Introduction to Critical Rationalism

Something has a purpose if it serves a goal or an objective or solves a problem. (These are different ways of saying the same thing)

If something has a purpose, it has a reason for being like it is. It has some sort of function.

Like a way to decide whats an error and what’s not. Evolution requires replication with variation and selection. If you just have one or the other then evolution wont happen. So evolution requires selection criteria (and different things to select between i.e variation).

I think that Elliot is just deciding to be approximate because a precise answer is perhaps complicated in a way that is not useful for the point of the article. That’s my guess.

1 Like

Yes, the approximate answer is easy to understand and useful. I see two main ways to be precise. You can give lots of details (e.g. the contents of multiple Dawkins books, and you could still keep studying a lot more after that). Or you can make a short, abstract statement and try to get it logically perfect, which would be hard and less practically useful for most readers than thinking about grandchildren.

BTW maybe I should have specified having more grandchildren. More is better.

EDIT: I added “(more is better)” to the article.

1 Like

Ah i think i see what you’re saying. Idk if i should say some examples to check my understanding

You can say the same thing with different words?
Good to know

Coming back to this statement:

Genes have a purpose. They serve a goal

Oh ok is selection criteria another word for error correction?

I think I see what youre saying. Like being more precise requires more words. I dont think a single sentence would be enough to go into the details about the criterion in nature.

1 Like

it’s not synonymous no but they’re related.

error correction is the process or act of finding and fixing errors.

selection criteria are rules, standards, or principles for deciding what is or isn’t an error. They are criteria for selecting. So selection criteria kinda means goals. If you’re trying to achieve a goal, then that goal is the selection criteria. It is the standard by which you judge your attempts to achieve it.

Ahhh those two are different. Hard for me to visualize examples of each one, but I think ill know the difference if the two come up

Critical Discussion and Justificationism pre-ideas

Critical Discussion

Finding errors in ideas and correcting them is how evolution improves the ideas, rather than increasing the justification for the ideas. Thus a critical mindset and critical discussion is important for the growth of knowledge.

CR says that the ideas that best survive criticism are the ones we ought to believe in (would CR not talk about belief here?). Ideally after refuting a bunch of ideas we are left with only one idea that has survived all the criticism, and that’s the idea we should believe in (is this a CF original idea?). CR says that we can’t prove ideas correct by tests, but they can be corroborated by tests if they pass the tests and thus we can have greater confidence in the idea (would CR talk about confidence in ideas?).

We can also view some solo thinking as self-discussion. We might have two contradicting ideas in which we don’t know which to accept and which to reject. In such situations we ought to approach it much like a critical discussion with another person. We ought to not be biased for one side and we ought to look for errors in both sides.

Critical discussion is often viewed as adversarial where each tries their best not to lose and tries to defend their side as much as possible. But since criticism is the beginning of the growth of knowledge we ought to generally think of receiving criticism as a gift. If I’m wrong in a debate it is I that learned something, while my discussion partner merely had to be my teacher. In reality it is the person who was wrong who actually “won”, it is he gained the most. But if you were right from the start that doesn’t mean you wasted your time either. Critical discussion doesn’t have to end in agreement in order to be productive. By exposing your ideas to criticism you can improve them and/or improve your arguments for them.

Review

I focused on discussion while the article spoke a lot about criticism. I found it necessary to say a bit about criticism in the beginning to connect, but I intentionally limited it. I think it’s generally better to just write more and not worry about things being covered in other sections. The purpose here is partly to expose what I already know and believe and look for contradictions with the article and errors generally. So if I write more than the article then that doesn’t really matter. I still benefitted from writing it as well.

So confidence in ideas would be wrong according to CR, it’s about preferring ideas that better survived criticism.

The rest of what I said wasn’t relevant. I actually a bit puzzled why this stuff would come so early in the article, but it was all I could think of on discussion. I was also puzzled because I knew Popper didn’t say that much about specifics on how to do critical discussion.

My focus wasn’t the same so I don’t think I can say much about what I missed. Unlucky.

