Critical Fallibilism (CF) is a philosophy I developed which deals with rationality, knowledge, and critical thinking and discussion. It builds most on Karl Popper's Critical Rationalism, which says we learn (create knowledge, solve problems) by an evolutionary process of conjectures and refutations. Popper rejected positive arguments (justifications) and induction. He advocated fallibilism and error correction.
CF's most important idea involves distinguishing between decisive and indecisive arguments. CF evaluates ideas by whether they succeed or fail at a purpose (achieve a goal, solve a problem, answer a question). Ideas are for something and can't be evaluated in isolation. The same idea can succeed at one purpose and fail at another (and indeed all ideas do that). Negative arguments (criticisms) argue that one or more ideas fail at one or more purposes. A purpose could be avoiding dehydration, having true ideas, or both. Any group of purposes joined with "and" (or other logical operators like "or") is a purpose.
A decisive argument (or group of arguments) contradicts the negation of its conclusion, so both can't be true. A decisive positive argument contradicts its target being false (it says the idea must succeed at its purpose). A decisive negative argument contradicts its target being true (it says the idea must fail at its purpose). If you accept the argument, you shouldn't accept the thing it contradicts, because they're incompatible.
Decisive negative arguments are reasonably common and accessible. We can find and point out errors such as counter examples or flawed logic. One error is often enough to fail at a purpose.
Decisive positive arguments are either rare or entirely inaccessible. Pointing out 1000 good things isn't enough to prove an idea will succeed at its purpose. Despite all those merits, there could still be an error that causes failure. Basically, errors have logical priority over positive traits. No matter how many good traits an airplane has, a single mistake in the engine can cause a crash.
Karl Popper argued that negative arguments are better than positive arguments. While he made several good points, I think there's another issue that Popper missed. Decisive arguments are better than indecisive arguments, and most or all of our decisive arguments are negative. Why? Decisive positive arguments require 100% proof; otherwise you could accept the argument, and accept that its conclusion is false, without contradiction. I think fallibilism excludes 100% guarantees against error in math, logic and every field. Even if you disagree about how broad fallibility is, you could still accept CF for science, philosophy and most fields.
Criticism is easier than proof because proof requires addressing all possibilities. A single counter-example or error can be a decisive criticism without considering vast numbers of possibilities. Popper discussed this with examples like "all ravens are black", which is refutable with one contradictory observation, but still not proven with a million compatible observations.
All our arguments are fallible and can be reconsidered with new ideas and evidence. But if you accept an observation of a white raven and some background knowledge, that contradicts "all ravens are black". There's no comparable way to prove it, even fallibly.
You can make decisive positive arguments using universal premises (e.g., "All men are mortal.") which assert their own completeness. But that just moves the problem: how do you prove that premise since observing a million mortal men is inadequate?
Indecisive negative arguments have the same basic flaw as indecisive positive arguments: you can accept the argument but still reject the argument's conclusion without contradicting yourself. There's no logical connection between the argument and the conclusion. Logically, indecisive arguments don't do anything. Contradiction is a powerful, useful logical tool used with decisive arguments, but we don't have good alternative tools to use with indecisive arguments. Compatibility (non-contradiction) lacks logical power despite sometimes misleadingly being called "support" or "confirmation".
If I observe a purple hamster or red tree, that is compatible with "all ravens are black"; compatibility is basically worthless without also using another concept like relevance. But even with a million relevant, compatible observations – e.g., observations of black ravens – a white raven could still exist. Relevance is very difficult to define and evaluate, but even if we had perfect knowledge of what was relevant, compatibility plus relevance would still be inadequate to make indecisive arguments effective. Nothing about seeing a lot of black ravens actually means that there couldn't be a white one.
If indecisive arguments are logically flawed, then why do they often appear to work moderately well? Many of them are convertible to decisive arguments. They have a valid point which is presented imperfectly. Similarly, many positive arguments are convertible to negative arguments. I hypothesize that arguments which can't be converted to decisive, negative form are wrong. Skipping the conversion step is often reasonable in low-stakes, low-precision, friendly contexts.
How do you convert arguments? Positive arguments point out good trait(s). To convert to negative form, criticize alternatives for lacking the good trait(s). Indecisive arguments involve more creativity to convert. They point out good or bad things without logically connecting that to success or failure at a purpose. To make them decisive, clarify the purpose, figure out criteria for success and failure, and point out how the bad things (or missing good things for alternatives) cause failure.
