Falsifiability

Why do we trust physics but not astrology? Both use elaborate calculations, both have a thousand-year history, both make 'predictions'. The difference is one key property: physics can be refuted, astrology cannot. That is what turns physics into knowledge and astrology into entertainment.

• **Medicine:** the FDA requires randomized trials, falsifiable tests of efficacy. Without them, a drug is not approved
• **Engineering:** engineering theories are falsifiable. A bridge either holds the load or collapses. That is what makes reliable construction possible
• **Critical thinking:** when someone offers a 'theory of everything', ask what could refute it. If nothing could, it is not a theory

What falsifiability is

**Falsifiability** is the property of a statement that makes refutation by experience possible. If no experience could refute the statement, it says nothing concrete about the world. This is the key criterion of scientific status.

**Falsifiable statement:** 'All swans are white' - Refutation: find a black swan - Status: scientific (refuted in Australia) **Unfalsifiable statement:** 'Everything happens by the will of fate' - Any event 'confirms' it (was predestined) - No event can refute it (also predestined) - Status: non-scientific (carries no information)

**Important:** falsifiability is not the same as falsity. 'The Earth is round' is falsifiable (could have been refuted by finding an edge) but true. 'God loves us' is unfalsifiable (any event can be read as love or a test), but that does not make it false. It just sits **outside science**.

Popper's criterion

**Karl Popper** (1902-1994), a philosopher of science, proposed falsifiability as the **demarcation criterion**: the boundary between science and non-science. Popper was after one question: why is Einstein's physics science, while Freud's psychoanalysis is not?

**Popper's contrast:** **Einstein (1919):** 'If starlight is not deflected by the Sun by the predicted angle, my theory is wrong.' → A risky, testable prediction. It could have failed. **Freud:** 'A person is aggressive because they suppress aggression.' 'A person is gentle because they sublimate aggression.' → Any behavior gets 'explained'. Nothing refutes the theory.

**Critique of Popper:** real science is messier. Scientists do not discard theories after the first refutation. Often the issue is in the experiment, not the theory. Quine-Duhem showed: you can always 'save' a theory by adjusting auxiliary hypotheses. Still, Popper remains useful for spotting pseudoscience.

Pseudoscience

**Pseudoscience** is a field that looks like science (terminology, 'studies', 'experts') but does not follow the scientific method. The main signal: unfalsifiability. The theory explains everything, regardless of the result.

**Signs of pseudoscience:** • **Immunity to refutation:** the theory 'explains' any outcome • **Ad hoc hypotheses:** patches added when problems appear • **No progress:** the theory does not evolve for decades • **Appeal to authority:** 'the great X said so' • **Conspiracy of silence:** 'official science is hiding it' • **Reverse logic:** absence of evidence = evidence of conspiracy

**'Science cannot say' vs 'It is unscientific':** some questions sit outside the competence of science (ethics, aesthetics, the meaning of life). That is not pseudoscience, just a different mode of inquiry. Pseudoscience is when something claims scientific status without meeting scientific standards.

Testable predictions

The best test of scientific status is the ability to make **risky, testable predictions**. Not retrospective 'explanations' of known facts, but predictions of what has not yet been observed. The more specific the prediction, the stronger the theory if it comes true.

**Examples of risky predictions:** • **Einstein:** light bends near the Sun by 1.75 arc seconds (not 0, not 100, a specific number) • **Mendeleev:** elements with atomic weights 68, 72, 45 and specific properties should exist → Ga, Ge, Sc were later discovered • **Evolution:** there should be a transitional form between fish and land animals → Tiktaalik was found • **Higgs:** a particle with mass around 125 GeV should exist → detected at the LHC

**Postdictions** (explanations of already known facts) are weaker than predictions. It is easy to invent an explanation after the fact. Evolution would be stronger if it had predicted DNA before the discovery (and in part it did, with the 'hereditary code').

**In practice:** when someone offers you a 'theory', ask what concrete, risky, not-yet-tested prediction it makes. If the answer is fuzzy or 'everything is already explained', it is probably not science.

Key Ideas

• **Falsifiability** is the possibility of refutation by experience. Popper's criterion of scientific status
• **Asymmetry:** one refutation is logically stronger than a million confirmations
• **Pseudoscience** is a claim to scientific status without meeting scientific standards (unfalsifiability, ad hoc patches)
• **Risky predictions** are the best test of a theory: concrete, testable, made in advance

Related Topics

Falsifiability connects to the scientific method and critical thinking:

• Scientific method — Falsifiability is the key criterion of a scientific hypothesis
• Confirmation bias — We look for confirmations, but Popper says look for refutations

Вопросы для размышления

• Which of your beliefs are unfalsifiable? (For example: 'Everything happens for the best', 'Karma is real'.) Does that make them wrong?
• Recall a time you changed your mind because of refuting evidence. How did it feel?
• How would you design an experiment to test the claim 'positive thinking improves health'?

Связанные уроки

• dm-01