[Previous] Me on Critical Preferences | Home | [Next] Using False Theories

Critical Preferences and Strong Arguments

This post is a followup. For context, click here to read the first post.

The following is intended as a statement of my position but does not attempt to argue for it in detail.

The concept of a critical preference makes the common sense assumption that there are strong and weak arguments, or in other words that arguments or ideas can be evaluated on a continuum based on their merit.

The merit of an idea is often metaphorically stated in terms of its weight (e.g. Popper wrote "weighty though inconclusive arguments", Objective Knowledge p 41). It's also commonly stated in terms of probability or likeliness. And it's also stated in terms of ranking or scoring ideas to see which is best.

Ideas do have merit, and they can be closer or further from the truth (more or less truthlike, if you prefer). However, we never know how much merit an idea has. We can't evaluate ideas that way.

(BTW suppose we could evaluate how much merit ideas have. A second assumption is that doing so would be useful and that it would make sense to prefer the idea with more merit. That should not be assumed uncritically.)

Popper did not give detailed arguments for the idea that we can or should evaluate arguments by their strength or amount of merit. That's why I call it an assumption. I think he uncritically took it for granted without discussion, as have most (all?) other philosophers.

In the strength based approach, an idea could score a 1, or a 2, or a 20. In Popper's view, the numbers don't have an absolute meaning; they can only be compared with the scores of other ideas. Or in other words, we never know how close to the truth we've come on an absolute scale. In this approach, an idea can have infinitely many different evaluations.

In my approach, an idea can only have three possible evaluations. An idea can be unproblematic (non-refuted), problematic (refuted), or we're unsure. Ignoring the possibility of not taking a stance, which isn't very important, an idea gets a boolean evaluation: it's either OK or not OK.

If we see a problem with an idea, then it's no good, it's refuted. We should never accept, or act on, ideas we know are flawed. Or in other words, if we know about an error it's irrational to continue with the error anyway.

On the other hand, if we have two ideas and we can see no problem with either, then we can have no reason to prefer one over the other. This way of assessing ideas does not allow for the middle ground of "weighty though inconclusive arguments".

If an idea is flawed, it may have a close variant which is unproblematic. Whenever we refute an idea, we should look for variants of the idea which have not been refuted. There may be good parts which can be rescued.

My approach is in significant agreement with Popper's epistemology because it does not allow for the possibility of ideas having support. Some people would say we can differentiate non-refuted ideas by how much support each has, but I follow Popper in denying that.

Popper's alternative to support is criticism. I accept the critical approach. Where I differ is in not allowing an idea to be both criticized and non-refuted. I don't think it makes sense to simultaneously accept a criticism of an idea, and accept the idea. We should make up our mind (keeping open the possibility of changing our mind at any time), or say we aren't sure.

As I see it, a criticism either points out a flaw in an idea or it doesn't. And we either have a criticism of the criticism, or we don't. A criticism can't contradict a theory and be itself non-refuted, but also fail to be decisive. On what grounds would it fail to be decisive, given we see no flaw in it?

Let's now consider the situation where we have conflicting non-refuted ideas, which is the problem that critical preferences try to solve. How should we approach such a conflict? We can make progress by criticizing ideas. But it may take us a while to think of a criticism, and we may need to carry on with life in the meantime. In that case, the critical preferences approach attempts to compare the non-refuted ideas, evaluate their merit, and act on the best one.

My approach to solving this problem is to declare the conflict (temporarily) undecided (pending a new idea or criticism) and then to ask the question, "Given the situation, including that conflict being undecided, what should be done?" Answering this new question does not depend on resolving the conflict, so it gets us unstuck.

When approaching this new question we may get stuck again on some other conflict of ideas. Being stuck is always temporary, but temporary can be a long time, so again we'll need to do something about it. What we can do is repeat the same method as before: declare that conflict undecided and consider what to do given that the undecided conflicts are undecided.

A special case of this method is discussed here. It discusses avoiding coercion. Coercion is an active conflict between ideas within one mind with relevance to a choice being made now. But the method can be applied in the general case of any conflict between ideas.

My approach accepts what we do not know, and seeks a good explanation of how to proceed given our situation. It is always possible to find such an explanation. It may sound difficult, but actually you already do it dozens of times per day without realizing it. Just like people must use conjectures and refutations to understand each other in English conversations (and must use them in all their thinking), and when they first hear that idea it sounds bizarre, but they already do it quickly, reliably, and without realizing what they are doing.

