Key Principle
Popper constructs a formal autonomous probability calculus — a six-axiom system (A1-A3, B1-B2, C) where relative probability p(a,b) is primitive, interpretation-free, and remains defined even when p(b) = 0. From this he derives two results that close the door on probabilistic induction:
Universal laws have zero probability: p(x) = 0 = p(x,y) for any universal law x and finite evidence y. A universal law is an infinite conjunction of singular statements, each with probability less than 1; the infinite product converges to zero (Appendix *vii).
Degree of corroboration cannot be a probability: Proven by counterexample — identifying "degree of support" with probability produces a logical contradiction (Appendix *ix).
Why This Matters
The zero-probability result is the formal linchpin of the entire book. Without it, probabilistic inductivists could concede every philosophical argument in Chapters 1-4 and still retreat to: "Yes, but evidence at least raises the probability of laws." Zero probability blocks this retreat absolutely — evidence cannot raise zero to anything useful.
The autonomous calculus is infrastructure: standard probability systems (including Kolmogorov's) define conditional probability as p(a,b) = p(ab)/p(b), which is undefined when p(b) = 0. Since universal laws have zero probability, the standard definition makes it impossible even to state the likelihood of evidence given a law. Popper's system removes this technical barrier, making the zero-probability result and the corroboration formula mathematically coherent.
Good Examples
The die counterexample: Let z = "even number," x = "six," y = "not six." Evidence z supports x (raising it from 1/6 to 1/3) while undermining y (lowering it from 5/6 to 2/3). Yet p(y,z) = 2/3 > p(x,z) = 1/3 — the undermined hypothesis has higher "probability-support" than the confirmed one. If support = probability, we must say z simultaneously supports x and y-has-higher-support-than-x, which is contradictory (Appendix *ix).
The corroboration formula: C(x,y) = (P(y,x) - P(y)) / (P(y,x) - P(xy) + P(y)). A tautology (P(x) = 1) yields C = 0 regardless of evidence. A bold law (P(x) = 0, content = 1) can reach C approaching 1 if it predicts evidence that would otherwise be maximally surprising. "The content of a theory — which is the same as its improbability — determines its testability and its corroborability" (Appendix *ix).
Confirmability ceiling: C(x) = P(x̄) = 1 - P(x). Only theories with high content (low probability) can achieve high corroboration. A safe, high-probability theory cannot be well-corroborated because it was never at risk.
Counterpoints
Bayesian response: Bayesians argue that prior probabilities need not be zero — they can be assigned subjectively. Popper rejects subjective priors as arbitrary: the zero-probability result holds for logical probability, and any system that identifies confirmation with probability faces the die counterexample regardless of prior assignments (Appendix *ix).
Carnap's confirmation theory: Carnap attempted to build inductive logic by defining "degree of confirmation" as a probability. Popper's results show this programme is mathematically self-defeating: if confirmation is probability, then universal laws (the targets of science) can never achieve positive confirmation (Appendix *ix).
Incompleteness of formalization: Popper acknowledges that no formal measure captures the "ingenuity and sincerity of attempts at refutation." The quality of tests remains partly a judgment call — but the formal results establish lower bounds on what any adequate measure must satisfy (Appendix *ix).
Key Quotes
"A high degree of probability is therefore not an indication of 'goodness' — it may be merely a symptom of low informative content." — Karl Popper, Appendix *ix
"I regard the doctrine that degree of corroboration or acceptability cannot be a probability as one of the more interesting findings of the philosophy of knowledge." — Karl Popper, Appendix *ix
"The content of a theory — which is the same as its improbability — determines its testability and its corroborability." — Karl Popper, Appendix *ix
Rules of Thumb
- Never confuse the probability of a theory with its scientific standing — universal laws have zero probability but can be highly corroborated
- Any claim that evidence "makes a theory probable" faces the zero-probability barrier for universal laws
- If a measure of scientific support behaves like a probability, it cannot track what scientists actually value (bold, informative theories)
- The retreat from "certainty" to "high probability" does not rescue induction — it is a retreat to zero
Related References
- Corroboration — How a Theory Stands Up to Tests — The qualitative account that this formalizes
- Content, Testability, and Degrees of Falsifiability — The conceptual chain that the formula encodes
- Critique of Positivism and the Vienna Circle — The broader anti-positivist programme these results serve