How Does Error Correction Create Knowledge?

TL;DR

Knowledge grows through conjecture, criticism, and error correction, not by proving beliefs permanently true. The same pattern appears in natural selection, memetic evolution, invention, and markets: alternatives encounter reality, failures are removed, and better explanations survive provisionally while remaining open to correction.

Transcript

Deutsch has this great view of the world where he believes that everything important is understandable by a single human. By important, he means underlying base theories that drive most of reality, and he fixates on four theories. I could argue maybe there's a few more, especially if you start getting into Adam Smith and The Wealth of Nations and a... Read More

Key Insights

• Knowledge is conjectural rather than permanently justified as true. A serious theory earns provisional acceptance by surviving attempts to expose its errors, while remaining open to later criticism, correction, or replacement by an explanation that accounts for reality more accurately.
• A good explanation is not equivalent to a random guess. It is a conjecture that has encountered demanding tests and criticism without being refuted, which distinguishes serious knowledge from unsupported speculation while avoiding the false promise of certainty.
• Bayes’ theorem is a legitimate statistical formula, but the discussion rejects Bayesianism as a complete epistemology. Updating confidence can compare existing theories, yet it cannot generate the new explanations required to solve problems or account for previously unexplained observations.
• A crucial experiment distinguishes between competing theories by producing an outcome that rules out one while leaving another viable. Newtonian gravity and general relativity illustrate this process, with the surviving theory retained only until new problems reveal further errors.
• General relativity is the best available theory of gravity in the discussion, but it is not treated as final. Scientific progress produces better approximations to reality, and even successful theories remain vulnerable to correction because no theory can be guaranteed free of error.
• Natural selection is an error-correcting process in which environmental feedback removes organisms that are insufficiently fit. The neo-Darwinian account identifies genes, rather than groups, herds, or whole individuals, as the units whose variations are selected across generations.
• Memetic evolution applies an evolutionary pattern to ideas. Humans create explanatory knowledge, criticism removes ideas that fail, and better ideas persist, making genetic evolution a prelude to a future increasingly shaped by the creation and selection of knowledge.
• Humans are exceptional because they can explain visible events through unseen mechanisms. Other animals primarily learn from repeated events, while human explanatory reasoning enables deliberate criticism, correction, and open-ended progress beyond instinctive pattern recognition.

Questions & Answers

Q: What is Deutsch’s theory of knowledge?

Deutsch’s theory of knowledge, as presented here, says that knowledge consists of conjectures that have survived attempts at refutation. A theory is not established by proving it permanently true. Instead, people propose explanations, search for flaws, discard explanations that fail, and provisionally retain those that withstand criticism. Because every theory can contain errors, knowledge can continue improving without a final endpoint.

Q: Why can no truth be permanently justified?

No method can demonstrate that a piece of knowledge is completely and permanently true, according to the discussion. Evidence may fail to refute a theory, but future criticism or experiments can still uncover an error. Treating knowledge as conjectural acknowledges this limitation while preserving a rational standard: explanations are accepted because they have survived serious tests, not because certainty has been achieved.

Q: Why is Bayesianism not a complete epistemology?

Bayesianism can update confidence in theories as evidence accumulates, and Bayes’ theorem remains useful as a statistical formula. The criticism is that confidence does not establish truth, and Bayesian updating cannot generate new explanatory theories. At best, it ranks theories already under consideration. Knowledge creation also requires imaginative conjectures and criticism capable of revealing why an existing explanation fails.

Q: What does Newtonian gravity show about scientific confidence?

Before the experiment discussed in connection with 1919, observations of gravity had repeatedly appeared consistent with Newtonian theory. A Bayesian might therefore have been most confident immediately before the theory was shown to be false. The Popperian interpretation avoids this problem by saying Newton’s theory was always an imperfect approximation that contained useful truth but required correction by general relativity.

Q: Can there be a final theory of gravity?

The discussion argues that there can be no final theory of gravity. Newtonian gravity approximated reality but was corrected by general relativity, which is currently presented as the best available explanation. General relativity must also remain open to criticism and future correction. Scientific theories therefore progress as increasingly accurate approximations rather than becoming guaranteed, complete, and permanently unrevisable descriptions of reality.

Q: How is natural selection an error-correcting process?

Natural selection tests organisms against their environment. Organisms that are insufficiently fit may die without passing on their genes, while fitter genetic variations survive and spread. This environmental feedback removes unsuccessful variations without requiring a creator or purposeful stretching of inherited traits. The neo-Darwinian account identifies the gene as the unit of selection rather than the group, herd, or individual.

Q: What is the relationship between genetic and memetic evolution?

Genetic evolution produces variation through mutations and filters it through natural selection. Memetic evolution follows a comparable structure: humans create ideas, then criticism removes ideas that fail. Deutsch’s view treats genetic evolution as a prelude because explanatory knowledge lets humans conduct a faster, open-ended evolutionary process in which ideas, rather than only genes, increasingly influence the future of the universe.

Q: Why are humans considered exceptional in this framework?

Humans are considered exceptional because they can create good explanations and describe what is seen through unseen underlying mechanisms. Other animals are characterized as relying mainly on repeated experience, such as expecting a recurring event to continue. Human reasoning can instead propose theories, criticize them, identify errors, and improve knowledge, enabling memetic evolution and open-ended progress beyond biological adaptation alone.

Summary & Key Takeaways

• Deutsch’s epistemology replaces justified true belief with conjectural knowledge. A theory qualifies as knowledge not because evidence proves it true, but because it has survived serious attempts at refutation. This fallible view leaves every explanation open to correction and makes continued intellectual progress possible rather than treating current conclusions as final.

• Bayesian reasoning can legitimately update statistical confidence, but the discussion argues that it cannot generate explanations or serve as a complete theory of knowledge. Newtonian gravity accumulated supporting observations before a crucial experiment favored general relativity, illustrating why increasing confidence does not establish that a theory is true or immune to replacement.

• Popperian epistemology and Darwinian evolution share an error-correcting structure. Organisms undergo selection through environmental feedback, while human ideas undergo criticism. Humans are presented as exceptional because they can create explanatory knowledge, moving evolution beyond genetic change toward memetic evolution in which ideas increasingly shape the universe’s future.