Elliot Temple on March 4, 2010

Comments (13)

Verisimilitude

"Popper did not give detailed arguments for the idea that we can or should evaluate arguments by their strength or amount of merit. That's why I call it an assumption. I think he uncritically took it for granted without discussion, as have most (all?) other philosophers."

I think he did discuss it in detail. In Chap 10 of CR, for example, where he explains his idea of verisimilitude. Note, however, that Popper points out that verisimilitude "is not a probability in the sense of the calculus of probability", so he would not agree with the common idea that theories should be evaluated in terms of probability or likeliness.

"Ideas do have merit, and they can be closer or further from the truth (more or less truthlike, if you prefer). However, we never know how much merit an idea has. We can't evaluate ideas that way."

Popper would say that although you can never know for sure that one idea has a higher degree of verisimilitude than another idea you can nevertheless guess and examine your guess critically.

"I don't think it makes sense to simultaneously accept a criticism of an idea, and accept the idea."

What about the situation where all the ideas proposed so far are known to have flaws?

Brian Scurfield at 3:03 PM on March 8, 2010 | #1975

Re: Verisimilitude

> I think he did discuss it in detail. In Chap 10 of CR, for example, where he explains his idea of verisimilitude.

He discusses verisimilitude in general, e.g. by offering criteria for greater verisimilitude. But I don't see that he anticipated and addressed my criticism there or anywhere else. Do you think he does? Page?

BTW I reread C&R ch10 before posting this -- which is a nice chapter -- and assorted other Popper chapers before the original post.

> Popper would say that although you can never know for sure that one idea has a higher degree of verisimilitude than another idea you can nevertheless guess and examine your guess critically.

Yes, and I'm saying that isn't a sensible way to approach conflicts between ideas. Basically because it doesn't do anything about the conflict. My approach does.

(Of course when we guess and examine critically we may end up creating new criticism and resolving the conflict. Then it's easy to proceed. But the problem at issue is what to do before that new resolving criticism gets invented.)

> What about the situation where all the ideas proposed so far are known to have flaws?

Then the solution to the question must be deemed undecided or unknown, and we can ask, "Given our situation, which is that we don't have a correct solution, but we do have these various ideas and some knowledge of their flaws, what should we do?"

This question can lead to one idea about what to do, which is not in conflict with any of our ideas, and has no known flaws.

Elliot at 9:30 PM on March 8, 2010 | #1978

Re: Verisimilitude

On page 293 of C&R, Popper writes that although science is not subject to the operation of anything resembling a law of historical progress, "we can speak clearly and sensibly about making progress there". So, he takes as his first thesis that "Within the field of science we have ... a <i>criterion of progress</i>: even before a theory has ever undergone an empirical test we may be able to say whether, provided it passes certain specified tests, it would be an improvement on other theories with which we are acquainted." I think he has identified and discussed the assumption that we <i>can</i> evaluate theories by their relative strength (which he says is objective and to do with explanatory power, as you note, but not to do with probability or plausibility or subjective weight) and hasn't taken it onboard uncritically.

I can't see that he discussed in any detail whether we <i>should</i> want to do this given we have an unresolved conflict between ideas, so I think you are right on that. And I agree that, in general, weighting ideas is not something we should want to do in this situation. Rather, as you say, our approach should be to seek a good explanation for what to do now. We are here answering a different problem situation to the original and this helps us become unstuck, as you point out. Do you think, though, that if verisimilitude is objective, then it can sometimes (often?) supply useful information to help us?

Brian Scurfield at 5:49 AM on March 9, 2010 | #1979
> Do you think, though, that if verisimilitude is objective, then it can sometimes (often?) supply useful information to help us?

No, not without modification.

Popper offers criteria of verisimilitude and basically says they can, in certain circumstances, be used to support theories over other theories. I think they are good criteria, but only for critical use.

Consider the situation where T1 and T2 are both non-refuted (uncriticized) and we claim T1 is better in any way. Can we convert this into a criticism of T2? Point out that it fails to meet some criteria? If we can, just do that. Why leave T2 uncriticized if we know a criticism? On the other hand, if all our arguments that T1 is superior do not contain any criticism of T2 at all, then I think they must be mistaken.

