The archive
Writing on healthcare technology, systems, and incentives — some new, some republished from earlier work with its original date. Roughly one a month.
An admiral who ran the Joint Strike Fighter fly-off taught me one rule about politics: every two years, hand the people who funded you a win they can take credit for. Miss the cadence and a better plan loses to a good one that delivers.
Read →I spent the '90s fixing Mercedes door locks that someone else broke. Every design decides who pays when the next person touches it, and the good ones admit the trade.
Read →Everyone argues open versus closed on quality. Put a volume axis on the cost chart and it becomes a capital decision, and I'd guess most companies are too small to make it.
Read →The hours AI gives back are real. They just get spent on the work people used to put off, which is worth a lot, but it isn’t a cost line, and it shouldn’t be scored like one.
Read →If the person asking for approval gets to choose who approves, you don't have a control. You have a signature, and an audit trail of preferences.
Read →Aviation’s safety rules mostly came after the crashes. That’s how the safety regimes that actually work got built, and a lot of AI policy is trying to skip the step that made them work.
Read →The AI story everybody tells comes from a movie the machines lose. The one that actually happened is about a shared package cache, a missing air gap, and who signs for what the agents did.
Read →Almost every real business knows what its product costs. Healthcare mostly doesn't, and that's on purpose: knowing the cost would expose the spread that everybody in the chain is living on. The first provider to name its real number is going to get flooded with business.
Read →The things that feel effortless, like seeing, walking, and knowing who's in the room, feel easy because they're the most optimized things you have. Evolution was the pretraining, and the machines inherited the wrong five thousand years of it.
Read →I had a model doing the same job every morning and getting the same answer every morning. It took me a while to notice that what I’d really built was an expensive cron job.
Read →Every lever has an inverse, and nobody has beaten the second law. AI is a massive lever, so the bill is out there somewhere. The people who worry me are the ones convinced there isn't one.
Read →Postponing work looks like weakness. Priced correctly, it’s buying an option, and the more expensive mistake is finishing something just because you started it.
Read →A generation lived for decades beside a technology that really could end everything, and built hotlines and treaties instead of panicking or pretending. The AI risk debate talks as if this is the first time.
Read →Every airplane compass is a little wrong, and every airplane carries a card saying by how much. Nobody is going to ship a neutral AI model, so that card is the standard we should be asking for.
Read →Thanks to an office remodel, I spent September 11th in Mission Control, the one room built so somebody can read a signal and know what it means. My wife called me twice that day, and I misread both calls.
Read →I thought I bought a device. What I really bought was a lease on a service, and the expiration date got set the day somebody bigger bought the company.
Read →Frontier labs are racing each other on Millennium Prize problems. I’d rather somebody modeled American healthcare end to end and told us which part is holding the whole thing up.
Read →Quality arguments go in circles because everybody is arguing a verdict. The trend is already sitting in your release history, and it costs about an afternoon to pull.
Read →The same commitment turns into a different thing depending on who was in the room when it got made. And the kind move, it turns out, is giving someone a date they can actually hit.
Read →When a team is drowning in rework, everyone assumes it ships badly. Usually the bill is older than the team that’s holding it.
Read →Every unpriced "can you also" teaches people that change costs nothing, and every role you refuse to fund ends up built on whoever is standing closest. You can keep saying yes. Just stop hiding the bill.
Read →AI watermarks are shipping this month, and everyone is busy combing through the writing for an answer that was never in it. You can't lift authorship off the page like a fingerprint. It's a claim made by the person willing to say, "I wrote this."
Read →We praise Toyota's culture because it lets us pretend the magic was mindset. Toyota engineered consequences, and the culture showed up afterward.
Read →A competent essay used to be evidence of something. Now it's evidence you can prompt a chatbot. The work didn't go away. It moved.
Read →The Space Shuttle had 2,000 switches and stopped flying. The Soyuz had 63 and still flies.
Read →One in five claims gets denied on the first pass. A system that says no to a fifth of everything is not making mistakes. It is doing what it was built to do.
Read →Clinicians spend more time clicking than caring, and everyone treats it as inevitable. It isn't. Good clinical UI is a design choice, not a law of nature.
Read →Care coordination depends on every provider seeing the whole patient record. Document-based exchange can't get us there. API-based exchange on FHIR can.
Read →