Every healthcare organization says it wants people to speak up. Almost none of them can name the people who actually do.

That is the real test — and it is not whether an institution honors its critics once they have been proven right. Institutions are glad to do that. It costs nothing. The test is whether it protects them while they still sound wrong, while they are still inconvenient, while they are still making everyone in the room uncomfortable.

Twelve years ago, Brian Powers, Christine Cassel, and I published a short essay in the Journal of General Internal Medicine called "The Power of Embedded Critics." The argument was that the people best positioned to change an institution are the ones who belong to it. We wrote that such reformers "derive unique insight and power from their position," and that within the profession sat a reservoir of embedded critics waiting to be tapped.

I still believe every word. But I got one thing wrong. I treated embedded criticism as an underused asset — as if the barrier were awareness. The barrier is not awareness. The barrier is cost. And in the years since, I have watched how few people in this industry are willing to pay it, and how efficiently we punish the ones who are.

Walzer's connected critic

We borrowed the frame from Michael Walzer, who spent much of his career arguing that the most effective social critic is not the exile or the outsider but the person who stands, in his phrase, only a little to the side of his own society — a distance measured in inches. The disconnected critic is easy to dismiss; he shares neither your vocabulary nor your stakes. The connected critic is not, because he argues in the moral language the institution has already claimed for itself. He is not saying your values are wrong . He is saying you are not living them, and I can prove it, because I am here.

Walzer's paradigm case was the Hebrew prophet — Amos, Isaiah — and the point is easily missed. Isaiah was no wild man shouting from the desert; he had access to kings, and the indictment in his first chapter is aimed not at some foreign enemy but at the rituals of his own temple, the ones he himself belonged to.

Peacetime leadership and wartime leadership

Most of what we teach as leadership is peacetime leadership: stewardship, continuity, coalition, brand, the careful management of reputation and risk. In peacetime, defending the institution is serving the institution. Loyalty and advocacy point in the same direction.

Wartime begins when an institution's routines become more dangerous than the external threats they were built to manage.

Peacetime leaders optimize. Wartime leaders repent.

In peacetime, loyalty means protecting the institution. In wartime, loyalty means protecting the institution's purpose — including from the institution itself.

The difficulty is that nobody announces the transition. There is no declaration and no bugle. The org chart looks identical on the first day of the war and the last day of peace, which is why so many capable people keep running a peacetime playbook against a wartime problem: another listening tour, another rebrand, another set of values on the wall.

And in wartime, the necessary work can only be done by people inside the establishment turning against the establishment. Outsiders can indict, but they cannot convict. They lack standing, access, and specific knowledge of where the bodies are actually buried. The insider has all three — which is precisely why the insider's silence is the expensive kind.

American healthcare entered wartime gradually, which is why so few of us noticed. We normalized administrative burden that everyone doing the work knew was waste. We accepted prices almost no one inside the system can explain. We learned to call preventable delay scheduling , documentation with no clinical purpose compliance , and incentives drifting away from patients market reality . Somewhere in there, the enemy stopped being outside the walls.

We keep sending peacetime leaders anyway. They issue statements. They commission task forces. They pass the trade association's talking points through their own mouths and call it engagement.

The lineage is right in front of us

Nearly everything good in modern American healthcare came from embedded critics, and we have somehow filed them under "pioneers" instead. The list is long — Don Berwick alone has spent four decades telling a profession he loves what it does not want to hear — but four cases make the pattern plain enough.

Avedis Donabedian is remembered as the father of quality measurement, as though he were a methodologist. He was a scold, and a loving one. He insisted the foundations of quality were moral rather than technical — the secret of quality, as he put it, is love — and he reserved particular contempt for physicians who took refuge in the claim that they were fine clinicians trapped in a bad system. The system, he said, is the responsibility of physicians and hospital leadership.

Howard Hiatt used the authority of a Harvard deanship to ask a question the profession did not want answered: how often do hospitals injure the patients they admit? The Harvard Medical Practice Study came back with the number, and it became the empirical spine of the patient safety movement and, a decade later, of To Err Is Human . Hiatt did not go looking for someone else's failure. He commissioned an audit of his own house.

Barbara Starfield spent a career at Johns Hopkins arguing that American medicine had built itself wrong — that a system organized around specialists and procedures would underperform one organized around primary care, no matter what we spent. When she published a JAMA commentary tallying the deaths caused by medical care itself, she later recounted that the first journal she sent it to had rejected it on the grounds that readers would not find it interesting.

Troyen Brennan offers the sharpest case. As chief medical officer of CVS Caremark — the largest pharmacy chain in the country — he co-authored a 2014 JAMA viewpoint arguing that pharmacies should stop selling tobacco. He was not describing an industry he observed. He was describing his employer’s shelves. Partnering closely with CVS CEO Larry Merlo, the company exited tobacco that year at an estimated cost of roughly $2 billion in annual revenue.

The power of self-indictment

Criticism from outside is free. Criticism from inside costs something, and the cost is precisely what makes it credible.

When you name a practice you benefit from, you strip your listener of the easiest way to dismiss you. Interest no longer explains your position. This is why self-indictment carries a force no amount of analysis can manufacture — and why it is so rare. It is not merely uncomfortable to say my industry does this, and I have done this . It is professionally expensive. Your own communications team will ask you, gently, whether this is the right time. It is never the right time. That is a feature of the mechanism, not a scheduling problem.

Richard Baron did a version of this from the chair. In February 2015, with internists in open revolt over Maintenance of Certification, the president of the American Board of Internal Medicine wrote to every diplomate that the board had gotten it wrong, apologized without qualification, and suspended the requirements that had provoked the rebellion. He did not defend the program's intent, explain that critics had misread the evidence, or announce a listening tour. He conceded the substance in public, on behalf of the institution he ran. Pressure explains the timing but not the response: most leaders facing an identical revolt issue a defense, which is why a decade later people still cite the apology.