Elliot at 8:44 AM on March 9, 2010 | #1980

Re: Verisimilitude

http://www2.warwick.ac.uk/fac/soc/philosophy/people/associates/miller/foreword2009.pdf

You may be interested in this passage:

"By the time of the second edition of Objective Knowledge (1979), it had become known that Popper’s own account of verisimilitude, and most others, are seriously defective. It is a theorem that if T2 possesses more truth content than T1 , and is false, then it possesses also more falsity content. It follows that if we prefer theories with greater verisimilitude in Popper’s sense, then preferences among false theories are never possible. Popper acknowledges all this, and other logical di?culties, in (3)–(5) of the newly added Appendix 2, but looks forward hopefully to the time when an adequate theory of verisimilitude will be at our disposal. (This time, in my judgement, has not yet arrived.) Assuming that happy state, he goes on to make a determined e?ort to explain more clearly in what sense the empirical record sometimes gives us ‘reasons’ for preferring one theory to another. At my prompting he added an explanatory remark at the end of item (4): ‘Whenever I say . . . that we have reasons for believing that we have made progress . . . I do (not) claim . . . that T2 is, in fact, nearer to the truth than T1 . Rather, I give an appraisal of the state of the discussion of these theories, in the light of which T2 appears to be preferable to T1 , from the point of view of aiming at the truth’ (p. 372). Whether this statement contains another of those ‘whi?s of induction’ that Popper’s critics so often manage to detect in his writings, I leave for the reader to judge. But the truth is that our theoretical preferences are quite as conjectural as our theories are; and there is nothing wrong with this, provided that we can, as suggested in the previous paragraph, test these preferences against the facts, and sometimes overthrow them."

If I have understood you correctly, there will never be a good theory of verisimilitude. The reason is that the problem of forming critical preferences is a justificationist mistake, so the whole problem situation is misbegotten. Correct?

Brian Scurfield at 10:33 AM on March 10, 2010 | #1981
@quote: I think seeking a *highly formal* theory of critical preferences is a mistake. But what about an informal theory of critical preferences which says we should prefer theories which are simpler, better testable, clearer, contain better explanations, are harder to vary, are less arbitrary, make fewer assumptions, and so on? The formal criticism doesn't address that. I think Miller is accepting the informal theory of critical preferences when he says "our theoretical preferences are quite as conjectural". I think he's saying we have critical preferences, and they may not have formal justification, but they're still good in the usual fallibilist way.

Rafe accepts the informal theory too. He summarized some informal criteria of critical preferences (and of course one may always add or subtract from this list by critical discussion, it's not set in stone):

http://www.criticalrationalism.net/2010/03/08/comments-on-mises-and-gordon-on-popper/#comments

> To repeat, the criteria include explanatory power (or theoretical problem-solving power), capacity to unify different fields, capacity to stand up to tests (not just evidence but logic, consistency with other well-tested theories), practical or instrumental value and the capacity to stimulate good research.

I think these informal attempts are better than the highly formal one, but fail. My original post was aimed at them, not at the formal version.

> The reason is that the problem of forming critical preferences is a justificationist mistake

I'm not sure if it's a justificationist mistake or something else. I'm leaning towards something else.

> so the whole problem situation is misbegotten. Correct?

No. I see nothing wrong with the problem being addressed. My approach solves it.

The problem, in its best formulation, is: Consider a problem and some solutions to it which have been subjected to criticism to your tentative satisfaction. Count the non-refuted solutions. If the number is one, how to proceed is very easy, do that one. If the number is not exactly one, then it's not so clear how to proceed, there is a problem of what to do.

Critical preferences tries to solve the problem by taking theories that are all non-refuted (possibly also theories that are all refuted if the non-refuted count is zero) and differentiating between them in some way other than by considering whether they are refuted. I think that doesn't work and the refuted or non-refuted criterion is the only correct way to differentiate theories. This approach, prima facie, is not justificationist because the criteria for differentiating suggested are things other than quantity of justification.

I solve the same problem by doing stuff to arrive at a situation of having exactly one non-refuted theory.

Elliot at 11:28 AM on March 10, 2010 | #1982
> The problem, in its best formulation, is: Consider a problem and some solutions to it which have been subjected to criticism to your tentative satisfaction. Count the non-refuted solutions. If the number is one, how to proceed is very easy, do that one. If the number is not exactly one, then it's not so clear how to proceed, there is a problem of what to do.

This, though, isn't about forming a preference, it is about finding a non-refuted explanation of what to do. Preferences to me seem to imply some sort of weighting process on a continuum and this is exactly what you are not doing. Whatever you call it, however, I think your approach is correct.