Justificationism

Justificationism is the idea that you can improve ideas by providing positive supporting arguments. It says that knowledge is justified true belief. By giving an idea supporting arguments our belief that the idea is true becomes more and more justified.

Evidence is usually highly regarded and preferred over arguments by justificationists. According to them when our ideas fit the evidence our justification for the ideas increases. Even if we accepted that evidence can justify ideas we could still see there’s an logical asymmetry here. You only need one decisive criticism to deem an idea false, whereas a single instance of evidence isn’t usually regarded by justificationists as fully confirming an idea.

However CR rejects that any supporting argument or evidence can actually provide any justification. CR says we can only use negative arguments to refute ideas and use the surviving ones.

If we claim “all ravens are black” then justificationism says that observing black ravens increases the justification for this statement. But we also know that if we observed a single nonblack raven then the statement would be false. No matter how many black ravens we have observed, the possibility of a nonblack raven still exists. We can never prove that a nonblack raven doesn’t exist.

The best justificationism could do then is to say that observing black ravens increases the probability that all ravens are black. An interesting consequence here would be that observations of non-ravens would also increase the probability. Since all observing a black raven does is to confirm that this material is not a nonblack raven. A black raven doesn’t actually say anything about nonblack ravens. One could argue that observations of black ravens provide more justification since ravens are a subset of all other material. You wouldn’t really know what the probability change was anyway since you have no way of knowing with certainty how many ravens exist in the universe (there could exist ravens on other planets!).

Actually I don’t know why CR says the probability doesn’t increase. If I assume there exist a finite number of ravens then I could use an easier example to illustrate the same point. Instead I say I draw a ball from a bag with X amount of balls and observe that it’s black, if we assume uniform probability for black vs nonblack on every other ball (X + 1 different configurations being equally likely) then we can calculate the probability for any X, and the probability would increase for each black ball we drew out. Does CR say the principle of indifference is wrong, can I not extrapolate from the balls in the bag to ravens in the universe?

I asked Claude AI: “let’s say I take out a ball from a bag of which know there are 100 balls and the ball is black. what is the probability that rest of the balls are also black”. I didn’t about the principle of indifference before this.

Review

I didn’t say anything about “true” or “belief”.

I talked more about justifying beliefs/ideas rather amount of goodness. I think they’re almost the same?

I think I don’t the variety in justificationist thinking. So I assumed they conceded some things that I thought were true like decisive criticism and that at least most them think in terms of probable and high credence rather absolute certain true knowledge. Which is ironic given I’ve been talking with @actually_thinking who believes in absolute certain true knowledge.

Critical Discussion and Justificationism outline

Critical Discussion

  • search for flaws in ideas in order to improve them
    • we need to error correct instead of looking for support for ideas
    • we cannot prove an idea is not wrong
    • there can always exist a mistake that we are not aware of
  • criticism works logically because they can contradict something
    • logical asymmetry between criticism and supporting arguments
      • if the criticism is correct then the criticized idea must be incorrect
        • it’s impossible to make supporting arguments that if they are true then the idea they support must also be true, there’s no logical connection, other than compatibility, i.e. non-contradiction
          • there’s nothing you can do to guarantee that there doesn’t exist a flaw you aren’t aware of
            • it’s like asking to guarantee that god doesn’t exist
            • it’s asking to prove the non-existence of something you don’t have contradictory arguments against
      • supporting arguments cannot rule out flaws in ideas, but criticism can rule out ideas as true
        • this means criticisms are decisive while supporting arguments aren’t
      • adding more supporting arguments doesn’t help against criticism or lessen the likelihood for there being flaws in the idea
  • thus the point of discussion should be find and fix errors
  • CR says to judge ideas by how well they survive criticism
    • the more ideas are tested the better
    • critical preference: preferring ideas that are more vigorously criticized and have survived
    • CF disagrees and says we should judge ideas as refuted or non-refuted
      • we should only act on ideas that haven’t been decisively refuted
      • CF’s approach is binary whereas CR’s approach is analog
  • CR does’t give much details on how to do critical discussion but CF has invented new methods

