Falsificationism

In brief

Karl Popper (1902 to 1994) asked a sharp question: what marks the line between science and everything that dresses up as science? His answer, first set out in Logik der Forschung (1934, English as The Logic of Scientific Discovery in 1959), was falsifiability. A theory is scientific only if it forbids certain observations, so that some possible result would show it false. "All swans are white" is scientific because a single black swan sinks it. A theory that fits every conceivable outcome forbids nothing and explains nothing. From this Popper drew a picture of how knowledge grows: not by accumulating confirmations, which prove little, but by making bold conjectures and then attacking them. The good theory is the one that has survived the fiercest attempts to refute it, and it is never proven, only not yet refuted.

The full treatment

The problem it answers

Popper wrote against two things at once. The first was the problem of induction, stated most cleanly by David Hume (1711 to 1776): no number of observed instances can logically establish a universal law, because the next instance might differ. If science cannot be founded on confirming observations, on what can it be founded? The second was a live intellectual scene in 1920s Vienna. Popper noticed that Einstein's general relativity, Marx's theory of history, Freud's psychoanalysis, and Alfred Adler's individual psychology were all presented as sweeping explanatory triumphs. But they behaved differently. Einstein's theory made a risky prediction: light from distant stars would bend by a precise amount near the sun, and if it did not, the theory was wrong. Marxism and psychoanalysis, by contrast, seemed able to absorb any event. A person's behavior confirmed Freud whether they did the thing or its opposite. That flexibility looked like strength to their admirers. To Popper it was the fatal weakness. A theory compatible with every possible observation tells you nothing about the world.

How it works

The engine of falsificationism is a logical asymmetry. Universal statements cannot be verified by observation, but they can be falsified by it. No finite run of white swans proves "all swans are white," yet one black swan refutes it outright. Confirmation is logically weak; refutation is logically decisive. Popper turned this asymmetry into a method. Science does not begin with neutral observation and grind out laws by induction. It begins with a problem, leaps to a bold conjecture, deduces testable consequences, and then subjects those consequences to severe tests designed to break the theory. A test is severe when the theory forbids a likely-seeming outcome, so that passing the test is genuinely surprising. A theory that survives such tests is corroborated, Popper's careful word, meaning it has proven its mettle so far, not that it has been shown probable or true. Knowledge, on this account, is permanently conjectural. We never reach certainty; we only eliminate error and press on with the survivors.

The demarcation criterion

The heart of the view is demarcation: the line between science and non-science (Popper stressed this is not the line between sense and nonsense). Metaphysics, he held, can be meaningful and even valuable, but it is not science if nothing could count against it. Falsifiability is the criterion. A statement earns scientific status by its degree of testability, by how much it forbids. "It will rain or not rain here tomorrow" forbids nothing and is empty. "It will rain here tomorrow" forbids the dry sky and says something. The more a theory prohibits, the more it risks, and the more it is worth. This reframes a virtue: a good scientific theory is not one that explains a great deal comfortably, but one that could easily have been refuted and was not.

The distinction that matters most

Falsificationism is often confused with two neighbors it explicitly rejects. It is not verificationism, the logical positivist demand that a statement be confirmable by experience to be meaningful. Popper inverted that: confirmation is cheap and the mark of science is refutability, not verifiability. Nor is it the claim that scientists actually spend their days trying to destroy their own theories, which as a matter of history they often do not. It is a claim about logic and about what makes a theory scientific in status, a normative account of what good science should look like, not a tidy report of laboratory life. Missing this distinction is the source of most misplaced objections.

Lineage

Falsificationism descends from the problem of induction. Hume showed that inductive inference has no logical justification; Popper accepted this fully and, remarkably, agreed that induction is a myth, then built a philosophy of science that does without it. Where the empiricist tradition of Francis Bacon (1561 to 1626) and John Stuart Mill (1806 to 1873) held that science gathers facts and generalizes, Popper made deduction, not induction, the logic of testing. His immediate context was the Vienna Circle of logical positivists, whose verifiability principle he knew well and defined himself against, though he was never a member. The swan example that anchors the view runs back through the same skeptical lineage that produces the black swan as a symbol of the unpredictable. Popper's later ally and critic Imre Lakatos (1922 to 1974) extended the framework, and his rival Thomas Kuhn (1922 to 1996) recast the whole picture in historical terms.

