Episode 138: “Popperian” vs “Deutschian” Epistemology

• Listen to this episode: Apple Podcasts

Unofficial, reviewed transcript. It may still contain mistakes; check the podcast when wording matters.

Speakers are labeled by first name where identified. Unidentified-speaker labels may refer to different people. Bracketed question marks indicate uncertain wording.

Transcript

[00:00:00]  Unidentified speaker: Hello, out there.

[00:00:02]  Peter: This week on the Theory of Anything podcast, Bruce discusses various differences he perceives between modern CritRats, so Critical Rationalists on Twitter, and more old school Critical Rationalist epistemology. Bruce is, as always, passionate about this, but I believe he is respectful to those he has disagreements with. I hope someone out there also gets something out of this.

[00:00:36]  Bruce: Welcome to The Theory of Anything podcast.

[00:00:38]  Unidentified speaker: Hey, Peter.

[00:00:39]  Unidentified speaker: Hey, Bruce.

[00:00:40]  Unidentified speaker: How are you doing today?

[00:00:41]  Unidentified speaker: Good.

[00:00:42]  Bruce: You're ready for an epic long, longest episode ever of the podcast?

[00:00:48]  Unidentified speaker: Honest answer, I hope.

[00:00:53]  Bruce: So I couldn't figure how to split this episode up.

[00:00:55]  Unidentified speaker: So it's huge.

[00:00:56]  Unidentified speaker: Okay.

[00:00:57]  Unidentified speaker: I'm gonna do my best with it.

[00:00:59]  Bruce: It's something that I've been thinking about for a while that I wanna capture kind of altogether in one episode.

[00:01:04]  Unidentified speaker: All right.

[00:01:05]  Bruce: So now I've argued that CritRats today have departed from Popper's originally intended epistemology. And I've blamed Popper for that. I feel his wording has misled rather than led. Now I know that's a rather bold, risky sort of claim, right? In addition, Popper's original epistemology had numerous real problems that later Critical Rationalists really did need to solve. So not surprisingly, we don't all agree on how to solve these problems. Now I've heard through the grapevine that I offended another Critical Rationalist podcaster who felt that I was saying something like, you're not a real Critical Rationalist. Now I admit my view is likely to offend. But let me try my best to situate it in such a way that I think it doesn't lose its intended punch, which is that I think that the popular version of Popper is wrong in some ways. Because we don't want to lose the intended punch. But I think I can situate it so that it comes across less offensive as to what I'm really trying to get at here. I've talked on this show about the children of Popper and how Critical Rationalism isn't really a single epistemological view, but a family of views, each developed by various students of Popper. David Miller probably represents the view closest to Popper's original view. Though I feel Miller helped start the CritRat war on words, that has proved problematic. Lakatos doesn't get a lot of love, but his work was a needed corrective, though he worded his corrective in a way that I can see why Popper felt straw-manned. Taleb is one of the most famous important Critical Rationalists today.

[00:02:52]  Bruce: He's also a really big jerk. He adapted Critical Rationalism, claiming survival of the idea doesn't matter as much as survival as those that hold the idea. So, for example, if the Jews had various supernatural ideas and it caused them to cause them as a people to survive and thrive, then that idea worked regardless of whether it was true or not. So this is an idea that may or may not be correct, but it deserves serious attention. Danny Frederick created an instrumentalist version of Critical Rationalism. That almost sounds like a contradiction, but that is what he created. Now, I think instrumentalism is a completely wrong path, but his argument is absolutely worth taking seriously, and then carefully refuting. It's not easy to carefully refute his ideas. Plus, he was a critical part of developing my own views, and he helped me make sense of where some of Deutsch's views had gone astray compared to Popper's original epistemology, which is one of the things when I would talk with Danny Frederick online, he would argue that that was the case. There are so many others. I'm not even mentioning Agassi, Bartley comes to mind. Then that doesn't even include Popper-adjacent people that weren't students of his, like Jonathan Rauch, who we had on the show, one of the truest Popperians I've ever read. Though he claims he gets his ideas more from Charles Sanders Peirce, one of Popper's main influences rather than Popper, or at least for that particular book, The Constitution of Knowledge, that was true.

[00:04:23]  Bruce: Which might make sense, since when we'd refer to Popper's epistemology, the real truth is that what we call Popper's epistemology predates Popper by centuries as embodied in the scientific method. So it's technically incorrect to call it Popper's epistemology. Though I think that is not a bad label for it, since Popper looms large over it, and is the first person to try to put it down explicitly what it was. And of course, David Deutsch, he's done more to popularize Popper modernly than anyone else. He's added to it, and I'll argue here, argue he's also subtracted from Popper in important ways that have directly influenced the development of Critical Rationalism. He was my introduction to Karl Popper. And this is just a few examples. There are many more children of Popper that took Critical Rationalism and developed it in various directions. It would be very strange indeed if only one of them got everything right and no one else had something important to add. It would be very strange indeed if they didn't all make mistakes that needed to be corrected. So the goal in my mind should be to synthesize all their ideas into a single improved theory. At a minimum, the various children of Popper were all intentionally trying to error correct Popper. So all of them attempted to change, modify, or improve upon Popper in some way because they felt Popper either got something wrong or didn't explore some area well enough. But none of them agreed on what the best approach was. So let's use Lakatos as our example.