Justificationism

  • the idea that says positive arguments can say how good an idea is
    • can use supporting arguments, supporting evidence, proofs etc.
    • single ideas aren’t enough but multiple can add up to a justified belief
    • judge ideas by amount of goodness
  • justified true belief
    • true is infallibilist and perfectionist
      • anything later found out to contain an error wasn’t knowledge at all
        • but in reality lots of ideas are useful even though we later found errors in them
    • belief deny books and genes have knowledge in them
  • JTB has lots of known problems but they think the stuff CR criticizes is fine
  • some are fallibilists and look for probable ideas or justified high confidence in ideas
  • evidence/observation doesn’t speak for itself. evidence has to be interpreted to be used
  • many accept decisive criticism in addition to positive arguments
    • ideas refuted by evidence
    • ideas with internal logical contradiction
    • then ideas they say can’t be ruled out with decisive criticism has to weighed with their pros and cons
  • none of this works because establishing ideas as good or true is impossible
    • an ideas can be false no matter how many good traits it has
    • there are infinite ideas with the same positive traits but reach different conclusions
      • I think this point is relevant because the positive arguments doesn’t differentiate the idea from the other ideas with the same same positive traits
    • criticism contradicts an idea
      • contradiction is a meaningful relationship whereas positive arguments only say there’s a failure to contradict
        • non-contradiction is not support

And there are infinitely many other ideas, in the set of all logically possible ideas, that have all of those good traits but reach wildly different conclusions.

At first I didn’t understand why this was relevant. Is it relevant because the supported idea isn’t differentiated from all the other ideas that share the good traits, and they have different conclusions so the supported conclusion isn’t actually supported?

This applies to non-refuted as well, right? But there the assumption is that better ideas exist but the one non-refuted idea we have is the best we have, and so we accept it for that and nothing more like guaranteed true.

From my pre-idea writing:

Actually I don’t know why CR says the probability doesn’t increase. If I assume there exist a finite number of ravens then I could use an easier example to illustrate the same point. Instead I say I draw a ball from a bag with X amount of balls and observe that it’s black, if we assume uniform probability for black vs nonblack on every other ball (X + 1 different configurations being equally likely) then we can calculate the probability for any X, and the probability would increase for each black ball we drew out. Does CR say the principle of indifference is wrong, can I not extrapolate from the balls in the bag to ravens in the universe?

I couldn’t figure out after reading the section.


I think this all took like 4 hours.

1 Like

A key issue is whether you can easily find/construct some of the many unknown ideas.

For any data set, you can easily construct many ideas that perfectly match it. You can design them to have various conclusions. So these infinitely many theoretical ideas should be treated as part of the discussion, and as conclusions we might actually reach, because we have easy access to them on demand.

For non-refuted ideas, it’s often hard to come up with better or equally good alternative ideas.

I think I understand.

For pattern matching the criteria is just to have ideas that fit the data. This criteria is easy to meet, there are infinitely many ideas that fit.

But for non-refuted ideas the criteria is to pass all the criticism that the non-refuted idea passed. This is a hard criteria, we cannot easily create new ones.

You could add irrelevant details to the non-refuted idea to create infinitely more ideas, but then you could critique the ideas for including irrelevant parts, and so they wouldn’t actually be non-refuted ideas.

The same doesn’t work for pattern matching. The simplest pattern isn’t guaranteed to be the right one.

1 Like

There can be an unlimited number of new ravens born in the future. Also we don’t know how many ravens there are now and could discover more. Also we could discover evidence about ravens from the past. Also the number of ravens that have lived so far, while finite, is very vast. Part of the balls in bags scenario is that no new balls may be created/added and also the total number of balls is small enough that taking a few out matters. If there were trillions of balls, taking some out by hand would be a tiny sample of the balls.

Another issue is how the bag was populated with balls and how the ravens were (and will be) created. Depending on how you think the bag was populated, pulling out a black ball could increase, decrease or make no difference to your probability estimate for the next ball being black.

Fallibilism and Induction Preconceptions

Fallibilism

Fallibilism is the idea that we can always make mistakes. It says that there are no guarantees against making mistakes. No matter what you do there could always be a mistake that you’re not aware of.

