Healthcare isn'tunderfunded.It has forgottenhow to learn.

The United States spent $5.3 trillion on healthcare in 2024—more than the GDP of Japan. For that money, Americans live shorter lives than their peers, clinicians face widespread burnout, and patients repeat their medical histories to every new face in every new room.

We keep misfiling this failure. We call it a funding problem and pour in more money. We call it a staffing problem and ask more of exhausted clinicians. We call it a software problem and buy more software—then ask clinicians to serve one more system.

Four decades of reform have rearranged who pays, how care is delivered, and what gets measured. Almost none of it has touched how the system thinks.

Somewhere in every hospital, right now, a hard problem is being solved for the fortieth time—by someone with no clear way of knowing how it was solved the previous thirty-nine times.

The reasoning that runs healthcare—why this diagnosis, why this protocol, why this exception—lives in the worst possible places: hallway conversations, buried message threads, the memory of a charge nurse who retires in June, a PDF scanned from a fax. The decision survives. The why evaporates.

So guidelines calcify into rituals. Workarounds outlive the problems that justified them. Every merger, every rotation, every shift change erases another piece of institutional memory. The system's best explanations, the reasoning behind them, and the knowledge they carry all fade—often so gradually that the loss goes unnoticed.

A system that cannot remember why it does what it does cannot stop doing the wrong thing.the consequence of forgetting why

Here is the strange part: medicine already has the method. The randomized controlled trial is the most honest epistemological machine humans have ever built—a formal apparatus for trying to prove your own idea wrong, and trusting it only if it survives.

Karl Popper said it plainly: knowledge grows by conjecture and refutation, not by accumulation and authority. A rigorous system asks of every belief: what evidence would show that this is no longer our best explanation?

Medicine applies that discipline rigorously when testing treatments. Healthcare applies it far less consistently to the systems around them—the handoffs, staffing models, discharge rules, and prior-authorization rituals—where assumptions are protected from criticism by hierarchy, liability, and exhaustion. Guidelines and protocols harden into settled authority instead of remaining revisable explanations.

A claim that survives criticism is not proven—nothing ever is. It is not yet refuted, which is the strongest thing an honest system can say. Too much of healthcare runs on claims that have never been given a serious chance to fail.

Popper Labs builds infrastructure for exactly this: reasoning as a first-class object. Not just notes about decisions—the reasoning itself: the claim, the evidence behind it, the criticism it faced, and whether it survived.

In the system Popper is building, “why do we do it this way?” is not an archaeology project. It is a link. Every protocol keeps its provenance. Every exception keeps its argument. When the evidence changes, the system traces which decisions depend on it and surfaces what needs reconsideration—instead of waiting fifteen years for a generation of practice to retire.

We are building this as a connected Popper suite—a thinking environment where hard problems face structured criticism, and a working environment where teams keep their reasoning attached to their decisions. Healthcare has the most to gain: the highest stakes, the deepest expertise, and the least infrastructure for keeping what it learns.

The first product we're launching from this thesis is PopperSIM, a training ground for clinical reasoning. Students work through patient cases by making their judgments, evidence, and assumptions visible. PopperSIM challenges their reasoning, shows them where their explanations hold up or fail, and gives them repeated practice revising their thinking—not just recalling the right answer.

Institutional memory should compound like interest. In healthcare it evaporates like ether.popper labs

This is not an AI doctor. It is not a diagnosis tool, and it is not another electronic health record. Popper does not practice medicine—clinicians do, and the last thing they need is another system pretending otherwise. What we're building is the layer underneath: a place where a health system's thinking can be seen, questioned, and preserved over time—never accepted on authority.

PopperSIM is not a test bank or a boards-prep app. It is not designed to reward memorization or replace clinical instruction. It gives the next generation of clinicians a place to fail, be challenged, practice their skills, and revise their reasoning across simulated cases—pressure-testing how they think, not merely what they remember. It does not diagnose patients, recommend treatment, or make decisions about their care.

If you teach clinical reasoning but see students rewarded for what they remember rather than how they think; if you run a health system and can name a protocol nobody can explain; if you build healthcare software and know the ritual of re-learning; if you are a clinician who has watched good reasoning evaporate at shift change—we want to compare notes.

Healthcare doesn't need another dashboard. It needs a reasoning layer that remembers why—and makes every decision answerable to criticism.

if this is you

Help us build the system
that learns.

Explore PopperSIM→meet Popper Labs →