Focusing on negative arguments can take some getting used to because it's more indirect. Instead of arguing in favor of an idea, you criticize alternatives to that idea. You also try to criticize the idea. The conclusion you reach is the idea you can't find any decisive error in, despite trying.
If you have multiple ideas that you think will work, you can use any of them, or you can aim for a more ambitious purpose/goal. CF has methods for narrowing it down to exactly one non-refuted idea, but that's often not worth the effort. If you have no ideas you think will work, you can brainstorm/research more, adjust your purpose/goal to be easier, or give up and do something else.
If you don't find an error with an idea, it could still be wrong. Why use it? Because ideas you don't know are wrong are preferable to ideas you do know are wrong (already found a decisive error in). It never makes sense to use an idea for a purpose if your best understanding is that it will fail at that purpose.
Why?
What is the motivation for CF? A reasonable first impression would be that CF doesn't sound strictly wrong (you don't see a reason it can't possibly work), but it sounds inconvenient or cumbersome. So what's the upside to make it worth pursuing?
CF's main motivation is logical arguments showing that various other approaches cannot possibly work. CF prioritizes correctness over everything else. (I actually think CF is reasonably convenient and elegant once you get used to it, and has various merits besides correctness, but I acknowledge there's an initial learning curve.)
Two main alternatives to CF are induction and weighted factors. Serious problems with these are actually well known in the academic literature and have been written about by advocates of these approaches, not just by opponents like Popper. I'm not actually saying anything very new by claiming these approaches are flawed. I think people try to use them despite the known errors because they've largely given up on finding alternatives, and errors seem somewhat ignorable to people accustomed to indecisive thinking. Some people are ignorant of the problems with their approaches, but I think many experts are instead pessimistic about finding something better. Ignorance of problems reduces motivation to consider alternatives. And experts who understand the problems may not want to spend time studying or debating a new system that they find initially counterintuitive and aren't optimistic about.
The various schools of inductive thought can be seen as attempts to say that indecisive arguments sometimes partially work. Compensatory weighted factor approaches (where a high score at one factor can compensate for low scores at other factors) also try to use multiple indecisive arguments to reach a conclusion. Non-compensatory multi-factor approaches are less common but exist and have some overlap with CF.
I've written criticisms of induction and weighted factors before. For this article, I'll just say that I'm open to discussion if someone disagrees but wants to try to reach a conclusion about the matter. In a discussion, I only have to address one person's concerns instead of all potential concerns, so it can be easier and more focused (and provide valuable feedback and potentially criticism for me, too). I can provide a free account on my forum if someone accepts this invitation (email me).
Examples
The idea "go to McDonald's for a burger" succeeds at the purpose "get lunch" but fails at the purposes "get a gluten-free lunch" and "learn knitting". We should evaluate {idea, purpose} pairs rather than ideas alone. Being more precise, we can include context and evaluate a triple: {idea, purpose, context}.
Note: In other articles I've said "goal" instead of "purpose" and abbreviated the triple to IGC, but they mean the same thing. It's {idea/solution/answer/plan/explanation/argument/reasoning/option/alternative, purpose/goal/question/problem/objective/criterion, context/problem-situation/background-knowledge/scenario} triples. The triple is meant to work with any of the terms separated by slashes; you can use whichever is most natural. Sometimes answer and question work well, but in other contexts solution and problem make more sense. {idea, purpose/goal, context} is particularly generic and works well when you want to keep terminology consistent. Plurals like "goals" or "criteria" may be used because the goal is often a group: the goal could consist of multiple conjoined sub-goals or be to meet multiple criteria. Interestingly, goals and contexts are types of ideas which can be put in the first spot of their own IGC triples. Similarly, arguments are used to evaluate IGCs but they are also part of their own IGCs: arguments have goals and contexts and an argument's IGC can itself be criticized.
A criticism can apply to multiple {idea, goal} pairs. If my goal is to get a gluten-free lunch, the criticism "buns have gluten" applies to McDonald's and also to other ideas like Burger King and Wendy's. The same criticism can also work for other goals, e.g., getting a gluten free breakfast or dinner. It's often important to think using principles and make your criticisms broad enough to cover whole categories of ideas and goals instead of just an individual idea-goal pair.
