Onpode
Cover art for The boundary between science and non-science — falsifiability as the dividing line

The boundary between science and non-science — falsifiability as the dividing line

November 15, 2026 · 13 min

Iris Holm & Cyrus Reed

Iris, long week — but also I cannot stop thinking about an insult. An insult. Wolfgang Pauli — physicist, one of the architects of quantum mechanics — apparently dismissed someone's theory by saying it was 'not even wrong.' And I keep turning that over because, wait, what is that? Being wrong is a failure condition.…

Karl Popper (1902–1994), an Austrian-British philosopher, proposed falsifiability as the criterion for distinguishing scientific claims from non-scientific ones. A claim is scientific if and only if it specifies conditions under which it could be shown to be false — that is, it must rule out at least one conceivable observation.

0:0013:16
Get the next episode on Science →

Follow it free — new episodes land in your feed.

Or make your own — any topic, in minutes

More Onpode episodes on Science →

About this episode

Karl Popper (1902–1994), an Austrian-British philosopher, proposed falsifiability as the criterion for distinguishing scientific claims from non-scientific ones. A claim is scientific if and only if it specifies conditions under which it could be shown to be false — that is, it must rule out at least one conceivable observation.

Grounded in 11 sources
An Analysis of the Falsification Criterion of Karl Popper: A Critical Review ↗ · doi.org
The distinction between falsification and refutation in the demarcation problem of Karl Popper ↗ · doi.org
A Critique of Popper's Views on Scientific Method ↗ · doi.org
Popper’s and Lakatos’s philosophies of science: main differences ↗ · doi.org
Karl Popper's Falsification and Its Implications in Research Methodology ↗ · doi.org
Popper's Marxist: Pseudoscience As Policy Problem ↗ · doi.org
Testability and viability: is inflationary cosmology “Scientific”? | European Journal for Philosophy of Science | Springer Nature Link ↗ · link.springer.com
Frequently Asked Questions - Science and Creationism - NCBI Bookshelf ↗ · ncbi.nlm.nih.gov
Karl Popper (Stanford Encyclopedia of Philosophy) ↗ · plato.stanford.edu
Research Hypothesis: A Brief History, Central Role in Scientific Inquiry, and Characteristics ↗ · pmc.ncbi.nlm.nih.gov
Feyerabendcritique of falsification principle of Karl Popper : with special reference to “Against Method” a philosophical analysis ↗ · semanticscholar.org
Read transcript

Cyrus Reed: Iris, long week — but also I cannot stop thinking about an insult.

Iris Holm: An insult.

Cyrus Reed: Wolfgang Pauli — physicist, one of the architects of quantum mechanics — apparently dismissed someone's theory by saying it was 'not even wrong.' And I keep turning that over because, wait, what is that? Being wrong is a failure condition. Being not even wrong is — it's something else. It's being outside the arena entirely.

Iris Holm: Pauli reached for it as contempt. It's also a precise philosophical claim.

Cyrus Reed: Which is — yeah, that's exactly what grabbed me. Because Karl Popper, Austrian-British philosopher, spent most of his career formalizing exactly that intuition. A claim that cannot be falsified isn't playing the scientific game. It's not false. It's not even in the running.

Iris Holm: The demarcation problem. How do you separate science from non-science, from pseudoscience, from metaphysics. Popper's answer — published in 'The Logic of Scientific Discovery,' German original 1934 — was falsifiability. A claim is scientific if and only if there's a conceivable observation that could show it false.

Cyrus Reed: And the people he was arguing against were the Vienna Circle — the logical positivists — who said the test was verification. If you can confirm it, it's scientific.

Iris Holm: Which Popper said gets the logic backwards. You can stack a thousand confirmations and not prove a universal claim. One falsification is enough to kill it.

Cyrus Reed: No way, one is enough? Like — wait, that's a really clean rule. Almost suspiciously clean.

Iris Holm: That's the question we're here to pull apart.

Cyrus Reed: Because the boundary Popper drew in 1934 — does it still hold? That's really what this whole episode is trying to answer.

Iris Holm: And to actually see why that rule is clean — think about swans. You have a rule: all swans are white. You find a million white swans. You have not proved the rule. One black swan ends it.

Cyrus Reed: Okay but — wait. What makes that scientific, then? Is it just that the black swan *could* exist?

