Epistemic status: The numbers are sourced and current. The factory frame is mine, and it is a frame, not a proof. Section 05 is the strongest case against this piece, written as well as I can write it.
I’ve spent the last year and change studying first principles thinking the way you’d study for a license. Mostly through one man’s factories. Walter Isaacson’s biography of Elon Musk, the Founders podcast working through it, Charlie Munger on inversion, Brad Jacobs on buying and building. I run a home health and hospice company in Southern California. On paper none of that material is for me.
01A Year on the Factory Floor
Two ideas from that year didn’t just stick. They got installed.
The first is the algorithm. Musk makes every factory run the same five steps, in order. Question every requirement, and every requirement has to come with the name of the person who made it. Delete every part and process you can, and if you aren’t adding ten percent of them back, you didn’t delete enough. Simplify what survives. Accelerate it. Automate it, last. The order is the whole point. Most organizations automate first, which means they build robots to perform work that shouldn’t exist.
The second is the idiot index. Take the cost of a finished part. Divide it by the cost of its raw materials. A rocket valve that costs thirty times its metal has an idiot index of thirty, and the index isn’t measuring the valve. It’s measuring the organization that made it. A high idiot index means the waste isn’t in the materials. It’s in the process, which means it’s in decisions people made and could unmake.
We built CultureAI.com on those two ideas, and my companies run on it now. That part of the story I’ve told elsewhere. This piece is about what happened when I pointed the same two ideas at my own industry.
02The Wrong Question
My whole career, every serious conversation about American healthcare has started from the same question. What does it cost. Cost curves, cost drivers, cost containment. Twenty years of conferences about the cost.
First principles doesn’t start with cost. It starts with the product. A factory is judged by what comes off the end of the line, and only then by what the line consumed. So: what does a health system make?
Strip it all the way down. Every rocket, every company, every hospital is built out of exactly one raw material, and it isn’t metal or money. It’s hours of a healthy person’s attention. Ask anyone who has built anything what the true constraint was. It was never the steel. It was always the people, and specifically the people functional enough to show up and think.
Follow that logic and the finished product of a health system is not a visit, a claim, or a discharge. Those are process. The product is healthy years. Years a person can work, build, raise kids, take care of someone else. Everything between the raw material and that output is manufacturing, and manufacturing is where waste hides.
03The Yield
So look at the line’s output like a manufacturer would.
The United States spent 14,775 dollars per person on healthcare in 2024. That is close to double the average of comparable wealthy countries and roughly five thousand dollars more per person than the next highest spender on earth. If spending bought output, we would be the healthiest population in history.
Life expectancy just reached a record 79 years. That’s real, and I’m glad of it. But a manufacturer doesn’t celebrate units produced. He checks how many pass inspection. The inspection number is healthy life expectancy, the years lived in full function, and on the World Health Organization’s latest comparable data ours is 63.9. Japan and Singapore are above 73. Cuba and Vietnam are ahead of us.
Hold those two numbers next to each other. We live 79 years. We live well for about 64 of them. The gap, 12.5 years, is tied with Australia for the largest in the world. Except Australia spends half what we do to produce it.
That is the yield of the highest-spending health factory ever built: double the input, a decade less functional output per person than the countries we’re supposed to be racing. No plant manager on earth keeps his job with that scorecard. The idiot index of American healthcare isn’t the price of any pill, though the pills are bad too. It’s the ratio of what the system consumes to the healthy years it ships.
04Where the Years Actually Go
I see the 12.5 years up close, because my company works in the part of the system where they’re spent. They are not mostly spent in hospitals. They’re spent in living rooms. A father who can’t manage the stairs anymore. A daughter running his medications between her own shifts, becoming an unpaid care coordinator with no training and no relief. A recliner that has quietly become a bed.
The nurse in that living room is not overhead. She is the line worker at the exact station where the yield is decided.
From inside the factory frame, home-based care stops looking like what the industry calls it, a cost center at the soft end of the system. Every year my teams move a person from the dependent column back toward the functional one, at home, is raw material handed back to the country. And the daughter she trains and relieves gets hours of her own attention back, which is a second unit of product nobody counts.
That’s the reframe that changed how I run the company. We are not managing decline at a discount. We are a yield operation at the margin where America loses more functional years than anywhere else.
05The Honest Complication
Now the strongest argument against everything above, because this piece doesn’t deserve to exist unless it can survive it.
The factory frame implies the health system controls its yield. It controls less than half of it. The research on what actually drives health outcomes puts clinical care at somewhere between ten and twenty percent. The rest is behavior, environment, and social circumstance: what we eat, obesity, opioids, cars, guns, loneliness, poverty. A huge share of America’s missing healthy years is manufactured outside the clinic, upstream of anything a doctor or a nurse touches. Blaming the delivery system for all 12.5 years is like blaming the paint shop for a warped chassis.
Here’s why I think the argument survives. First, the frame doesn’t actually require the clinic to own the whole gap. It requires the country to notice that its healthspan factory, the whole thing, food policy to street design to my industry, is shipping a defective yield, and no one is accountable for the finished product. Every department optimizes its station. Nobody owns the line. Naming the product is step one of the algorithm: question the requirement, and every requirement needs a name attached. Right now, healthy years has no name attached.