Infallibilism attempts to prove ideas as certainly true. It’s problem is that whatever arguments is given in order to prove that the idea is true must also be proved true themselves. And so you have to make sure ad infinitum that every supporting argument in the proofs don’t contain any mistakes. This is an infinite regress which means we can never get to the bottom of it.

Infallibilist propose axioms as a way to escape the infinite regress. Axioms are fundamental ideas that we can be certain are true without giving proof that they are. Many of ideas held as axioms are knowledge, but there really is no way we can be 100% certain they are true. Any axiom which you use reasoning to establish must have the reasoning be proved true as well, so those cannot escape the infinite regress. The other option is to claim that the axioms are simply self-evident. They say we don’t need any reasoning we just know they are true because of their simpleness, obviousness or something else. But ideas once held as self-evident have later been found to be wrong. There isn’t anything that’s truly obvious; anything can be doubted. Claiming self-evidence is arbitrary and is never a rational reason to accept any idea.

If we view truth as correspondence with reality then theoretically we can express the truth perfectly (however there could be issues with language or other thinking tools in getting perfect correspondence). Fallibilism doesn’t necessarily say this is impossible. However it says that if we did manage to match reality perfectly then we can’t be certain that we did. Any idea that claims that we know for certain that it’s true suffers from infinite regress. Still we can hold the true idea.

Fallibilism says we can’t have certainty, but it doesn’t say we can’t have knowledge. According to JTB, fallibilism would amount to denying knowledge, but according to evolutionary epistemology certainty isn’t necessary for knowledge. CR says you have to start with guesses. It then says that we can improve upon the guesses by criticizing the guesses and then coming up with new guesses that don’t have the same flaw. The possibility of the improvement of our ideas is what let’s us say we have knowledge. For each time we reject ideas and come up with alternatives that are better we have come closer to the truth. It is also always possible that what thought was an improvement was actually a regression.

We can also view attempted refuted ideas as true and useful in some limited context. Refuted ideas is still information which is not random. They will have a definite structure and an appearance of design for some purpose. And so we can think of them as less useful and/or true knowledge. Because if something was designed then that means knowledge had to be used in order to create it. Either that could be a watchmaker using his knowledge to create watches, or it could be nature using the process of evolution to create genes, which contain knowledge, to create the appearance of design in animals.

Fallibilists are often accused as being skeptics because the accuser thinks that we need certain ideas for knowledge or at the least that we need some certain ideas to base the less certain ideas on. But what is the danger of being a skeptic? Pyrrho, a skeptic, was said to have to be pushed out of the way from a carriage since he would not move himself because he could not be certain that the carriage was real. Whether or not the story was true it shows that the problem with skeptics is that they’re indecisive. With certainty you would be absolutely decisive, so infallibilists fear that without certainty you would be like Pyrrho not knowing what to do. But so long as we have a method to figure out what our best idea that we know is, then we can be decisive and follow that idea. With CR we can use critical preference and choose the ideas that have best survived criticism. With CF we can refute ideas and come up with new selection criteria until we are left with one non-refuted idea to follow.

Fallibilism can also mean that mistakes are common. Which they in fact do seem to be. Given that we improve our knowledge by fixing mistakes, that mistakes are common and that we can never be certain there is no mistake in our knowledge, then we really ought to always be on the look out for mistakes.

I wanted to say:

Because we can never be certain that we are correct then there is always room for infinite progress

But that doesn’t follow. I think the infinite progress idea has to do with fallibilism, but I can’t really connect them. Here’s one way:
Since we can never perfectly define our all of our knowledge without making circular definitions, we can always add definitions and thus improve our body of knowledge as a whole.
Hmm. Just adding definitions doesn’t necessarily improve our knowledge. We could have already added all the useful definitions.

Review

I can’t find any contradictions.

I think the biggest thing I missed out on was not talking more about being open to criticism. I only said we should always be on the look out for mistakes.

Induction

Induction is the (false) idea that you can extrapolate patterns from observations to create theories and/or validate them. Induction is a justificationist theory which says we can justify theories by making repeated observations.