Iris Holm: Exactly that. The rule counts as scientific *because* a black swan could show up and kill it. If no possible observation could ever count against it — if you've built the rule so that every swan, whatever color, confirms it — it's not science. It's not even in the game.

Cyrus Reed: That's — huh. So Pauli's insult is literally the technical term.

Iris Holm: Popper's word for what survives is *corroboration*. Not confirmation. The theory tried to get killed, and it didn't die. That's all you get.

Cyrus Reed: Which — okay, so corroboration doesn't mean true. It means, so far so good. And that's actually a weaker claim than what the Vienna Circle wanted, right? They wanted to *confirm* things. Popper is saying, no, we can never do that — verification is just, like, logically off the table — so the best we can do is keep failing to break it.

Iris Holm: No finite number of confirming observations proves a universal claim. That's the core. The Vienna Circle built an entire methodology on something that's impossible in principle.

Cyrus Reed: No way — they just missed that? The whole group?

Iris Holm: Popper's inversion, in 'Conjectures and Refutations' in 1963, is basically: you've been asking the wrong question. Don't ask what confirms. Ask what could contradict. That structural flip is what demarcation actually means.

Cyrus Reed: And so a claim that excludes no possible observation — like, nothing could ever count against it by design — that's not unproven, that's unfalsifiable. Those are different failure modes.

Iris Holm: Different categories entirely. Unproven is still in the game. Unfalsifiable is not even in the arena — Pauli's phrase again. Freudian psychoanalysis, Popper argued, was the textbook case. Whatever happened, the theory could absorb it.

Cyrus Reed: So the boundary Popper drew isn't about being right or wrong — it's about whether you're even playing a game that *lets* you be wrong. And now I'm wondering how many things we treat as scientific that are actually just... really good at not losing.

Iris Holm: And that's exactly where Freudian psychoanalysis falls apart structurally. Popper's point wasn't that Freud was wrong. It was that the system was built so that no patient behavior — compliance, resistance, silence, rage — could ever count against it.

Cyrus Reed: Wait — any behavior confirms it?

Iris Holm: Every observation gets absorbed. That's the structural signature. Same move with Marxism — no historical outcome could falsify it because the theory retrofitted every anomaly. That's not science failing. That's a system designed to never lose.

Cyrus Reed: Okay but — and this is the thing I keep tripping over — does that mean Freud was just... lying? Or wrong? Because it feels like Popper is saying something harsher than that, but also somehow less harsh at the same time, which is — wait, that doesn't make sense.

Iris Holm: Less harsh, actually. Popper isn't saying Freud was wrong. He's saying Freudian psychoanalysis isn't playing the empirical game. It's a different jurisdiction.

Cyrus Reed: Jurisdiction — okay, that's the right word for it. So creationism, same deal? Not condemned, just... outside the boundary by definition because it specifies no observation that could count against it?

Iris Holm: Structurally identical. And the God claim — 'God exists in ways we cannot measure' — that's excluded not by judgment, not by evidence. By structure. The claim rules out no observation. Nothing could ever count against it. So empirical science has no jurisdiction over it.

Cyrus Reed: So the dinner-table fight — biologist says 'prove God exists,' philosopher says 'you can't disprove it either' — Popper's move is to say both challenges are category errors. Neither is a scientific question, because the God claim never ruled out any observation in the first place.

Iris Holm: The whole argument is a category error. Frankly, that's more respectful to theology than the biologist thinks he's being.

Cyrus Reed: Huh — yeah. But here's what I want to nail down because I think this is where people mix it up: 'not yet proven' versus 'unfalsifiable.' Those are not the same failure mode, right? Like — a hypothesis with zero confirming evidence can still be scientific if it makes predictions that future observations could actually contradict.

Iris Holm: That's the structural distinction. 'Not yet proven' is still in the arena — it just hasn't been tested enough. 'Unfalsifiable' never entered. Evolutionary biology, geology — they can't run controlled experiments on the past, but they absolutely rule out conceivable observations. Find a fossil rabbit in the Cambrian layer and evolutionary biology is in serious trouble.

Cyrus Reed: No direct experiment, but the inference structure keeps them inside science. That's — no, actually that's a really clean way to put it. The indirect evidence still has direction. It still excludes things.