[00:05:57]  Unidentified speaker: I don't even know if I'm pronouncing that right by the way.

[00:05:59]  Bruce: Everyone else hates him. Let's say there is some fan of Lakatos and you want to explain why you feel Lakatos got some important ideas wrong. The very fact that you are claiming Lakatos was wrong on something, that does in some sense, imply that you think that person's epistemology is mistaken, at least somewhat. Frankly, maybe more than just a bit. If that person got upset and said, you're saying I'm not a real Critical Rationalist, what would be an appropriate response? You surely are arguing they're mistaken in their epistemology, at least to a degree, aren't you? What I want to do is explore what I've sometimes called CritRat epistemology by which I mean something like the current most popular Deutschian interpretation of Popper's epistemology, at least among laymen, particularly on Twitter. CritRat epistemology is one version of Critical Rationalism that makes some important adjustments to Popper's original epistemology, often for very good reasons, by the way. It also reinterprets Popper's epistemology in various ways. As with what I've called the pseudo-Deutsch theory of knowledge, it's often difficult to separate what Deutsch intended and what his superfans interpret him as saying. For our purposes, let's make some distinctions, just so that I've got some words that can be shorthand for what I'm talking about. I'll call Deutsch's specific interpretation that he himself says as Deutschianism. And in contrast, we'll call what Popper originally said as Popperianism.

[00:07:34]  Unidentified speaker: Okay?

[00:07:35]  Bruce: And the CritRat's specific interpretation, often but not always coming from Deutsch, we're going to call CritRatism.

[00:07:42]  Unidentified speaker: Okay?

[00:07:43]  Bruce: Now, I mean these only as shorthand. So Popperianism is just, for the sake of this podcast only, by the way, a shorthand for Popper's original interpretation of critical rationalism. And Deutschianism is just, and only for the sake of this podcast, a shorthand for David Deutsch's adjusted version of Karl Popper's epistemology. And CritRatism is just a shorthand for the popular critical rationalist interpretation of critical rationalism today out on Twitter, largely inspired by David Deutsch. Each of those is a mouthful, so we need shorthand labels for them. I'm going to claim that CritRat epistemology makes a number of epistemological mistakes, due in part to some mistakes on David Deutsch's part. But I respect that David Deutsch also gets a lot right, often correctly identifying problems, sometimes big problems, with Popperianism, which now means just Popper's original epistemology. Just as an obvious example of this, Popper was entirely wrong about quantum mechanics, and Deutsch has proved entirely right about it. And Deutsch arrived at the conclusion precisely because Deutsch understood Popper's epistemology better than Karl Popper did, at least on some important points. So let's first take a look in-depth at Deutschianism, what Deutsch has specifically said to adjust Karl Popper's epistemology, and compare it to Popper's original epistemology, Popperianism for short. Then we'll circle back and we will talk about CritRat epistemology and its relationship to all of this.

[00:09:18]  Bruce: Okay, so let's talk about the differences between Deutsch and Popper on various issues and kind of just take Deutsch at his word for how he differs from Popper, okay? So Deutsch isn't a big fan of Popper's concept of verisimilitude. On Brett Hall's podcast, episode 100, this exchange takes place. Brett, turning to another word, a word that you in fact don't use, and it brings me to chapter two, which is called Closer to Reality. But Popper had this term verisimilitude, which means something like closer to truth. And it seems to me that you've deliberately avoided that word.

[00:09:55]  Unidentified speaker: But why?

[00:09:56]  Bruce: Is it because it contains a misconception? It's misleading? Or it's a neologism? What would be the reason to avoid such a word? David says, I think all of those things. Deutsch then goes on, he first admits that Popper's idea about getting closer to the truth was probably initially helpful. But then he says, but on the other hand, when you do introduce truth and especially verisimilitude, then the idea is that somehow we can utter truth or at least we can utter something that's 90% true or something like that, Brett, and would have thereby a method of measuring that 90%, David. I think verisimilitude was not intended to be something that you could measure. But if you can't measure it, and then it's, measure it, then its only use is a sort of philosophical regulating principle. And I think knowledge and problems are much better concepts. So, Deutsch believes, though he doesn't explain in detail, that concentrating on knowledge and problems are better replacement concepts than the idea of getting closer to the truth. Deutsch says this is why chapter 2 of Beginning of Infinity is titled Closer to Reality rather than Closer to the Truth. However, Deutsch is incorrect that Popper didn't originally intend verisimilitude as something you could measure. Popper spends quite a bit of time trying to measure it and makes attempts to do just that in, say, his book Conjecture Refutation. His famous initial attempt involved trying to count up the size of the set of true and false assertions a theory makes.