Induction relies on the premise that the future will resemble the past. It says that patterns we observe now and have observed will continue into the future. But only some patterns will continue into the future while others won’t. For induction to work we would need some epistemological technique to know which patterns are going to continue and which won’t. We can make (fallible) explanations for why certain patterns will continue or won’t, but induction can’t use those since induction is supposed create such explanations. It would be a stolen-concept. Induction has no way of telling which patterns will continue and which won’t.

For any sequence of observations we have observed so far there logically exists infinitely many patterns that fit these observations. The sequence of observations may repeat again, but it could just as well be followed by any different observation afterwards. There are infinite logical possibilities for what observation will follow. If the next observation breaks what looks like our previous pattern then a new pattern has actually emerged, just repeat the sequence we have seen up until now. If we look at the observations we have now and imagine the logically possible patterns they could follow we can see that there are an infinite amount patterns that logically fits.

I have a concern with the above. Shouldn’t we only count what is logically possible as things following non-contradiction and the laws of physics. So when I say there are infinite amount of logically possible observations that could follow any sequence of observations, then that’s false because I’m counting observations which would not follow the laws of physics and non-contradiction? Maybe there aren’t infinite amount of possible observations that could follow, but if there multiple logically possible observations then that could make infinitely many different patterns. You can make infinite patterns with 1’s and 0’s so long as there doesn’t stop being a possibility for either observation after some point of time.

Induction also faces the problem of determining how well a pattern fits a situation. When you’re making observations how does induction tell you that the situations are similar enough for the pattern to hold? For example you may find that water boils at different temperature at different altitudes. That can be explained by the different atmospheric pressure at different altitudes, but induction can’t tell you that. Over time the world changes, and sometimes relevant changes are made to the situation for the theory. Induction says to make multiple observations for similar enough situations in order to confirm the theory. But induction cannot tell you when the situations are similar enough and how well the pattern fits the situation.

Induction assumes that observations can be understood directly. It says pure impressions of observations can generate ideas and verify them. It says that truth is manifest, just repeatedly observe nature and the truth will occur to you. But in reality what you get from observations is just sense data. And the data doesn’t make sense unless you interpret it. The idea you get from making observations is what you interpret it to be. There is no mechanism for sense data to automatically generate true ideas in your mind.

Induction also assumes that observation takes the leading role, guiding the mind towards true theories. It says that after making repeated observations the pattern underlying the phenomena becomes apparent to you. But we have already explained that there are an infinite amount of possible patterns that are compatible with our observations. So, how does the method of induction pick out the correct pattern among them? It can’t. In order to discover the correct pattern you have to already have it in mind, otherwise you cannot recognize it in your observations. When we have a pattern or theory in mind we can observe to see if they are compatible with nature. So it is the mind which guides the observations towards theories and patterns such that we can look for ways nature might contradict our ideas. You need to have a theory in mind in order to focus your attention towards the correct observations.

Review

Induction is an attempt to use evidence without arguments or intelligent thoughts.

That’s a good sentence. I said induction couldn’t use explanation since that would be a stolen-concept and that induction is supposed to explain how explanations are made. But the quote above said it more explicitly.

I didn’t say anything about how evidence can only provide contradiction or non-contradiction.

I didn’t criticize induction for not trying to explain and understand the world conceptually.

Fallibilism and Induction Outline

Fallibilism

  • we’re always capable of making mistakes
    • we can never get 100% certain proof
      • there can always be a mistake that we missed
      • progress can still be made
  • errors are common
    • there are logical arguments for errors always being possible but it’s not just a theoretical issue that only sometimes comes up in practical reality
  • justificationism tries to fight against fallibilism by trying to prove ideas as true or let us know errors are unlikely in an idea
    • CR embraces fallibilism and says we can still learn by error correction
  • we are never done thinking and solving new problems
    • big mistakes can be made but can also fix those mistakes later
  • to deal with fallibility we ought to seek criticism as much as possible
    • we should take criticism from sources without authority as well
    • ignoring critics prevents error correction and is thus the best way to stay wrong
    • have a critical mindset and suspect even very popular ideas