The strongest case for it

Falsificationism captures something real and durable about scientific integrity. It gives a clean, usable test for pseudoscience: ask what would prove the claim wrong, and if the answer is nothing, be suspicious. Astrology, conspiracy theories, and certain grand systems fail exactly here, always able to explain away the misses. The account also honors the courage of good science. Einstein staked general relativity on the 1919 eclipse observations by Arthur Eddington, results that could have destroyed it; that willingness to be wrong is the opposite of the self-sealing theory. Falsificationism dissolves the ancient problem of induction rather than solving it, sidestepping a puzzle that had resisted centuries of attack. And it connects to parsimony: a simpler theory forbids more and is easier to refute, so falsifiability supplies a hard-nosed reason to prefer it, one that Popper himself drew. As a warning against theories that cannot lose, the criterion has never been improved upon.

The strongest case against it

The objections are serious, and most philosophers of science now regard strict falsificationism as untenable in its original form.

The deepest is the Duhem-Quine thesis. Pierre Duhem (1861 to 1916), in The Aim and Structure of Physical Theory (1906), argued that a hypothesis is never tested alone: any prediction depends on a web of auxiliary assumptions (about instruments, background conditions, other theories). When a prediction fails, logic alone cannot tell you which part of the web is at fault. You can always save the main hypothesis by revising an auxiliary. Willard Van Orman Quine (1908 to 2000), in Two Dogmas of Empiricism (1951), pushed this to holism: no statement is immune to revision and none faces the tribunal of experience by itself. Clean refutation, the linchpin of Popper's asymmetry, becomes elusive in practice.

The historical objection came from Thomas Kuhn in The Structure of Scientific Revolutions (1962). Real scientists, Kuhn observed, do not abandon a theory at the first anomaly. During "normal science" they work within a paradigm and treat failures as puzzles to be solved, not as refutations. Mercury's misbehaving orbit did not refute Newton; astronomers hunted for a hidden planet. Theories are discarded only when a rival paradigm exists to replace them, in a revolution, not by isolated falsification. If Popper described how science should work, Kuhn described how it does, and the gap is wide.

Lakatos (1922 to 1974), sympathetic to Popper, conceded much of this and offered a repair in Falsification and the Methodology of Scientific Research Programmes (1970): we should judge not single theories but research programmes over time, calling a programme progressive if it predicts novel facts and degenerating if it only patches failures after the fact. Paul Feyerabend (1924 to 1994), Popper's former student, went further in Against Method (1975), arguing that no single rule, falsifiability included, fits the messy reality of scientific discovery, and that rigid falsificationism would have strangled theories that later triumphed. There is also a nagging internal problem: strict falsificationism has no account of why we should trust a corroborated theory for prediction, since corroboration is explicitly not evidence of truth or probability.

Where it stands now

Popper reshaped how scientists talk about their own work. "Falsifiable" and "testable" entered the working vocabulary of science, and the demarcation instinct, ask what would prove this wrong, remains one of the sharpest tools against pseudoscience and a staple of textbooks and courtrooms. As a complete philosophy of science, though, naive falsificationism is widely held to have failed, undone by Duhem-Quine holism and Kuhn's history. The living descendants are subtler: Lakatos's research programmes, Bayesian accounts that treat evidence as raising or lowering probability by degrees rather than as clean refutation, and a general recognition that theory choice weighs many virtues at once. The mature verdict is that Popper identified a genuine and vital feature of good science, its openness to refutation, while overstating it as the single mechanism by which science works. The criterion endures as an ideal and a discipline, not as a mechanical rule.

Test yourself

Take a belief you hold strongly, about people, politics, or the world, and ask the Popperian question in earnest: what observation, if you saw it, would make you give this up? If you find that nothing could, notice what that means. It does not prove the belief false. It shows the belief is doing no scientific work, and that you may be holding it in a way that no evidence could ever touch.