Karl Popper, writing in the twentieth century, proposed a criterion for distinguishing genuine science from things that merely resemble it: falsifiability. On Popper's account, what makes a theory scientific isn't primarily how much confirming evidence has accumulated in its favour, but whether the theory makes specific, risky predictions that some conceivable observation could, in principle, prove false. A theory so flexible that it can be reinterpreted to accommodate literally any possible outcome, Popper argued, isn't making a genuine empirical claim about the world at all, however scientific it might otherwise sound.
Confirming evidence is cheap, a genuine risk of failure is not
Popper's key insight was that confirming evidence, by itself, is a surprisingly weak test of a theory's scientific status, because a sufficiently vague or flexible theory can often be made to fit almost any observed outcome after the fact, regardless of whether the theory is actually true. What genuinely tests a theory, on his account, is whether it forbids specific possible outcomes in advance — sticking its neck out by predicting something risky enough that failing to observe it would count as a real, specific failure for the theory, not something that can simply be reinterpreted or explained away without cost. A theory that predicts everything, in this sense, actually predicts nothing, because there's no possible observation that could ever count against it.
Falsifiability as a line between science and non-science, not truth and falsehood
Crucially, falsifiability is a criterion for what counts as science, not directly a test of whether a given theory is actually true — a theory can be perfectly scientific, in Popper's sense, and still turn out to be false once it's actually tested against the evidence; in fact, on Popper's account, that's exactly the kind of genuine testing that makes a claim scientific in the first place. Conversely, a theory that isn't falsifiable at all — one so flexible it can absorb any observation without ever risking a clear failure — doesn't thereby become automatically wrong; it simply falls, on Popper's account, outside the category of empirical science altogether, whatever other value it might have as philosophy, ethics or interpretation.
What we're still unsure about
Popper's original falsifiability criterion, and its historical influence on how scientists and philosophers think about demarcating science, are well documented in the philosophy of science literature. What remains genuinely, actively debated among philosophers of science is whether falsifiability alone can fully and cleanly separate science from non-science in every real case — later philosophers of science have raised significant challenges, pointing out that scientists in practice often don't abandon a theory the moment a single observation seems to contradict it, preferring instead to investigate whether the anomaly reflects a flaw in the experiment, an auxiliary assumption, or a genuine problem with the theory itself, a messier real-world picture that some philosophers argue Popper's original, cleaner criterion doesn't fully capture on its own.
This sits inside Scientific Epistemology & Falsificationism, one of seven topics in Epistemology, one of five domains in Philosophy, one of seventeen subjects the app can quiz you on.