Second, the share the delivery system does own, it is fumbling in exactly the way the idiot index predicts. The waste isn’t in the raw materials. American nurses and doctors are as good as any in the world. The waste is in process: documentation that consumes a third of a clinician’s hours, prior authorizations, thirty-day payment cycles for work already delivered, turnover that replaces an entire hospital workforce in five years. Process waste is decisions, and decisions can be unmade. That’s my station on the line, and I don’t get to point upstream while it’s wasteful.
06Running the Algorithm on the Factory
So run the five steps on healthcare, in order, the way you’d run them on any line with a bad yield.
Question every requirement. Ask who owns each rule that consumes clinician hours, by name. An enormous amount of what my industry does traces back to no one. The requirement survives because it has always survived. In my company the standing question in every operations meeting is the same one Musk asks: who asked for this? If the answer is a department instead of a person, the requirement goes on notice.
Delete before you optimize. The instinct in healthcare is to add. Add a form, add a committee, add a compliance layer, add software to manage the forms and committees. First principles says the cheapest process step is the one that doesn’t exist. If you aren’t adding ten percent of what you deleted back, you didn’t cut enough. We’ve deleted reports nobody read, meetings that existed to prepare for other meetings, and approvals that approved nothing. Not one has been missed.
Simplify, then accelerate. Cash flow is a yield problem too. Care delivered but not documented is inventory rotting on the dock. The fastest acceleration in our industry isn’t clinical. It’s the distance between the visit and the signed note.
Automate last. This is where the timing gets interesting, because the automation wave is arriving in healthcare right now, whether the process was cleaned up first or not. I wrote about that gate opening in Three Septembers. The operators who run the first four steps before the tools flood in will automate clean processes. Everyone else is about to automate their waste, at scale, and lock it in.
07The Operator Application
If you run a post-acute company, here is the whole piece in one move. Stop presenting yourself as a cost to be managed and start measuring yourself as a yield operation. Pick the functional outcome you actually manufacture, days at home, falls prevented, hospitalizations avoided, a caregiver’s hours returned, and put it on the same dashboard as revenue. Then run the algorithm on everything between your clinicians and that number, starting with question and delete, not with software.
The payers are ahead of the industry on this, whatever you think of their motives. Value-based contracts are a clumsy first attempt to pay for yield instead of activity. The operators who can walk into that negotiation with their own yield data, instead of a visit count, are negotiating the future. The ones who can’t are a line item in someone else’s spreadsheet.
08The Ledger
Every field note that makes a prediction gets a dated line on the public ledger. Here’s this one.
Ways I’m Wrong
The HALE number is the soft spot. Healthy life expectancy is modeled from self-reported disability data, it lags by years, and cross-country comparisons carry real measurement noise. Cuba’s statistics in particular deserve skepticism. If the 2021 figures get revised and the American gap shrinks materially, the sharpest sentence in this piece dulls with it. The direction of the gap is robust across every dataset I can find. The precise size is not.
The frame could also fail the way frames fail. Healthcare might resist factory logic for reasons that aren’t waste: bodies aren’t chassis, patients aren’t parts, and a yield mindset applied stupidly becomes the worst version of managed care, denying grandmothers as defects. If the metric gets captured, my prediction could resolve true and the country could still lose. And the ten-to-twenty percent problem never goes away. If healthy years move in the next decade, food, drugs, and despair will decide more of it than my industry will. I’m running my station as if the line depends on it anyway, because that’s the only station I’ve got.
No comment box here on purpose. The discussion for every field note lives on LinkedIn, in public, with names attached.
It also continues weekly: one field note in your inbox, text and audio.
Weekly · Unsubscribe anytimeReferences · 10 sources
1. US national health spending, $14,775 per person in 2024. CMS National Health Expenditure data. cms.gov/nhe
2. US health spending versus comparable countries, nearly double the peer average. Peterson-KFF Health System Tracker. healthsystemtracker.org
3. US life expectancy at a record 79 years. CDC National Center for Health Statistics. cdc.gov/nchs
4. Healthy life expectancy (HALE): US 63.9 years, Japan and Singapore above 73, and the 12.5-year lifespan-healthspan gap. WHO Global Health Estimates, 2021 comparable series. who.int/data/gho
5. Clinical care contributing roughly 10 to 20 percent of health outcomes, with behavior, environment, and social factors driving the rest. County Health Rankings model, University of Wisconsin Population Health Institute. countyhealthrankings.org
6. Documentation consuming a large share of clinician hours. Time-motion research in Annals of Internal Medicine and AMA studies on EHR burden. acpjournals.org (Sinsky et al., 2016)
7. Hospitals turning over more than 100 percent of their workforce over five years. NSI National Health Care Retention & RN Staffing Report. nsinursingsolutions.com
8. The five-step algorithm and the idiot index. Walter Isaacson, Elon Musk, Simon & Schuster, 2023.
9. The gate opening for healthcare AI. Three Septembers, The Culture Lab, Nº 13.
10. The yield argument in its founding form. Culture Is Not a Values Poster, Part Five, The Culture Lab, Nº 06.