An indecisive positive argument is "I'll go to McDonald's because their food tastes good". It's indecisive because the reasoning (McDonald's food tastes good) doesn't contradict "I won't go to McDonald's". Both can be true. Also, alternatives like "I should go to Burger King" aren't contradicted even though they imply not going to McDonald's (in a context where you're picking one restaurant for a specific meal). Indecisive means there's no logical problem with accepting the reasoning but also accepting a contradictory conclusion.
To convert to a negative argument, figure out what positive trait is being praised (food taste) and criticize alternatives for lacking it. E.g., "I won't go to Burger King because their food tastes bad." After converting, always check if the result is true. If you actually like the taste of Burger King's food, then it's wrong. In that case, the original indecisive, positive argument was also wrong to conclude that you should go to McDonald's specifically (not Burger King) because of taste.
To convert to a decisive argument, you often need to clarify your purpose/goal. Let's say I'm hungry now and I want to get restaurant food within five minutes of walking. Then an indecisive argument is "I'll go to McDonald's because it's close." A decisive argument is "I won't go to Burger King because it's over a five-minute walk away." More generically, "I won't go to any restaurant that's over a five-minute walk away." can rule out the idea of going to Burger King, Wendy's and more with a single criticism. To make location decisive, we figure out what proximity constitutes success or failure and we criticize ideas that fail.
Next Steps
If CF interests you, a good place to start is by categorizing arguments you make, read or hear as decisive or indecisive and positive or negative. You can do that without changing what arguments you use or like; it can start as a research project for informational purposes. A later step could be trying to convert some arguments to be decisive and negative and paying attention to how some convert well but others don't.
A more advanced issue is considering whether enough decisive, negative arguments exist (including via conversion) to do all of our reasoning with them, or if we need to use some non-convertible positive and/or indecisive arguments. Many people like decisive, negative arguments and see them as highly powerful and useful, but believe they're too scarce to use exclusively. A related issue to ponder is: Can you ask a series of yes or no questions to explore any issue, or are other types of questions (or non-questions) absolutely necessary? Yes or no questions are good at exposing decisive issues. "Does it cost under $50?" invites a decisive answer while "How expensive is it?" or "How happy am I with the price?" don't.
Critical Fallibilism (CF) is a philosophy I developed which deals with rationality, knowledge, and critical thinking and discussion. It builds most on Karl Popper's Critical Rationalism, which says we learn (create knowledge, solve problems) by an evolutionary process of conjectures and refutations. Popper rejected positive arguments (justifications) and induction. He advocated fallibilism and error correction.
CF's most important idea involves distinguishing between decisive and indecisive arguments. CF evaluates ideas by whether they succeed or fail at a purpose (achieve a goal, solve a problem, answer a question). Ideas are for something and can't be evaluated in isolation. The same idea can succeed at one purpose and fail at another (and indeed all ideas do that). Negative arguments (criticisms) argue that one or more ideas fail at one or more purposes. A purpose could be avoiding dehydration, having true ideas, or both. Any group of purposes joined with "and" (or other logical operators like "or") is a purpose.
A decisive argument (or group of arguments) contradicts the negation of its conclusion, so both can't be true. A decisive positive argument contradicts its target being false (it says the idea must succeed at its purpose). A decisive negative argument contradicts its target being true (it says the idea must fail at its purpose). If you accept the argument, you shouldn't accept the thing it contradicts, because they're incompatible.
Decisive negative arguments are reasonably common and accessible. We can find and point out errors such as counter examples or flawed logic. One error is often enough to fail at a purpose.
Decisive positive arguments are either rare or entirely inaccessible. Pointing out 1000 good things isn't enough to prove an idea will succeed at its purpose. Despite all those merits, there could still be an error that causes failure. Basically, errors have logical priority over positive traits. No matter how many good traits an airplane has, a single mistake in the engine can cause a crash.
Karl Popper argued that negative arguments are better than positive arguments. While he made several good points, I think there's another issue that Popper missed. Decisive arguments are better than indecisive arguments, and most or all of our decisive arguments are negative. Why? Decisive positive arguments require 100% proof; otherwise you could accept the argument, and accept that its conclusion is false, without contradiction. I think fallibilism excludes 100% guarantees against error in math, logic and every field. Even if you disagree about how broad fallibility is, you could still accept CF for science, philosophy and most fields.