[00:11:29]  Bruce: I think Deutsch would likely argue here that Popper never intended this more than as a thought experiment. However, David Miller, working with Popper, put a lot of effort into actually trying to measure verisimilitude and failed, ultimately culminating in Miller producing a proof that it was impossible to measure verisimilitude. But Popper did, at one point at least, think you could measure verisimilitude. After Popper failed, and only then did he start to treat it as a mere regulating, philosophical regulating principle. Also note the specific objection Deutsch raises, that if you can't measure verisimilitude, and surely as of today, you can't, then what use is it other than as a mere philosophical regulating principle? Which, I agree with Deutsch, isn't all that particularly useful. So there really is a real problem here that Deutsch has tried to solve. However, it is understandable why Popper tried to hold on to his verisimilitude, why he was stubborn about it, as it tied into other aspects of his epistemology in a way that wasn't easy to excise. The main problems was Popper's stance that theories are a hundred times more valuable than concepts, as we discussed in episodes 126 and 127. Popper's whole argument was that theories have truth content, while concepts do not, and that what we should care about is what truth assertions our theories make, not what concepts we use to explain it. So Popper argued, quote, this is Objective Knowledge, page 123, What should interest us are theories, truth, argument.

[00:13:14]  Bruce: If so many philosophers and scientists still think that concepts and conceptual systems and problems of their meaning or the meaning of words are comparable in importance to theories and theoretical systems and problems of their truth or the truth of statements, then they are still suffering from Plato's Main Errors, i.e.

[00:13:34]  Unidentified speaker: essentialism.

[00:13:35]  Bruce: He goes on, this is page 310 now, Always keep to assertions, to theories, and the question of truth. I hope people can see why there is no easy way to make this same argument utilizing Deutsch's interpretation. Verisimilitude is deeply wired into Popperianism. Note also that this view that theories are about finding truth, or at least getting closer to it, is deeply related to Popper's view that we shouldn't argue over what words and even concepts we use to explain theory. Thus, Popper continues, Objective Knowledge, page 310, So one should never quarrel about words and never get involved in questions of terminology. One should always keep away from discussing concepts. What we are really interested in are real problems, are factual problems, or in other words, problems of theories and their truth. This is Popper's view that questions of meaning of words are empty, words are just instruments for theories in his view. And instead, we should just hold to truth assertions. Note that this fits perfectly with Popper's view of testability and corroboration. A theory's value is directly related to how far you can test it and how far you have seriously tested it so far. Absent the idea that our theories can get closer to the truth, this whole view starts to crumble. This is why Popper felt strongly about the idea of getting closer to the truth, or in other words, verisimilitude, and saw it as central to his epistemology and stubbornly refused to give up on it.

[00:15:08]  Bruce: This is also why Deutsch, for what seems like pretty good reasons, discarded this idea and tried to replace it with getting closer to reality and replacement concepts like knowledge and problems, as he just said.

[00:15:22]  Unidentified speaker: Next is Popper's demarcation criteria.

[00:15:28]  Bruce: For convenience, I'm going to misuse the term criteria as both singular and plural.

[00:15:32]  Unidentified speaker: Being a layman, I get to do things like this.

[00:15:34]  Bruce: I realize I really should be saying Popper's demarcation criterion.

[00:15:37]  Unidentified speaker: I just don't like wording it that way.

[00:15:40]  Bruce: Deutsch does seem to accept Popper's demarcation criteria, but he understands it differently than Popper did. Here is Deutsch's interpretation from his Sam Harris interview. Popper's criterion of demarcation is not intended to be either sharp or pejorative or a criterion of meaning or worthiness, worthwhileness or anything like that. It's just a matter of convenience, a matter of convenient classification of subject matter. Both Deutsch and Popper agree that his demarcation criteria is not a criteria of meaning. A non-scientific explanation is not meaningless, like the logical positivists thought. Whether Popper would agree that the demarcation criteria is sharp, would probably depend on what you mean by sharp. Popper does take the stance that a theory can have degrees of testability. For example, given that he says, this indicates that the criterion of demarcation cannot be an absolutely sharp one, but will itself have degrees. There will be well testable theories, hardly testable theories, and non-testable theories.

[00:16:48]  Unidentified speaker: That's from Conjecture Refutation, page 346.

[00:16:51]  Bruce: Note that when Popper says that the demarcation criterion is not sharp, he explicitly clarifies that this means theories have degrees of testability. This is probably why Popper often talks pretty straightforwardly, as if we can generally lump theories into two buckets, scientific or non-scientific. Thus, the above quote continues, still on page 346, Those which are non-testable are of no interest to empirical scientists. They may be described as metaphysical. Notice that he treats them as fairly sharp buckets. So to Popper, the sharpness or lack thereof is more a matter of how testable a theory was. But Popper still seems to have believed, or I can't find anywhere so far where he contradicts this, that theories are either testable or they are not, which is why he always talks of theories as either being scientific or metaphysical. See episode 132, Roughly Testable Theories, where I disagree with Popper on this and challenge this idea. In that episode, I had an ANCAP friend that was trying to argue to me that economics should be considered metaphysical on the ground that you can't test it empirically. I really wish that I could have responded to that friend with a really nice Popper quote that said something like this, The demarcation criteria isn't meant to be sharp like that. There are actually theories, including economics, that don't fall into either bucket easily. But I was unable to find such a quote.

