Clara Bennett: Max, I read something this week that I keep turning over — a physicist defending string theory by basically saying, look, we don't have testable predictions yet, but the mathematics is too beautiful to abandon. And I thought: Karl Popper would have had a field day with that sentence.
Max Rivera: Popper, yeah — the falsifiability guy.
Clara Bennett: Right. So — The Logic of Scientific Discovery, 1934. Popper introduces what he calls the demarcation criterion: the thing that separates science from non-science. A theory is scientific if it makes predictions that could in principle be contradicted by observation. If no possible observation could ever prove it wrong, it isn't science.
Max Rivera: That's the version I learned. Clean, almost satisfying.
Clara Bennett: And string theory right now produces zero testable predictions. Zero. By Popper's own measure, it belongs in the same category as — well, as Freudian psychoanalysis. Which is the example Popper himself used as his canonical pseudoscience.
Max Rivera: Except Freud turns out to be the wrong example too, doesn't it — because there were testable claims?
Clara Bennett: That's the part that doesn't resolve neatly. Historians have found that psychoanalytic theory did generate some testable claims, and researchers actually tested them. So the gate Popper built — it's open for string theory and only partly closed for Freud. And we still teach it as the dividing line.
Max Rivera: So the demarcation problem — the thing falsifiability was supposed to solve — kind of resurfaces inside the solution itself.
Clara Bennett: Which is exactly why Popper's move was so specific — he wasn't just saying 'be testable.' He was solving a logic problem. The idea is that confirming evidence can pile up forever and never actually close the case. You observe one white swan, two, a million — and you still can't prove all swans are white. But one black swan ends it. That asymmetry is the whole engine.
Max Rivera: Wait — so it's not about whether you've tested it. It's whether a test is even possible in principle.
Clara Bennett: Exactly that. A theory is scientific if you can describe an observation that would prove it wrong — not whether anyone's actually run that experiment yet. Just that the door exists.
Max Rivera: Okay, so — like, a doctor says: if this drug works, your fever breaks in 48 hours. If it doesn't break, something's wrong. That's it, right? That's falsifiability. There's a specific outcome that would kill the claim.
Clara Bennett: That's a clean version of it, yes. And notice — the doctor isn't waiting to be proven right. They're naming what failure looks like upfront. Popper actually wrote, word for word: 'the criterion of the scientific status of a theory is its falsifiability, or refutability, or testability.' Three words for the same gate.
Max Rivera: That's from 'Science as Falsification,' yeah — I mean, that quote is almost too clean. It sounds like a definition someone polished for a textbook.
Clara Bennett: In practice, though, the hard part isn't the definition — it's what he called potential falsifiers. A theory has to actively name the observations it prohibits. Not just 'we could test this someday.' It has to exclude something specific. If every possible observation is compatible with your theory, the theory isn't saying anything.
Max Rivera: So — wait, that's the move that catches Freud, right? Any patient outcome gets explained. Recovery proves the therapy worked. No recovery proves the patient resisted. Nothing is ever prohibited.
Clara Bennett: That's the charge. And it's why General Relativity was Popper's good example — Einstein's theory predicted light would bend around the sun by a specific angle. If observations had come back different, relativity was dead. That's a genuine potential falsifier. It's not just risky in spirit; it's risky in a measurable number.
Max Rivera: A measurable number — and that's the part I keep getting stuck on. Like, Einstein didn't just say light bends. He said by how much. That's what makes it risky, right? A vague prediction can survive almost anything.
Clara Bennett: That's exactly the distinction. Precision is what gives a falsifier teeth. And the 1919 solar eclipse observation actually tested it — if the angle came back wrong, General Relativity was done. That's a theory putting itself on the line.
Max Rivera: Now compare that to Freud. Whatever happens to the patient — gets better, stays the same, gets worse — there's an explanation waiting. The theory never loses.
Clara Bennett: And that's not a small problem. If a theory permits every possible outcome, it has no potential falsifiers. It's not making a claim about the world — it's just redescribing whatever the world does.
