This is a brief, question-based outline of some Critical Fallibilist epistemology.
What is knowledge? Information that’s effective for (adapted to) a purpose. It has the appearance of design. It’s not arbitrary nor random.
How is knowledge created? Evolution (replication with variation and selection). The explanation of why evolution works is that it corrects errors.
What is learning? Creating knowledge.
What are ideas? Replicators used by intelligences. They’re a general purpose data structure which can hold any kind of information, including criticism of other ideas.
What is intelligence? The ability to learn, to create knowledge by evolving ideas.
What are degrees of intelligence? Good ideas enable more effective thinking. This only applies to intelligent agents (not e.g. ChatGPT, which doesn’t evolve or even have ideas).
The word “intelligence” has two meanings? Yes, which is confusing. It’s because other epistemologies look at the issue differently. You can talk about how smart someone is to refer to degrees of intelligence, and talk about whether or not something is a knowledge creator for binary intelligence.
What are the plausible alternatives to evolution for how knowledge is created? Nothing. Just like there are no known plausible alternatives to genetic evolution.
How does human thinking work in familiar terms? We guess ideas and try to refute our guesses in order to fix our mistakes. In other words, it’s brainstorming (replication with variation) and criticism (selection).
Are there methods of creating true, good or probable ideas? No. We have to use error correction, not avoid errors in the first place. (It’s possible in theory to build criticism filters into brainstorming/replication to prevent some ideas from being generated, rather than criticizing them immediately after they’re generated. Whether humans do that is a matter for future scientists. It might save computing resources but doesn’t affect the philosophical concepts.)
Most of the ideas I brainstorm seem pretty good, not random. Explain? You do lots of criticism/selection subconsciously. By the time an idea reaches the conscious level, it’s already had a lot of error correction.
How should ideas be evaluated? By whether they have a known error or not.
What’s an error? A reason a piece of knowledge fails at a purpose/goal.
What’s a criticism? An explanation of an error in an idea.
How do counter-arguments work? Criticisms are ideas. Criticisms, like all ideas, can be criticized.
What about positive arguments? Infinitely many positive arguments for an idea are logically compatible with it containing an error, so in some sense they don’t matter. By contrast, a single criticism can refute an idea.
Are positive arguments worthless? No. Many positive arguments can be translated to negative arguments. So they can be thought of as indirectly making valid, negative arguments.
How is evidence used? Evidence is used within arguments. Criticisms often refer to evidence. Evidence always requires interpretation by ideas; it can’t be used raw. You might say, “I saw some photons. My interpretation is there’s a dog over there. Your theory claims dogs don’t exist. Therefore your theory is refuted by counter-example.” Any part of this could potentially be questioned and be subjected to counter-arguments.
Which purposes should knowledge evaluated for? Whatever purpose(s) it was evolved for or whatever purpose(s) you want to use it for.
Can the same idea receive multiple different evaluations in relation to different purposes? Yes. CF evaluates IGCs not ideas. An IGC is group of three things: an idea, goal and context. Ideas shouldn’t be evaluated in isolation.
Are some purposes bad? Yes. Intelligences, unlike genetic evolution, can choose purposes. Purposes are ideas which can be criticized and evolved just like other ideas. A common mistake is to use vague purposes like “I want to buy a good house.” It’s hard to evaluate what is or isn’t an error for a vague purpose.
What about evaluating ideas by degrees of belief or goodness? No. Karl Popper allowed evaluating ideas by how well (to what degree) they survived criticism but not by positive justification. Popper didn’t give arguments to defend concepts like strong and weak arguments; I think he just took it for granted like everyone else. But errors, error correction, success and failure are all best viewed as binary issues. I think I improved on Popper by rejecting partial/indecisive/degree arguments in addition to rejecting positive, justifying arguments.