[00:18:23]  Bruce: The one I just made above where Popper says demarcation isn't sharp, really just contextually means the theory in the empirical bucket might not have very much testable content. Recall also that Hayek specifically claims economics is squarely in the scientific side of the demarcation line, even though it's very difficult to test empirically. Likewise, Popper did not see his demarcation criteria as merely a matter of classification, as Deutsch claims, quote from Popper, from an epistemological viewpoint, the problem of demarcation is in fact the only fundamental one, says Popper, that's from Two Fundamental Problems. Thus we find Popper saying things like, quote, Irrefutability is not a virtue of a theory, as people often think, but a vice, that's from conjectural refutation. Note how this is really nothing like Deutsch's version of the demarcation criteria, being, quote, just a matter of convenience or merely for classification. Popperianism, unlike Deutschianism, did place value on being scientific and did see scientific theories as, in some sense, superior theories, or at least non-empirical theories, as having a definitive flaw. For that reason, you will never find anything in Popper where it's okay to offer a non-testable competitor to an already testable theory and then find that non-testable competitor the better theory. In a competition between testable and non-testable theory, Popperianism always sees the testable theory as the winner in the contest.

[00:20:00]  Bruce: This is barring something extreme, like the testable theory solving the wrong problem or being contradictory or something like that. [With?] Deutsch, the demarcation criteria, is solely as a classification, so he doesn't place much significance on it. To Popper, the demarcation criteria was the main jumping-off point to him coming up with his entire epistemology in the first place. Popper started out wanting to demarcate actual scientific theories from pseudo-scientific ones, like in his view, Freudianism or Communism. Quote, it was in 1919 that I first faced the problem of drawing a line of demarcation between those statements and systems of statements which could be properly described as belonging to empirical science and others which might perhaps be described as pseudo-scientific.

[00:20:50]  Unidentified speaker: That's from Conjecture Refutation, page 344.

[00:20:53]  Bruce: So his initial impetus for trying to solve this problem was to tell a good theory from a bad one. He goes on to then expand that later to include metaphysical theories and even mathematical theories, not just bad pseudo-scientific ones. But initially, he was interested in separating out pseudo-scientific theories from real scientific theories. Popper, Conjecture Refutation, page 51 says, Today I know of course that this criterion of demarcation, the criterion of testability or falsifiability or refutability, notice that he treats those all as the same thing, is far from obvious, for even now its significance is seldom realized. But I did not yet realize its full implications or its philosophical significance. Notice how this is nothing like Deutsch's view, that it is merely a classification. Notice also how Popper saw the demarcation criteria as equivalent to his understanding of refutation. Almost every time Popper speaks of conjecture and refutation, which mind you, is almost constantly understand that he is talking about the demarcation criteria in his mind. As we just quoted Popper, he saw testability and thus demarcation as central to his epistemology. But given Deutsch's dimmer view of the demarcation criteria, it should not be surprising that Deutsch is not convinced that testability is central to epistemology. Deutsch offers an alternative view based around the idea of criticism. There is a subtlety here that I need to bring out because of course, Popper also suggested that criticism is key to his epistemology. To see how the two men differ, let's read a few quotes.

[00:22:30]  Bruce: First, consider Popper's view on this. Popper can't shut up about testability and falsifiability, which mind you, he sees as two words meaning the same thing. So Popper says, every genuine test of a theory is an attempt to falsify it, or to refute it. Testability is falsifiability.

[00:22:49]  Unidentified speaker: That's from conjecture and refutation.

[00:22:51]  Bruce: However, Popper did accept that even when so-called metaphysical non-empirical theory was being discussed, you could still rationally discuss it via criticism. Quote, in those days, I wrongly identified the limits of science with those of arguability. I later changed my mind and argued that non-testable, i.e. irrefutable metaphysical theories may be rationally arguable.

[00:23:14]  Unidentified speaker: That's from Objective Knowledge, page 40.

[00:23:17]  Bruce: Now, if we look up, look upon a theory, this is still a quote from Popper, if we look upon a theory as a proposed solution to a set of problems, then the theory immediately lends itself to critical discussion, even if it's non-empirical and irrefutable. That's from Conjecture Refutation, page 269. See episode 92, Popper on Philosophical Theories for a discussion on this, as well as episode 93, Philosophical Theories versus Bad Explanations, where I show a significant problem with Popper's view on this. But this should show that Popper wasn't directly at odds with Deutsch on this. He also believed that you could criticize untestable so-called metaphysical theories. To understand, then, how Popper and Deutsch subtly but profoundly differ on this, you first have to understand that to Deutsch, science is specifically about seeking increasingly good, quote unquote, good explanations. Thus, Deutsch says, I have already remarked that even in science, most criticism does not consist of experimental testing. This is because most scientific criticism is directly is directed not at a theory's predictions, but directly at the underlying explanation.