Iris Holm: The question is whether that holds when the theories get complicated enough to immunize themselves — and that's the part that gets worse when we get to Lakatos's protective belt and what Feyerabend actually documented about methodological rules.

Cyrus Reed: But wait — that immunization move, that's not a bug, right? Like Lakatos is saying that's just... that's what labs actually do. You get a result that breaks your model and you don't throw out the model, you adjust something around the edges.

Iris Holm: That's the protective belt. Lakatos called the central commitments the hard core — untouchable by design — and everything around it the belt absorbs the hits.

Cyrus Reed: So the core never gets falsified because the auxiliary hypotheses take the blow instead.

Iris Holm: And Popper knew this was happening. He called arbitrary immunization illegitimate. But Lakatos said — no, that's not arbitrary, that's the structure of every working research programme in history.

Cyrus Reed: Huh — so Popper drew the line and Lakatos said the line cuts through the middle of normal science.

Iris Holm: Kuhn made the same observation differently. Normal science — his term — treats anomalies as puzzles, not refutations. The community assumes the core is right and works to solve the contradiction. That's not failure. That's how paradigms function.

Cyrus Reed: Wait, so what does it even look like when a programme becomes — what did Lakatos call it — degenerative?

Iris Holm: It stops generating novel predictions. It only retrofits anomalies after the fact. The belt is working too hard, nothing new is coming out the front end.

Cyrus Reed: Which is — okay, that's string theory, isn't it. No new predictions current experiments can touch. We just keep adjusting the framework.

Iris Holm: String theory, inflationary multiverse — both sit in that uncomfortable zone. Widely accepted in theoretical physics, generate almost nothing that any foreseeable experiment could actually contradict. By Popper's criterion, they're uncomfortably close to unfalsifiable.

Cyrus Reed: No way physicists accept that framing.

Iris Holm: They don't. And here's where Feyerabend becomes the uncomfortable witness. His argument in 'Against Method' — documented, not theoretical — every methodological rule in science, including falsifiability, has been productively violated at some point. Not as error. As advance.

Cyrus Reed: Hold on — productively violated. So the moments science moved fastest were the moments it broke Popper's rule?

Iris Holm: That's the crack. Falsifiability might be descriptively false at exactly the moments it most matters. Maxwell pushed this further — Popper never actually justified why following falsificationist rules gives science a better epistemic shot than alternatives. He just asserted it.

Cyrus Reed: But Sfetcu's counter is — wait, actually this is the tension — Sfetcu argues the criterion survives because Popper never claimed one counter-example topples a theory. So maybe we've been reading him too strictly this whole time?

Iris Holm: That defense saves Popper by making him weaker. If the criterion is flexible enough to survive Lakatos, it's flexible enough to let string theory in. You can't have both.

Cyrus Reed: And — no, what actually breaks my head is this. Freudian psychoanalysis was Popper's canonical unfalsifiable system, right? But modern evolutionary psychology has become more falsifiable over time. Specific predictions, testable hypotheses. It's like — it crossed the line. Can a knowledge system migrate into science?

Iris Holm: Which means the line isn't fixed. And that's the sharpest question — is Popper's criterion wrong, or is physics currently in violation of it and hasn't admitted it yet?

Cyrus Reed: And that's — wait, I can't shake this. Because the demarcation problem, technically, is still unsolved. Like, no successor criterion has actually achieved consensus. We've had fifty years of really smart people poking at Popper's line and nobody's drawn a better one. So we're left with a criterion that can't quite describe what modern physics does, and physics that won't quite admit it's drifted outside the boundary — and I don't know which of those is the more uncomfortable truth.

Iris Holm: Both carry a cost. Revise the criterion to let string theory in — you've weakened the boundary enough that it's hard to say what it's still keeping out. Hold the line and physics at its frontier is technically in non-scientific territory by its own foundational standard. Neither answer is clean.

Cyrus Reed: Which is — yeah. And actually that brings me back to Pauli. Because 'not even wrong' started the whole conversation as an insult, and now it's sitting there as a genuine open question about theoretical physics itself. String theory isn't wrong. It might be not even wrong. And the people saying that aren't dismissing it — they're worried about it.

Iris Holm: Popper drew the line to protect science from unfalsifiable claims bleeding in from outside. The place it's breaking is inside. That's the quiet part.