Brian Scurfield at 12:35 PM on March 10, 2010 | #1983
I think there was a misunderstanding in my comment just above: Your approach seeks to solve a problem, the same problem that critical preferences attempted to solve. But you do it without using the idea of a critical preference. That idea is wrong. So you are well aware that your approach is not about finding critical preferences, it's about doing something else.

If the mistake of critical preferences is not a justificationist mistake, what kind of mistake do you think it might be?

Brian Scurfield at 2:09 PM on March 10, 2010 | #1984
Consider "weighty though inconclusive arguments" or critical preferences or other versions.

It doesn't actually make sense.

My guess is Popper didn't question it. So there's no justificationism behind it, or anything else. Just not noticing it was an important assumption that needed critical discussion.

What happens if we try to make sense of it? One interpretation is that it says to weigh the amount of justification of theories. I think you're noticing that possible attempt to make sense of it when you accuse it of justificationism. If justificationism was true, then that interpretation would make critical preferences a good theory. But justificationism is not true, so unless Popper advocated that way of making sense of it, I see no reason to blame it on him. I'm sure there are other possible attempts of making sense of it that aren't justificationist, though I haven't thought of an example that seems plausible.

If there is justificationism here, perhaps it's just b/c there is justificationism underlying so much of philosophy that if you accept any random philosophical theory it may well have justificationism lurking in it somewhere.

Looking at the weighty remark, there is a somewhat different mistake, which is the idea of criticized but non-refuted theories, which is nonsense that everyone takes for granted. People think arguments can survive being criticized unless the criticism is decisive, and most aren't decisive. This lets them ignore lots of criticism at their whim. The main issue here is: you can't accept a two theories that conflict at the same time! (One is a criticism of the other is a special case.) But people think you can. Perhaps that is the underlying mistake for critical preferences; it seems related. Is this other mistake justificationist? Well, it's not too hard to make a justificationist themed version of it, or to make a proof that it's actually true using justificationist ideas as premises, but I think there exist other roads to the same mistake including just accepting it for no particular reason.

Elliot at 5:05 PM on March 10, 2010 | #1985
Yes, people do take it for granted that you can accept two theories that conflict and it's not necessarily a justificationist mistake.

Even if your alternative to critical preferences turns out to incorrect - and I can't see that it is - it is right that critical preferences be put under the microscope and I hope you get more feedback than what you have appeared to get so far!

Some other questions: Although critical preferences is a false theory, what about the notion of "closer to the truth"? I guess you would say an unrefuted theory is [tentatively] closer to the truth than a refuted theory, though, perhaps, we can't formally quantify this? Say we have two refuted theories: can we say which was closer to the truth; for example, that Newton was closer to the truth than Kepler? I guess this can only be done within the context of our currently accepted unrefuted theory, namely Einstein? Correct?

Brian Scurfield at 12:38 PM on March 12, 2010 | #1986
I have no objection to the idea of things being closer to the truth, objectively. No criticism of it.

As for judging theories in that way, or trying to measure or approximate the closeness to the truth, there are two issues. The first is how? What counts as closer to the truth? I have no reason to think this question can't be answered, but I don't see that it is answered.

The second issue is: why? Who cares how close to the truth an idea is? What is the point of knowing that?

Also note: I would only expect to be able to judge closeness to the truth comparatively between ideas, not on an absolute scale.

Elliot Temple at 1:01 PM on March 12, 2010 | #1987
Yes, I would expect to be able to judge it only comparatively. If your current, one-and-only, non-refuted, theory could be judged absolutely as being at distance d from the truth, then the theory you used to measure d must contain knowledge about the truth that your current theory does not contain and could, I guess, be used to refute your current theory.

Who cares about closeness to the truth? Apart from the people who care about critical preferences? Popper was interested in what sense, if at all, we can speak of making progress in science. That seems to require some notion of closeness to the truth. In a practical sense you are correct: What we should care about is the one unrefuted theory that it is always possible to obtain in a given situation. You have no way of knowing how close to the truth it is and who cares?

Brian Scurfield at 5:02 PM on March 12, 2010 | #1988
We make progress by solving our problems. We can tell we've made progress when our lives have improved (including increased intellectual satisfaction), and when we know useful stuff we didn't used to know. This is of course a fallible metric -- we could be deluded -- but it's open to criticism so it's just as good as our other knowledge.

Elliot Temple at 5:21 PM on March 12, 2010 | #1989

What do you think?

(This is a free speech zone!)