Criticism is easier than proof because proof requires addressing all possibilities. A single counter-example or error can be a decisive criticism without considering vast numbers of possibilities. Popper discussed this with examples like "all ravens are black", which is refutable with one contradictory observation, but still not proven with a million compatible observations.
All our arguments are fallible and can be reconsidered with new ideas and evidence. But if you accept an observation of a white raven and some background knowledge, that contradicts "all ravens are black". There's no comparable way to prove it, even fallibly.
You can make decisive positive arguments using universal premises (e.g., "All men are mortal.") which assert their own completeness. But that just moves the problem: how do you prove that premise since observing a million mortal men is inadequate?
Indecisive negative arguments have the same basic flaw as indecisive positive arguments: you can accept the argument but still reject the argument's conclusion without contradicting yourself. There's no logical connection between the argument and the conclusion. Logically, indecisive arguments don't do anything. Contradiction is a powerful, useful logical tool used with decisive arguments, but we don't have good alternative tools to use with indecisive arguments. Compatibility (non-contradiction) lacks logical power despite sometimes misleadingly being called "support" or "confirmation".
If I observe a purple hamster or red tree, that is compatible with "all ravens are black"; compatibility is basically worthless without also using another concept like relevance. But even with a million relevant, compatible observations – e.g., observations of black ravens – a white raven could still exist. Relevance is very difficult to define and evaluate, but even if we had perfect knowledge of what was relevant, compatibility plus relevance would still be inadequate to make indecisive arguments effective. Nothing about seeing a lot of black ravens actually means that there couldn't be a white one.
If indecisive arguments are logically flawed, then why do they often appear to work moderately well? Many of them are convertible to decisive arguments. They have a valid point which is presented imperfectly. Similarly, many positive arguments are convertible to negative arguments. I hypothesize that arguments which can't be converted to decisive, negative form are wrong. Skipping the conversion step is often reasonable in low-stakes, low-precision, friendly contexts.
How do you convert arguments? Positive arguments point out good trait(s). To convert to negative form, criticize alternatives for lacking the good trait(s). Indecisive arguments involve more creativity to convert. They point out good or bad things without logically connecting that to success or failure at a purpose. To make them decisive, clarify the purpose, figure out criteria for success and failure, and point out how the bad things (or missing good things for alternatives) cause failure.
Focusing on negative arguments can take some getting used to because it's more indirect. Instead of arguing in favor of an idea, you criticize alternatives to that idea. You also try to criticize the idea. The conclusion you reach is the idea you can't find any decisive error in, despite trying.
If you have multiple ideas that you think will work, you can use any of them, or you can aim for a more ambitious purpose/goal. CF has methods for narrowing it down to exactly one non-refuted idea, but that's often not worth the effort. If you have no ideas you think will work, you can brainstorm/research more, adjust your purpose/goal to be easier, or give up and do something else.
If you don't find an error with an idea, it could still be wrong. Why use it? Because ideas you don't know are wrong are preferable to ideas you do know are wrong (already found a decisive error in). It never makes sense to use an idea for a purpose if your best understanding is that it will fail at that purpose.
Why?
What is the motivation for CF? A reasonable first impression would be that CF doesn't sound strictly wrong (you don't see a reason it can't possibly work), but it sounds inconvenient or cumbersome. So what's the upside to make it worth pursuing?
CF's main motivation is logical arguments showing that various other approaches cannot possibly work. CF prioritizes correctness over everything else. (I actually think CF is reasonably convenient and elegant once you get used to it, and has various merits besides correctness, but I acknowledge there's an initial learning curve.)
Two main alternatives to CF are induction and weighted factors. Serious problems with these are actually well known in the academic literature and have been written about by advocates of these approaches, not just by opponents like Popper. I'm not actually saying anything very new by claiming these approaches are flawed. I think people try to use them despite the known errors because they've largely given up on finding alternatives, and errors seem somewhat ignorable to people accustomed to indecisive thinking. Some people are ignorant of the problems with their approaches, but I think many experts are instead pessimistic about finding something better. Ignorance of problems reduces motivation to consider alternatives. And experts who understand the problems may not want to spend time studying or debating a new system that they find initially counterintuitive and aren't optimistic about.
The various schools of inductive thought can be seen as attempts to say that indecisive arguments sometimes partially work. Compensatory weighted factor approaches (where a high score at one factor can compensate for low scores at other factors) also try to use multiple indecisive arguments to reach a conclusion. Non-compensatory multi-factor approaches are less common but exist and have some overlap with CF.