[00:24:26]  Unidentified speaker: That's from beginning of infinity.

[00:24:28]  Bruce: So to Deutsch, an explanation is judged by its ability to explain or its explanatory power rather than to specifically, is specifically based on how well it makes predictions. Popper would not necessarily disagree with that as I just currently worded it. But Popper and Deutsch would disagree over what counts as a good explanation. To Popper, the goodness of an explanation is judged in terms of how independently testable, i.e. non-adhoc, the explanation or theory is, and how severely tested, that is to say corroborated, it is so far. Quote from Popper, ad hoc explanations are explanations which are not independently testable. Independently, that is, of the effect to be explained. They can be had for the asking and are therefore of little theoretical interest.

[00:25:18]  Unidentified speaker: That's Objective Knowledge, page 15.

[00:25:21]  Bruce: Thus Popper says, quote, The methodology of science and the history of science also becomes understandable in its details if we assume that the aim of science is to get explanatory theories which are as little ad hoc as possible. A good theory is not ad hoc while a bad theory is.

[00:25:41]  Unidentified speaker: That's Conjecture Refutation, page 81.

[00:25:43]  Bruce: Notice how Popper explicitly judges theories in terms of their independent testability. This is a consistent theme with Popper. Testability is never far from his mind.

[00:25:53]  Unidentified speaker: It is identical to the concept of falsification.

[00:25:56]  Bruce: But this way of judging theories is specific to scientific or rather empirically testable theories. Popper gives us little help on how to judge between non-empirical theories. And what help he does give, as per episode 92, isn't particularly helpful. It boils down to saying have a critical discussion about the theory. Deutsch instead judges a theory as good or bad, based on how, quote, hard to vary it is. In Deutsch's view, a good explanation is hard to vary, and a bad explanation is easy to vary. In fact, contrary to Popper's stated view, Deutsch uses this idea of hard to vary versus easy to vary to prove that testing isn't really the decisive factor in science. Deutsch argues, quote, testability cannot have been the decisive factor in the scientific revolution either. He points out that not only is the Persephone myth testable, but so is, quote, every traditional rule of thumb for making a flint blade or a campfire is testable. Every would-be prophet who claims the sun will go out next Tuesday has a testable theory. So does every gambler who has a hunch that this is my lucky night, I can feel it.

[00:27:07]  Unidentified speaker: That's from Beginning of Infinity, page 14.

[00:27:10]  Bruce: Deutsch's main example here is the story of Persephone and how it was used to explain seasonal cycles. Deutsch points out that you can easily test this theory by going to Australia where the seasons are all inverted, but he argues this does no good because it's so easy to adjust the myth on the fly to match any set of data. I wanted to check if Deutsch ever mentions Popper's no ad hoc rule or even the concept of ad hocness at all. Recall that one of Popper's main conventions is that you are never allowed to save a theory via adjusting, via adding an ad hoc auxiliary theory. I could not find any mention in Deutsch at all of this no ad hoc rule, nor could I find where Deutsch, unlike Popper, directly describes science as seeking non ad hoc explanations. However, Deutsch does sometimes refer to bad explanations as having quote ad hoc modifications or quote picked out with capricious and ad hoc magic or the like. Deutsch also takes the insight from Popper that you can still criticize a non-testable theory and he amplifies it. Deutsch argues that an empirical theory is merely one kind of criticism. Quote, even in science, almost all rejected theories are rejected for being bad explanations without ever being tested. Experimental testing is only one of many methods of criticism used in science.

[00:28:33]  Unidentified speaker: That's from Beginning of Infinity pages 119 to 120.

[00:28:37]  Bruce: Deutsch argues, I think correctly I might add, that quote, The Enlightenment has made progress by bringing those other methods to bear in non-scientific fields too. The basic reason that such progress is possible is that good explanations about philosophical issues are as hard to find as in science and criticism is correspondingly effective.

[00:28:58]  Unidentified speaker: That's from Beginning of Infinity page 120.

[00:29:01]  Bruce: CritRats have grasped on this idea that empirical tests are just one kind of criticism. It's a point that would be hard to argue with really. Surely it is the case that empirical tests are a special case of criticism. Deutsch then points out that criticism is, contrary to popular belief, is really specifically about, is not really specifically about empirical testing per se. He points out that, quote, the overwhelming majority of theories are rejected because they contain bad explanations, not because they fail experimental tests. We reject them without even bothering to test them.

[00:29:36]  Unidentified speaker: That's from Fabric of Reality, page 7.

[00:29:38]  Bruce: Here he uses this example—he uses the example of testing grass as a cure for cancer. Quote, for example, consider the theory that eating a kilogram of grass is a cure for the common cold, or I said cancer, but he actually said the cold. That theory makes experimentally testable predictions. If people tried the grass cure and found it ineffective, the theory would be proved false. But it has never been tested and probably never will be because it contains no explanation, either of how the cure would work or anything else. We rightly presume it to be false. Based on this adjusted view, Deutsch therefore emphasizes criticism over testability. So to Deutsch, the real point of science is, quote, attempting to solve problems by seeking good explanations, actively pursuing error correction by creating criticisms of both existing ideas and new proposals, beginning of Infinity, page 31. Or, quote, creative criticism interleaved with creative conjecture is how humans learn, he says.