Induction

  • an attempt to use evidence without arguments or intelligent thought
  • evolution explains knowledge creation without presupposing intelligence
    • replication, variation and selection don’t need intelligence. they can be used by a non-intelligent process
  • about finding patterns in data and extrapolate from them
    • not about error correction or understanding the world conceptually
    • tries to give justification to theories through evidence
    • also induction doesn’t explain how to find patterns without intelligent thought
  • logical problems
    • there are infinitely many compatible patterns for finite data sets
      • induction can’t explain which to to induce without using intelligence
    • “the future resembles the past”, but which patterns will continue? induction has no answer
      • it’s inevitable that some patterns continue while other stops, but that information isn’t very useful without knowing which will continue and which won’t
    • it assumes that correlation probably means causation, which it actually doesn’t
      • again which correlations will continue to hold?
    • how could we know what a “pattern” is and what counts as “similar” without intelligent thought?
    • any piece of evidence is compatible with infinitely many ideas
      • the evidence doesn’t support any idea more than another; all it can do is confirm non-contradiction, which is the same for all compatible ideas
        • there’s no guidance from induction on which ideas to prefer among the non-contradicting ones
  • criticism from Popper on probability in induction and epistemology

I spent somewhere around 3 to 3.5 hours on this.

1 Like

Depends what you were certain of.

Logically possible means following the laws of logic, which includes not being self-contradictory, but it doesn’t include following the laws of physics. If you just say “possible” then it often means possible according to our current understanding of both logic and physics.

2 Likes
Interpreting Observations and Decisive Criticism preconceptions

Interpreting Observations

Some epistemologies rely on observations giving us knowledge directly. According to CR observations are just sense data and can’t tell us anything by itself. Ideas are replicators and new ideas get created by replicating other ideas but with variation. A new idea does not get created independent of other ideas by observations.

To get knowledge out of observations we have to guess at what the observation is. The observation is just data and we have to interpret it. Without interpretation it would just look like arbitrary information, but with interpretation we can get knowledge. With interpretation we can see the structure of the data.

We also miss out on many things which are included in the sense data we get. Sometimes we experience that an object we were looking for was right in front of us in our sight. This happens because we have to focus our attention. And where to focus our attention is informed by our ideas. So in order to detect something we have to first have an idea to look for it. This means our theories comes before our observations.

For example any inventor will have had an idea of what it was he was going to invent. So he had to have an idea of the invention before he could observe the invention.

When we interpret an observation what we get is an idea. Our interpretation of the observation is fallible like all other ideas, and therefore we can criticize the interpretation. Observations don’t offer us infallible knowledge. The sense data itself can’t be wrong and it doesn’t necessarily get corrupted by our interpretation. The sense data are objective and real physical objects, but they are the result of both the object being observed and our measurement instruments, whether that is a telescope or just our eyes. Knowledge of the object and our measurement instruments are relevant in interpreting the observation.

There is no method by which we can stop thinking and still gain knowledge. There is no automatic method of acquiring knowledge. Gaining knowledge requires focus and brainstorming (it happen subconsciously as well).

Review

Elliot said our sensory organs may be wrong, I said the sense data itself can’t be wrong. I said though that the data is also dependent on whatever instruments we use to gather it. So knowledge of the limitations of the instruments are relevant.

For example we should know that our eyes can only detect a certain range of wavelengths as colors. We could say the eyes are wrong since there are “colors” there in reality which we don’t see, but if we align our expectations within the limits of the eyes then I wouldn’t call it wrong.

We also know that eyes get their data through the light emitted and/or reflected by other objects. If something happens to the light in between the object and our eyes then that is correctly reflected in our sense data. With only common sense we can forget the characteristics of our sensory organs and think we are getting direct impressions of objects, and thus think we are being fooled by the sense data, however it is our insufficient knowledge that makes us have the wrong interpretation of an observation.

What about impaired sensory organs? They are less precise and more limited compared to normal functioning. They give less information than normal. And so they’re wrong compared to normal. I think the genuine way for sense data to get corrupted by the sense organs is if they make random alterations to the incoming data.

