Critical rationalism

1. No observation, however abundant, proves a universal law: a thousand white swans do not establish that all are. Science therefore does not proceed by piling up proofs.
2. But a single counter-observation can refute a law: what makes a theory valuable is that it forbids observable things, that it runs the risk of being wrong. What is scientific is what is falsifiable.
3. This fallibility does not, however, make science merely one discourse among others. Against scientism, which holds science to be the only legitimate source of knowledge in every domain, critical rationalism acknowledges that science cannot answer every question. Yet, in domains of reality that lend themselves to objective experience, scientific knowledge remains the most legitimate and the most capable of approaching truth. Meaning, values, and lived experience may belong to other forms of discourse, such as art, ethics, or metaphysics, whose claims should likewise remain open to examination and criticism.

Therefore So science is neither the only thing entitled to say everything, as scientism would have it, nor one knowledge among others equally valid, as pluralism would have it: it remains the model of rigorous knowledge, one that acknowledges its limits without ceding its title to what does not expose itself to disproof.

1. Critical rationalism makes refutation the heart of science: a theory falls as soon as a fact contradicts it.
2. But one never tests a hypothesis alone, always with a bundle of auxiliary hypotheses. A disproof says that one piece is at fault, not which; one can always save the theory by correcting another piece.
3. In fact, no great theory has been dropped over a single anomaly: scientific communities change frameworks for historical, social, aesthetic reasons as much as logical ones. 'Falsifiability' does not by itself govern how the sciences move.

Therefore So clean demarcation by refutation is a reassuring myth: the boundary of knowledge is blurrier, more historical, than critical rationalism claims.