I run a Medicare Advantage plan. I have written critically about practices in my own industry — coding intensity, prior authorization, marketing incentives that have drifted from the program's purpose. I do not offer that as evidence of virtue. It is the minimum entry fee for being taken seriously on any of it.

Why institutions suppress the people they most need

The reason is not that leaders dislike criticism. Most of us like criticism in the abstract and solicit it constantly. The embedded critic is different because he removes the institution's favorite defense: the presumption of complexity. Our first line of protection is always that the critic does not understand the constraints, has not seen the real numbers, was not in the room, has never had to make payroll. That works beautifully against outsiders, and outsiders are most of who criticizes us.

It does not work against a member. He knows where the data came from because he helped produce it. He knows how the incentives work because he is paid by them. He was in the room. Ignorance is off the table and so is naivete, which leaves only motive — and so motive is what gets attacked. He is bitter. She is building a personal brand. He is positioning for a bigger job. Institutions rarely fail for want of good intentions. They fail because everyone who knows enough to name the problem also has a reason not to.

Two people I don't always agree with

The lineage above is safe to praise because it is largely historical. The live version is harder.

Anthony DiGiorgio is a neurosurgeon at a public safety-net hospital in San Francisco and one of the most persistent critics of the 340B drug pricing program — a revenue stream that runs through institutions like his own. Anish Koka is a practicing cardiologist in Philadelphia who writes, week after week, about the conformity of his profession and about what his colleagues believe privately and will not say publicly.

I disagree with both of them on real questions, sometimes sharply. That is not a disclaimer; it is the point. The test of an embedded critic is not whether he is right about everything. The test is whether he is describing something abnormal that the rest of us have quietly agreed to treat as normal — and whether he is paying a price to describe it. Prior authorization queues discussed like weather. Nonprofit status treated as an identity rather than an obligation. Rooms where margin gets three hours and patients get seven minutes.

We do not need agreement from our critics. We need accuracy and nerve.

Doctors have leverage they are not spending

Which brings me to the group with the least excuse. Physicians are among the most protected workers in the American economy. There is a national shortage; a physician fired for speaking uncomfortably will have another offer within weeks. And yet the profession is largely quiet. The binding constraint is not the employment agreement or the non-disparagement clause. It is cultural. We select for people who are good at being graded, train them inside rigid hierarchies, reward deference for a decade, and then call the resulting silence collegiality.

There is also a hole in the curriculum. We have spent decades teaching physicians to advocate for patients — against the insurer, against the utilization reviewer, against the disease. We spend almost no time teaching them to advocate against their own organizations when patients require it. The training runs out at exactly the point where the patient's interest and the employer's interest come apart.

The exceptions show what the leverage is worth. Helen Ouyang still works emergency shifts at Columbia while publishing long-form journalism about the failures of the system that employs her; she did not have to leave medicine to report on it, which is exactly why the reporting lands. Danielle Ofri has practiced at Bellevue for decades while writing that American health care's business model runs on the willingness of clinicians to absorb unbounded unpaid work. Both are describing the operating assumptions of their own employers, under their own names, between shifts.

If you hold leverage you are choosing not to spend, you do not have a hardship. You have a choice.

We honor our critics too late

Notice what those four names have in common now. Donabedian has awards named for him. Hiatt's study is a citation classic. Starfield's argument about primary care is closer to conventional wisdom every year. Brennan's tobacco decision is taught as a business-school case. We tell them as origin myths.

We did not treat them that way at the time. At the time, the connected critic is not a team player. He has an agenda. She is self-promoting, or bitter, or off-message, or naive about margin. The vocabulary varies by institution; the function never does. It recategorizes the person so that no one has to engage the content.

And the marginalization is rarely dramatic. Almost no one gets fired. They simply stop being invited. Their name comes off the working group. Their review acquires a line about collaboration. The org chart quietly routes around them, and three years later they are somewhere else and everyone agrees it was a fit issue.

Posthumous honor is the cheapest kind. A named lectureship costs nothing. A promotion costs something. If the only critics your institution honors are the dead ones and the ones who left, you have retrospective admiration paired with present-tense suppression — and everyone junior to you can tell the difference.

Every institution says it wants truth. The real question is whether it wants truth that carries consequences.

The change is cultural, not procedural. There is plenty of criticism in healthcare already — most of it anonymous, most of it aimed outward, none of it costing the speaker anything. What we lack is the habit of honoring connected critics in real time, while the criticism still stings and while it might still turn out to be wrong.

It starts with a question any leader can answer in about ten seconds. Who in my organization consistently makes me uncomfortable because they are usually right?

You know the name. Most leaders do. The only question is what you have done with it.

The practice from there is not complicated. Promote them. Protect them. Give them visibility. And — this is the one that gets skipped — do not require them to become agreeable before they advance. That is how institutions launder their critics: we rarely fire them, we sand them down, and then we promote the sanded version and congratulate ourselves for rewarding dissent. Seat the person who named the problem on the committee that owns it. Make sure their evaluation does not run through the person they criticized. Say their name on the record in the quarter the critique landed — not fifteen years later at a retirement dinner, when it is free.

And model it. The fastest way to make self-indictment survivable for a resident or a frontline nurse is for the person at the top to do it first, out loud, about something that costs him.

None of the names above left and threw stones. They stayed and did the harder thing.

Institutions do not become courageous by hiring courageous people. They become courageous when courageous people can survive inside them.