Decisive Criticism

CR says to prefer ideas which have best survived criticism. But what does it mean to “best survive criticism”? It would have to be by considering all the criticism for each idea and weighting each criticism by importance and then adding them together to get overall measure for how much the idea is criticized. This runs into the problem of determining how important each criticism is and putting into actual numbers. There’s really no way to do this given that ideas can deal in different dimensions so we can’t get a single survival of criticism score.

But CR says criticisms can refute theories, so would Popper talk about the weight of the criticism? Maybe it was about corroboration, i.e., passing tests, and then weighting the severity of the tests. So more difficult tests gives theories more corroboration and then we prefer the theories which are most corroborated? I guess that was what Popper said. Because it wouldn’t be positive corroboration vs negative criticism since the positive side would be justificationism. Popper said we could refute theories, so out of the non-refuted ones we prefer the ones with the most corroboration.

An alternative to a single survival score is to give ideas multiple pass/fail evaluations based on different contexts and goals. This limits the amount of criticism you have to deal with since you only need to consider criticisms which makes the idea fail for a specific goal in a specific context. And neither do we have to give the criticisms a numerical representation, since we’re only focused on whether the criticism stops the idea in succeeding for the goal in consideration, i.e., we’re only focused on decisive criticisms. If we evaluate ideas in this way we can always prefer ideas which are non-refuted instead of best survived. If all our ideas are refuted we can either think of new ideas or change our goals.

CR has a focus on refutation and it rejects positive supporting arguments. CF takes the focus on refutation further by rejecting weighted epistemology. CF is the natural progression of CR and is in a sense more “Popperian” than CR itself.

Review

So Popper did believe in strong and weak criticism, which is not just what I called corroboration. I guess I see decisive criticism as more natural and “obvious” so I thought Popper was closer to it.

I didn’t write one for “logical concerns” because I didn’t know where that one would go. I had already written some of what I thought to write and I thought it would be hard and take a long time.

Interpreting Observations, Decisive Criticism and Logical Concerns outline

Interpreting Observations

  • we have to interpret data
    • the interpretations are fallible
      • we may not be able to make sense of what we’re seeing, i.e. can’t think of a good interpretation
      • our sensory organs or instruments may have flaws and make errors
  • if our sensory organs were perfect we could still not automatically trust our interpretation since it depends on our other ideas
  • our ideas have to guide our observations
    • we need to know what to observe
      • and what to look after
    • there is too much raw data so we have to focus our attention and our focus is determined by our ideas
  • we should take note of when our expectations don’t match our observations

Decisive Criticism

  • contradiction and non-contradiction instead of positive support and counter-weight negative criticism
    • focus on negative critical arguments and error correction
  • CF says to evaluate as either refuted or non-refuted
    • a refuted idea has an idea we accept that contradicts it
      • a non-refuted idea is where we don’t know of such an idea
    • all correct criticism contradicts the critiqued idea, and is therefore decisive
      • therefore there can’t be stronger or weaker criticism. criticism is binary instead of being in degrees

Logical Concerns

  • if we believe two ideas to contradict they could both still be true due to mistake in evaluating them as contradicting or in our theory about logic
    • there could always be errors like this so how isn’t it paralyzing? by only concerning oneself with errors we actually have thought of instead of all theoretically possible errors
  • you can make bad faith ad hoc auxiliary arguments to rescue otherwise refuted ideas
    • we can criticize these by saying they’re arbitrary for example
    • on-topic criticism can’t handle much of such ad hoc arguments, but we can make general philosophical criticisms instead
      • most arbitrary ideas fit patterns of bad thinking which we can criticize on philosophical ground
        • for those that don’t fit we shouldn’t automatically assume it was arbitrary and rather figure out a new pattern which can be used in the future too, or we accept it as a legitimate idea

All physical objects that do computation, including sense organs, are capable of random errors even if they’re properly designed and not damaged. E.g. they can be struck by a cosmic ray which can flip a bit so the output comes out wrong. Atoms and electrons are always jiggling around and can go off course. Even if you add a lot of error correction to the design to account for this, it’ll never be literally 100% reliable.

3 Likes