Practice
Marrow
9 posts
The learning curve nobody warns you about
The 10,000-hour rule is told by people who never counted. The hours are not linear — and the plateau is where judgement is quietly built.
What a fracture clinic is actually optimised for
A fracture clinic is graded on throughput. The patient experiences something else entirely. The gap between the two scorecards is where all the interesting
The cloud is just someone else's computer — and your patient's data is asleep on it
Most clinical tools assume the data should live in someone else's data centre. For a great deal of what clinicians build, that assumption is quietly backwa
What the checklist actually fixes
A reading note on Gawande's Checklist Manifesto. The list is not about memory. It is a device for flattening hierarchy just long enough for someone to find
How long does a new hip last? It depends who is asking
A joint replacement does not have a use-by date. It has a survival curve, and reading one honestly means asking who the patient is. Poke the curve and see.
Two clocks, one surgeon, and the case of the missing morning
Fly east, operate at dawn, and your body insists it is the middle of last night. A live body-clock dial of the best and worst hours to operate.
I replaced an enterprise system with a laptop and a weekend
A single SQLite file did what a six-figure platform could not. Sometimes that is the right call, and sometimes it is a trap. Race the queries and decide.
Being Mortal: what "a good day" actually means
A reading note on Atul Gawande's Being Mortal — autonomy, the limits of medicine, and why a surgeon should ask what a good day actually looks like.
How to find a significant result in pure noise (please do not)
p<0.05 does not mean what most people think. Keep testing the same nothing and a significant result will eventually appear. Watch the trap spring, live.