We Locked Up Millions, Lobotomized Thousands. Psychiatry's Dark Secret Is Still Here.
In this article
Bottom line: Between roughly 1936 and 1967, about 40,000 Americans were lobotomized, and the physician who invented the procedure won a Nobel Prize for it in 1949.
The pattern behind it wasn't villainy. It was a cheap, scalable fix for an overwhelmed system, adopted before anyone measured the harm.
That pattern hasn't gone away: today it shows up in jail-based mental health care, in loosely validated diagnostic categories, and in AI therapy chatbots being deployed faster than they're being tested.
I used to think the lobotomy story was a horror anecdote, the kind of thing you shake your head at and move on from.
Then I spent a weekend reading about how it actually happened, and it unsettled me for a different reason. Nobody in that story thought they were the villain.
The doctors were exhausted. The hospitals were overflowing. A procedure showed up that was fast, cheap, and endorsed by prestigious people, and the system said yes because saying no meant doing nothing.
I build products for a living. I know exactly what it feels like to ship the thing that works well enough while the evidence is still arriving. That's why this story stays with me.
The Numbers That Should Bother You
At its peak in 1955, US state mental hospitals held around 559,000 people.
Over the century, millions cycled through those institutions, many committed without a meaningful hearing, some for conditions we'd now treat with an outpatient appointment or not treat at all.
Lobotomy was the efficient answer to that crowding. Egas Moniz won the Nobel Prize in 1949 for developing the leucotomy.
Walter Freeman then turned it into a production line, adapting it into the transorbital procedure: an ice-pick-style instrument through the eye socket, no operating room required, over in minutes.
Freeman personally performed thousands of these, and he drove a van he called the "lobotomobile" to demonstrate the technique across the country.
His patients included Rosemary Kennedy, 23, in 1941, who was left permanently incapacitated.
They also included Howard Dully, who was 12 years old in 1960 when his stepmother arranged it because he was, in her words, difficult.
Everyone Is Telling the Wrong Version
The popular retelling goes like this: a few reckless doctors did something barbaric, science corrected itself, and the field matured. It's comforting, and I think it's mostly wrong.
Science didn't correct itself in time.
Chlorpromazine (Thorazine) arrived in 1954, offered a pharmacological alternative, and lobotomy declined partly because something cheaper and more convenient replaced it.
Ethical reckoning came later, and slowly. Freeman did his last operation in 1967, and he was stopped by a patient's death, not by a consensus of his peers.
That distinction matters.
If the harm ended because a better convenience arrived, then the field didn't learn "don't do this." It learned "use the newer tool." Those are very different lessons, and only one of them protects you the next time.
The other comfortable myth is that this was a fringe. It wasn't. It was mainstream, celebrated, and published in respectable journals.
The Nobel committee has never rescinded the prize, and there is still a petition to do so that goes nowhere.
The Confident Cure Cycle
I've started thinking of this as a four-stage loop. I call it The Confident Cure Cycle, and once you see it you'll spot it everywhere, including in my own industry.
1. Capacity Crisis
A system is overwhelmed. Too many patients, too few clinicians, too little time. Pressure, not malice, is what creates the opening for a drastic intervention.
2. The Cheap Fix
Something arrives that is fast, scalable, and priced for volume. It doesn't need to be proven. It needs to be plausible and available. Lobotomy took minutes, and Thorazine meant a pill.
3. Measure the Convenience
Early success gets measured in what's easy to count: patients discharged, beds freed, wards quieter. A quiet ward looked like a success.
It took years before anyone asked what the quiet cost the person inside it.
4. The Late Audit
Years later, long after the infrastructure and reputations have been built, somebody audits the actual outcomes. By then there is a generation of people who can't be un-treated.
Here's the quotable version: a system optimized for managing symptoms will always mistake silence for recovery.
Where the Cycle Is Running Right Now
I'm not claiming we're lobotomizing people today. That would be sloppy, and it would insult people who've been helped enormously by modern psychiatry.
ECT, for instance, is still used and is effective for severe, treatment-resistant depression, and it looks nothing like its cinematic reputation.
But the cycle is running in at least three places, and I think each is worth sitting with.
The jail system as the new asylum. After deinstitutionalization, the beds disappeared and the people didn't.
Chicago's Cook County Jail is regularly described as one of the largest de facto mental health providers in the country.
We stopped locking people up in hospitals and started locking them up in cells, which is the same capacity crisis with a different budget line.
Diagnostic inflation. The first DSM in 1952 listed 106 categories. The current DSM-5 lists more than 300.
Some of that is genuine progress in recognizing real suffering, and some of it is category creep, since a diagnosis is what unlocks reimbursement.
And let's remember that homosexuality was a listed disorder until 1973. That one was removed after a vote and years of activism, not a lab result.
The chemical imbalance story. For decades, patients were told depression was caused by low serotonin.
A major 2022 umbrella review led by Joanna Moncrieff found no consistent evidence supporting that claim.
Antidepressants help many people, but the simple explanation that justified prescribing them was a marketing-friendly model, not a settled mechanism.
The Part That Hits My Industry
This is where I'll get uncomfortable, because I'm a tech founder and my peers are about to run the same cycle at a scale Walter Freeman couldn't have imagined.
Look at the setup. There's a genuine capacity crisis, with millions of people unable to get a therapist, waiting months, or priced out entirely.
A cheap, scalable, always-on fix arrives in the form of AI companions and therapy chatbots.
And the early measurement is engagement, retention, and session counts, which are the digital equivalent of a quiet ward.
I'm not saying these tools can't help. I use language models daily and I've seen them be kind and useful at 2 a.m. when no human was available.
But a model that is endlessly agreeable can feel like care while functioning as sedation.
I wrote something adjacent to this in You're Absolutely Right, about how sycophancy is a design property and not a bug. In therapy that property is dangerous.
A good therapist's job is often to not tell you you're right.
The audit stage of the cycle is where we are. Controlled, long-term outcome studies for these tools are thin, and the deployment is already global.
If history is any guide, we're in stage three right now, counting the convenience.
What I'd Actually Do About It
Skepticism without a plan is just cynicism, so here's the practical version, depending on where you sit.
- If you build or buy mental health tech: demand outcome data, not engagement data. Ask what happens to users at 6 and 18 months, and who audits it. If the vendor can't answer, that's your answer.
- If you manage a team or a benefits plan: treat "available at scale" as a feature and "validated" as a separate question. They are not the same thing.
- If you're a patient or a parent: ask any provider, human or software, what the evidence is, what the alternatives are, and what stopping looks like. A good clinician welcomes that. A bad one gets defensive.
- If you're an investor or a founder: build the late audit into the product on day one. Instrument for harm, not just usage.
None of this requires believing psychiatry is evil. It requires believing that good intentions plus scale plus weak measurement is a reliable recipe for harm, because that's what the record shows.
The Bigger Picture
What haunts me about Howard Dully's story isn't the procedure. It's that the people around him believed they were helping, and the institution handed them a tool and a reassurance.
Everyone behaved reasonably inside a system that had stopped asking whether it was working.
We love to imagine that we'd have been the one to say no.
The honest answer is that most of us would have signed the form, because the doctor was credentialed, the ward was full, and the alternative was nothing. Wisdom isn't refusing the cure.
It's insisting on the audit before the infrastructure gets built.
The lobotomy era ended, but the machinery that produced it, a crisis, a cheap fix, a convenient metric, and a late reckoning, is still humming.
The only question is which tool is on the cart this decade.
So here's what I keep asking myself, and I'd like to hear where you land: what's the "quiet ward" in your own field right now, the metric that looks like success while nobody has checked what it costs the person on the other end?