[00:30:36]  Unidentified speaker: That's from Possible Minds.

[00:30:38]  Bruce: Or, quote, to choose an option rationally is to choose the associated explanation. Therefore, rational decision making consists not of weighing evidence, but of explaining it in the course of explaining the world. One judges arguments by explanations, not justifications, and one does it creatively using conjecture, tempered by every kind of criticism, beginning of infinity, round page 341. The key point here, Deutsch explicitly says a theory is judged based on how well it explains available evidence. And notice that he always refers to criticism, not empirical tests. Note how that differs from Popper's no ad hoc rule. Popper's no ad hoc rule insists theories must not only explain the available evidence, but also make independently testable predictions, independent of the problem you are currently trying to solve. Quote, we require that a new theory, this is from Popper, should be independently testable, apart from explaining all the explained explicanda which the new theory was designed to explain. It must have new and testable consequences, preferably consequences of a new kind. It must lead to the prediction of phenomena, which have not so far been observed. This requirement seems to me indispensable, since without it, our new theory might be ad hoc, for it is always possible to produce a theory to fit any given set of explicanda.

[00:32:05]  Unidentified speaker: Conjecture Refutation, page 327.

[00:32:09]  Bruce: Note also that Deutsch explicitly says we temper our conjectures with every kind of criticism instead of every kind of testing.

[00:32:16]  Unidentified speaker: The rest of the passage after that reads, the real source of our knowledge is conjecture alternating with criticism, not testing, beginning of infinity, that not testing was me, the rest was a Deutsch quote.

[00:32:28]  Bruce: You won't find much in Deutsch about the importance of testability, because Deutsch does accept the demarcation criteria, at least for the purposes of classification, he does accept that science is tautologically always empirically testable. But try to go look up, every time he uses the term testability, and you'll find that he will only ever says the word to argue it isn't decisive to how science works, even if it defines what gets classified as science, or to downplay it or to situate it as merely one kind of criticism. The strongest quote I can find from Deutsch in favor of testing is this one.

[00:33:06]  Unidentified speaker: Testing is the prediction—testing the predictions is just an indirect way, albeit an exceptionally powerful one when available, of testing the explanations.

[00:33:18]  Unidentified speaker: Fabric of Reality, page 66.

[00:33:20]  Bruce: But I have to admit, I at least somewhat agree with Deutsch here. Testing really is just one form of criticism, and science does use other kinds of criticism besides testing. So Deutsch does not see testability and falsification as decisive in science, but instead prefers to think testability as just a form of criticism. And since Deutsch sees the demarcation criteria as merely a classification for convenience, it isn't too surprising that Deutsch disagrees with Popper that scientific theories are in some sense superior to philosophical or metaphysical theories. Consider the following quote from Deutsch, which indirectly makes the argument that some philosophical theories are as or even more important than scientific ones. Quote, some philosophical arguments, including the argument against solipsism, are far more compelling than any scientific argument.

[00:34:10]  Unidentified speaker: Indeed, every scientific argument assumes the falsity not only of solipsism but also of other philosophical theories, including any number of variants of solipsism that might contradict the specific part might contradict specific parts of the scientific argument.

[00:34:24]  Bruce: That's from Fabric of Reality, page 85. Again, I have to confess, I find this argument pretty compelling. Solipsism is the quintessential example of a, quote, metaphysical theory and Deutsch's argument against it that I cover in episode 99, Critical Rationalism and Solipsism, are as or more compelling than most arguments we accept as good scientific arguments. Nor do I deny that he's correct that every scientific argument assumes the falsity of solipsism. Note, though, in context, Deutsch was talking about the lack of hierarchy of reliability of theories, not actually about philosophical arguments being superior to scientific theories. However, many CritRats argue that this proves that Popper was wrong, that testable scientific theories are in some sense better than metaphysical theories. Another very large difference between Popper and Deutsch is how they understand the concept of refutation. I've covered this extensively in past podcasts, starting in episodes 41, The Problems of Refutation, and episode 42, Popper Without Refutation. And then I dug into a deeper, a lot deeper, in episode 115, is falsification falsifiable. The bottom line is that Deutsch, like Lakatos before him, believes that to refute a theory requires a second theory. Quote, but in any case, the existence of a problem with a theory has little import besides, as I said, in forming of research programs. Unless both the new and the old explicanda are well explained by a rival theory. In that case, the problem becomes grounds for considering the problematic theory tentatively refuted.

