Falsification | Kaelan

Karl Popper’s The Logic of Scientific Discovery (1934) solved a problem that had plagued philosophy of science: what separates science from non-science?

Verification can’t work. No finite number of observations proves a universal claim. You see a million white swans; you haven’t proven all swans are white. The next swan might be black.

Popper’s answer: falsifiability. A claim is scientific if it specifies what would prove it wrong. “All swans are white” is scientific because a black swan would falsify it. “Everything happens for a reason” is unfalsifiable (every outcome confirms it) and therefore not scientific.

The asymmetry is logical. You can’t verify universals, but you can falsify them. One counterexample destroys a general claim. Science advances not by confirming theories but by failing to falsify them. Theories that survive attempts to break them earn provisional trust — not because they’re proven true, but because they haven’t yet been proven false.

This reframes scientific method. Don’t ask “what supports this theory?” Ask “what would refute it?” The valuable experiment is the one that could kill your hypothesis. If nothing could, you’re not doing science.

Einstein’s general relativity made a specific prediction: light bends around massive objects by a calculated amount. Arthur Eddington’s 1919 solar eclipse expedition tested this. If the light hadn’t bent, or bent differently, the theory would have been falsified. It bent as predicted. Einstein wasn’t proven right — he survived a falsification attempt.

Contrast with Freudian psychoanalysis, which Popper used as a counter-example. Patient improves? The therapy worked. Patient gets worse? Resistance to therapy, which confirms the theory. Patient stays the same? Therapy is still ongoing. No outcome could falsify the theory. This doesn’t mean Freud was wrong — it means the theory isn’t scientific in Popper’s sense.

Falsification has limits. Theories don’t face evidence alone — they come with auxiliary hypotheses about instruments, conditions, and interpretation. When a prediction fails, you can save the theory by adjusting auxiliaries. The history of science is full of this: anomalies explained away until they couldn’t be.

Kuhn and Lakatos criticized Popper on this point. Scientists don’t abandon theories at first falsification. They protect productive frameworks and shift when a better alternative emerges. Falsification describes an ideal; actual science is messier.

The practical value isn’t in strict application but in orientation. Ask of any belief: what would change my mind? If you can’t answer, you’re not thinking — you’re defending.

The business plan that would work regardless of market conditions isn’t a plan. The political position that interprets all evidence as confirmation isn’t a position. The relationship diagnosis that explains every behavior isn’t insight.

Falsifiable claims are vulnerable. That vulnerability is the point. It’s how maps get updated when they don’t match territory.