I've written criticisms of induction and weighted factors before. For this article, I'll just say that I'm open to discussion if someone disagrees but wants to try to reach a conclusion about the matter. In a discussion, I only have to address one person's concerns instead of all potential concerns, so it can be easier and more focused (and provide valuable feedback and potentially criticism for me, too). I can provide a free account on my forum if someone accepts this invitation (email me).
Examples
The idea "go to McDonald's for a burger" succeeds at the purpose "get lunch" but fails at the purposes "get a gluten-free lunch" and "learn knitting". We should evaluate {idea, purpose} pairs rather than ideas alone. Being more precise, we can include context and evaluate a triple: {idea, purpose, context}.
Note: In other articles I've said "goal" instead of "purpose" and abbreviated the triple to IGC, but they mean the same thing. It's {idea/solution/answer/plan/explanation/argument/reasoning/option/alternative, purpose/goal/question/problem/objective/criterion, context/problem-situation/background-knowledge/scenario} triples. The triple is meant to work with any of the terms separated by slashes; you can use whichever is most natural. Sometimes answer and question work well, but in other contexts solution and problem make more sense. {idea, purpose/goal, context} is particularly generic and works well when you want to keep terminology consistent. Plurals like "goals" or "criteria" may be used because the goal is often a group: the goal could consist of multiple conjoined sub-goals or be to meet multiple criteria. Interestingly, goals and contexts are types of ideas which can be put in the first spot of their own IGC triples. Similarly, arguments are used to evaluate IGCs but they are also part of their own IGCs: arguments have goals and contexts and an argument's IGC can itself be criticized.
A criticism can apply to multiple {idea, goal} pairs. If my goal is to get a gluten-free lunch, the criticism "buns have gluten" applies to McDonald's and also to other ideas like Burger King and Wendy's. The same criticism can also work for other goals, e.g., getting a gluten free breakfast or dinner. It's often important to think using principles and make your criticisms broad enough to cover whole categories of ideas and goals instead of just an individual idea-goal pair.
An indecisive positive argument is "I'll go to McDonald's because their food tastes good". It's indecisive because the reasoning (McDonald's food tastes good) doesn't contradict "I won't go to McDonald's". Both can be true. Also, alternatives like "I should go to Burger King" aren't contradicted even though they imply not going to McDonald's (in a context where you're picking one restaurant for a specific meal). Indecisive means there's no logical problem with accepting the reasoning but also accepting a contradictory conclusion.
To convert to a negative argument, figure out what positive trait is being praised (food taste) and criticize alternatives for lacking it. E.g., "I won't go to Burger King because their food tastes bad." After converting, always check if the result is true. If you actually like the taste of Burger King's food, then it's wrong. In that case, the original indecisive, positive argument was also wrong to conclude that you should go to McDonald's specifically (not Burger King) because of taste.
To convert to a decisive argument, you often need to clarify your purpose/goal. Let's say I'm hungry now and I want to get restaurant food within five minutes of walking. Then an indecisive argument is "I'll go to McDonald's because it's close." A decisive argument is "I won't go to Burger King because it's over a five-minute walk away." More generically, "I won't go to any restaurant that's over a five-minute walk away." can rule out the idea of going to Burger King, Wendy's and more with a single criticism. To make location decisive, we figure out what proximity constitutes success or failure and we criticize ideas that fail.
Next Steps
If CF interests you, a good place to start is by categorizing arguments you make, read or hear as decisive or indecisive and positive or negative. You can do that without changing what arguments you use or like; it can start as a research project for informational purposes. A later step could be trying to convert some arguments to be decisive and negative and paying attention to how some convert well but others don't.
A more advanced issue is considering whether enough decisive, negative arguments exist (including via conversion) to do all of our reasoning with them, or if we need to use some non-convertible positive and/or indecisive arguments. Many people like decisive, negative arguments and see them as highly powerful and useful, but believe they're too scarce to use exclusively. A related issue to ponder is: Can you ask a series of yes or no questions to explore any issue, or are other types of questions (or non-questions) absolutely necessary? Yes or no questions are good at exposing decisive issues. "Does it cost under $50?" invites a decisive answer while "How expensive is it?" or "How happy am I with the price?" don't.