[00:36:02]  Bruce: So Deutsch, so to Deutsch, you refute a theory by having the existence of a better theory. That was, oh, by the way, quote from page 10 of Logic of Experimental Tests, Deutsch's paper. Not so for Popper. To Popper, a refutation or falsification is merely having a basic statement which contradicts it. Quote, as to falsification, special rules must be introduced, which will determine under which conditions a system is to be regarded as falsified. We say that a theory is falsified only if we have accepted basic statements which contradict it. That's from Logic of Scientific Discovery, page 66. Note, I'm actually taking that quote out of context a bit, but it should work for our purposes. Nowhere does Popper ever speak of requiring a second theory to refute a theory. I've often said that I'm rather sympathetic to Deutsch's desire to redefine refutation, because Popper's version of it is so difficult to make sense of. Popper explains that he never really meant that you could actually falsify a theory. He just meant evidence and theories had a logical relationship of falsification. Consider Popper's explanation of falsification in one of his last books, so quite late in his career. Quote, falsifiability in the sense of the demarcation criterion does not mean that a falsification can in practice be carried out, or that if it is carried out, it will be unproblematic.

[00:37:24]  Bruce: Falsifiability in the sense of demarcation criterion signifies nothing more than a logical relationship between the theory in question and the class of basic statements or the class of events described by them, the potential falsifiers. This kind of falsifiability is a matter of pure logic. We now come to a second sense of falsifiable and falsifiability, which has to be distinguished very clearly from my purely logical criterion of demarcation in order to avoid gross confusion. One can raise the question whether an actual falsification is ever so compelling that one must regard the theory in question as falsified and thus as false. Popper points out that as far back as his first book, Logic of Scientific Discovery, he admitted that you could never truly definitively falsify a theory and that doing so requires special conventions that you have to choose to adopt, specifically, of course, the no ad hoc rule. Quote, continuing from where we left off, hence, to repeat, we must distinguish two meanings of the expressions falsifiable and falsifiability. One, falsifiable, definition one. As a logical tactical term, in the sense of the demarcation criterion of falsifiability, this purely logical concept, falsifiable in principle, one might say, rests on the logical relationship between the theory in question and the class of basic statements, or the potential falsifiers described by them. Falsifiable two, definition two, in the sense that the theory in question can definitively or conclusively, or demonstratively be falsified, demonstratively false.

[00:39:01]  Bruce: I have always stressed that even the theory, which is obviously falsifiable in the first sense, is never falsifiable in the second sense. For this reason, I have used the expression falsifiable as a rule only in the first, technical sense. In the second sense, I have as a rule spoken not of falsifiability, but rather of the falsification of falsification and its problems. That's from Realism and the Aim of Science, page XXI to XXII. To call this answer profoundly convoluted would be the understatement of the century. Of course, of course, when Popper says, my epistemology is about conjecture refutation, or in other words, about testability, which is the same as falsification. People are not going to think that means. I'm talking purely about the relationship between a theory and the evidence, not about actually showing theories to be actually false. I feel this is where Popper loses people, really honestly at square one, on the first step. And that's why I feel this is entirely Popper's own fault. And it's why I'm so sympathetic to Deutsch's attempt to redefine refutation to mean something a lot more reasonable. This brings us to one of the least well understood differences between Popper and Deutsch. Popper misunderstood explanations, or sorry, Popper understood explanations and theories in terms of logic and deduction. He does this so much that many CritRats often accuse him of logicism, a charge that I once agreed with, by the way.

[00:40:33]  Bruce: Deutsch, by comparison, argues that, quote, the asymmetry between refutation, tentative, and support, nonexistent in scientific methodology, is better understood in this way, by regarding the theories as explanations, than through Popper's own argument about the logic of predictions appealing to what has been called the arrow of modus ponens. Scientific theories are only approximately modeled as logical propositions, but they are precisely explanations. That's Deutsch in logic of experimental tests. You asked about what is logicism.

[00:41:04]  Unidentified speaker: Honestly, I have no idea.

[00:41:06]  Peter: I got it on ChatGPT.

[00:41:08]  Bruce: I was going to say that it was a term of derision that was thrown around about a problem that was perceived with Popper amongst CritRats. Early in my CritRat career, I accepted it as meaning Popper was way too into logic, and you really don't need to try to make all explanations logical statements. He's just overdoing it when he tries to get into logic. He's off base. He's worried too much about the logic of it. That's just not what scientific explanations actually are.

[00:41:39]  Unidentified speaker: That's how I understood it at the time.

[00:41:40]  Unidentified speaker: What does it really mean?

[00:41:42]  Peter: Well, according to ChatGPT, it says, is a philosophy of mathematics that claims all mathematical truths can ultimately be reduced to or derived from pure logic. In other words, math isn't a separate kind of knowledge. It's just very sophisticated logic.

[00:42:03]  Unidentified speaker: [Well, good to me?]

[00:42:04]  Unidentified speaker: I think I actually agree with that statement.

[00:42:07]  Unidentified speaker: I guess I believe in logicism.

[00:42:09]  Peter: I guess you're a logicist.

[00:42:12]  Unidentified speaker: Yes, I'm a logicist.

[00:42:13]  Unidentified speaker: All right.

[00:42:15]  Unidentified speaker: So let me admit that I think Deutsch is correct, that science is about explanations and that scientific explanations are only approximately modeled as logical propositions.

