every bug you've personally fixed becomes a pattern the eye recognises next time. The best reviewers aren't smarter. They've simply seen more wrong things, carefully.