Max Rivera: Marxism is the same move, yeah? Failed prediction shows up, and suddenly there's a — wait, actually, what were the rescues? Like specifically?
Clara Bennett: The working class in Western Europe didn't revolt on schedule, so the explanation became: false consciousness, or late-stage capitalism, or the revolution was delayed by imperialism. Each rescue absorbs the failure without generating a new prediction. Popper called those ad hoc explanations — modifications that protect the theory instead of extending it.
Max Rivera: So the problem isn't the original theory being wrong — it's the move afterward. The patch.
Clara Bennett: Right — and this is the part worth holding onto. Falsifiability doesn't mean therapy doesn't work, or that Freud had no useful insights. The distinction is between scientifically structured and genuinely helpful. Those are separate questions.
Max Rivera: Though I'll say — what happens when scientists make that same ad hoc move for a theory we actually trust? Because that's uncomfortable, and honestly I think string theory and a few others are going to break the whole framework open.
Clara Bennett: That's actually where the framework starts to crack. Working scientists protect core theories from single refutations all the time — and the community accepts it as legitimate. Not as cheating. When Neptune's orbit didn't match Newtonian predictions, astronomers didn't abandon Newton. They postulated an undiscovered planet. That's an auxiliary hypothesis, and it worked. The criterion, in practice, is never applied to a theory in isolation.
Max Rivera: Wait — so the same move that Popper called dishonest in Freud and Marxism, physicists did with Newton and it was fine?
Clara Bennett: The difference Popper drew was whether the auxiliary hypothesis generates a new testable prediction. Predicting Neptune's position — that's a real bet. The Marxist rescue of 'false consciousness' generates nothing new you can check.
Max Rivera: Okay, but string theory doesn't even get to the auxiliary hypothesis stage — there's no original testable prediction to protect. Zero. It's not that an experiment failed and physicists patched it. It's that the experiment never existed.
Clara Bennett: And by Popper's binary, that puts string theory in the same column as Freudian psychoanalysis. Which is — I mean, that's the sentence that should feel uncomfortable.
Max Rivera: Especially because — wait, here's the Freud thing that keeps nagging me — Popper used psychoanalysis as his clean case of pseudoscience, right, the flagship example. But historians have actually documented that psychoanalytic theory generated some testable claims that researchers tested. So his clearest example is already fuzzy.
Clara Bennett: Which is why psychology is the more honest test case. Behaviorism makes tight, measurable predictions — falsifiable by Popper's measure. Classical psychoanalytic theory doesn't. The criterion ends up sorting *within* psychology, not just between psychology and physics.
Max Rivera: So falsifiability is a spectrum inside a single discipline — it's not this wall between science and non-science, it's a gradient that runs through every field.
Clara Bennett: The binary he promised doesn't survive contact with actual science. What it does do — and this is worth keeping — is tell you something about how a claim is structured. Whether it exposes itself to refutation at all. That's real. It just isn't the clean gate.
Max Rivera: But that's — I mean, that's the thing I can't settle. Popper introduces verisimilitude, right — this idea that Einstein advancing beyond Newton isn't just about falsifiability, it's about being closer to the truth. But that requires judgment. Someone has to look at two theories and say this one is better. And I don't know how you make that call in AI alignment or quantum gravity, where the experiments might be thirty years out.
Clara Bennett: No, that's genuinely open. Falsifiability tells you something real — it tells you about the logical structure of a claim, whether it even exposes itself to being wrong. But it doesn't tell you whether the claim is true, or whether the field is worth funding now. Those are different questions, and Popper never fully answered the second one.
Max Rivera: Which is what I keep saying. It's a direction, not a gate.
Clara Bennett: A direction toward bolder, more specific, more refutable claims — yes. That's probably the most defensible reading of what Popper actually left us. Not a mechanism. A north star.
Max Rivera: I'll take an unresolved north star over a broken gate. Thanks for thinking through this with me.
