Falsifiability

How it applies

• The asymmetry behind the idea: No finite series of confirming observations establishes a universal statement, but a single counterexample contradicts it — the textbook illustration is "all swans are white" against one black swan. Popper used this asymmetry between verification and refutation as his way around the problem of induction associated with David Hume, and as the reason to test theories by trying to break them rather than by collecting supportive cases.
• A criterion of demarcation, not of meaning: Popper presented falsifiability as the line between empirical science and non-science, and distinguished his criterion from the verifiability criterion of meaning discussed in the Vienna Circle. On his account an unfalsifiable statement is not thereby meaningless or worthless; metaphysical, mathematical and definitional statements simply are not empirical claims that observation can settle.
• Degrees of testability: The more a theory forbids, the higher its empirical content and the more falsifiable it is. Popper preferred bold, highly falsifiable conjectures and treated surviving severe tests as corroboration rather than proof — corroboration, in his vocabulary, never turns into verification.
• Methodological side: Popper also warned against rescuing a refuted theory by unprincipled auxiliary assumptions, a move he called a conventionalist twist; such repairs can leave the wording intact while removing what made it testable. The same failure mode appears in documentation when a claim is widened until nothing could contradict it.
• Writing checkable claims: Falsifiability is a useful drafting check for specifications, acceptance criteria, release notes and marketing-adjacent technical copy: state thresholds, conditions, tolerances and scope so that a reader or a test could show the statement wrong. Statements that forbid nothing observable cannot be reviewed, tested or traced.
• Review and diagnosis: In design reviews and root-cause work, the question "what would we expect to see if this assumption were false?" turns a belief into a test. It helps separate assumptions that have been checked from assumptions that have only been repeated, which is also why assumptions are recorded explicitly in systems engineering practice.
• Limits and criticism: Strict falsifiability sits awkwardly with statistical hypotheses, since no single observation contradicts a probability statement, and with the Duhem–Quine point that a failed test hits a whole bundle of hypothesis plus auxiliary assumptions and instrument theory, so the error cannot be localized by logic alone. Thomas Kuhn and Imre Lakatos criticized naive falsificationism on historical grounds; these debates are open in the literature, and the entry below presents them as discussion, not as settled results.
• Where it shows up outside philosophy: Falsifiability and testability are cited well beyond philosophy, including in US evidence law, where the Supreme Court's Daubert decision (1993) listed whether a theory or technique can be and has been tested among the factors for admitting expert scientific testimony. Invoking falsifiability is a reasoning standard only: it makes a claim checkable and does not by itself make a product safe, a document compliant, or a risk assessment or the required safety information dispensable.

Falsifiability vs. falsification

Falsifiability is a logical property of a statement — that some possible observation would conflict with it. Falsification is an event: a test or observation that actually does conflict with it. A falsifiable statement may never be falsified, and "falsifiable" says nothing about whether the statement is false; conversely, a statement that no observation could contradict is unfalsifiable but not thereby untrue. Confusing the two is the usual source of the misreading that calling a claim falsifiable is a criticism of it. The related distinction between deriving consequences from accepted starting points and generalizing from cases is covered under first principle (philosophy) and Inductive reasoning.

External references

• Stanford Encyclopedia of Philosophy: Karl Popper
• Stanford Encyclopedia of Philosophy: Science and Pseudo-Science
• Stanford Encyclopedia of Philosophy: Underdetermination of Scientific Theory