[00:42:26]  Bruce: So I guess I really, really strictly agree with Deutsch, what he just said here. But there is an interesting wrinkle here. Deutsch doesn't merely argue that logic isn't the correct way to model explanations. He goes on to argue it is impossible to model explanations at all. From Naval's The Deutsch Files 1, creativity and knowledge and explanation are all fundamentally impossible to define. Because once you have defined them, then you can set up a formal system in which they are then confined. And if you had the system that met that definition, then it would be confined to that and it could never produce anything outside the system. In addition, Deutsch argues that [explanationless?] science inhibits progress.

[00:43:10]  Unidentified speaker: That's from the beginning of infinity.

[00:43:11]  Bruce: So he argues, this form of [explanationless?] science is just bad philosophy disguised as science. Its effect is to suppress the philosophical debate by pretending that the issue has been settled scientifically.

[00:43:22]  Unidentified speaker: That's beginning of infinity, page 321.

[00:43:26]  Bruce: And continuing that, quote, another way in which [explanationless?] science inhibits progress is that it amplifies errors. Deutsch then gives his humorous city museum example where scientists count people going in and out of the museum and then call the difference in count either spontaneous human creation or spontaneous human destruction. So Deutsch rejects Popper's attempts to model explanations as logic, but feels there is no way to model explanations. And Deutsch argues that science that is explanationless can be dismissed on the grounds of being bad philosophy even if it makes testable predictions. This becomes an all-purpose weapon in the hands of CritRats—we will give some examples of this later today—that almost constantly dismiss science that they don't like as bad philosophy. Popper, by comparison, allows for explanations with any level of testability. A theory that vitamin C stops scurvy offers no deep explanation of the mechanism involved, yet Popper would still see it as a legitimate scientific explanation because it has empirical content that can be potentially falsified. Popper placed no limit on how deep or shallow your explanation must be to be considered legitimately scientific. So let's summarize the differences between Deutsch and Popper. To Popper, the central issue of epistemology was finding truth, and thus he saw verisimilitude as critical and the demarcation criteria as central. Deutsch rejects verisimilitude, doesn't see finding truth as critical, and accepts the demarcation criteria as merely a classification for convenience.

[00:44:58]  Bruce: Popper saw the goodness of an explanation as rooted in its independent testability, that is to say, its non-adhocness, and how severely tested and corroborated it is so far. Deutsch sees it rooted in the easy to vary versus hard to vary criteria, and sees testability as merely one specialized, though important form of criticism. Deutsch therefore makes criticism more decisive. Deutsch never mentions Popper's no-adhoc rule, and rarely mentions adhocness at all, and in fact rarely mentions testability. Popper felt that there could be meaningful metaphysical theories, but also felt lack of testability was a definitive vice. Popper often treats metaphysical theories as inferior to scientific theories and he never allows for metaphysical theories to be true competitors to an empirical or scientific theory. Deutsch takes a stance that philosophical theories can be as or even more important than scientific ones. Popper accepts any level of explanation, shallow or deep, so long as it has real empirical content that can be tested. So he does so based on modeling explanations as deductive logic. Deutsch dismisses shallow explanations as explanationless on the grounds that they are bad philosophy. He rejects any sort of attempt to model explanations. Popper saw refutation and falsification as a logical relationship between theories and evidence. Popper sees the asymmetry of support and falsification in terms of universal versus existential logical statements.

[00:46:30]  Bruce: Every time he brings up the asymmetry between verification and refutation, it's always for Popper the difference between a universal logical statement and an existential logical statement. He never separates those ideas. Deutsch sees refutation by comparison as related to having a second theory and prefers to think of the asymmetry of support and falsification in terms of theories being explanations rather than logical propositions.

[00:46:59]  Unidentified speaker: Okay.

[00:47:01]  Bruce: So that is the end of my first part here where I have now steel manned and tried to explain Deutsch's corrections to Popper and how he interprets Critical Rationalism different than Popper. I have also tried to flag my many agreements with Deutsch throughout this process. At a minimum, I think I agree with Deutsch on every single one of the points he makes, at least insofar as I think he's correctly identifying real problems with Popper's original epistemology, even if I don't necessarily agree with Deutsch on what the correct solutions are. To make this concrete, Verisimilitude really and truly failed miserably and ultimately ended up a purely philosophical regulating principle, so far at least, which, just as Deutsch points out, means it is, in current form at least, that it has no real value. I agree the idea of Verisimilitude has no real value for exactly the reasons Deutsch is giving. I think Popper's demarcation criteria is wrong. As currently formulated, I think it is, at best, certainly serves merely as a classification for convenience. Just like Deutsch is saying, Popper's way of judging the goodness of theories was useless for anything but empirical and scientific theories. Deutsch is essentially correct that realism is an underlying assumption of all scientific theories, and Popper's epistemology falls apart entirely without realism as its starting assumption. Therefore, making what appears to be a purely metaphysical theory more important and more foundational in many ways than any scientific theory.

[00:48:39]  Bruce: Popper really blew it when he chose conjecture and refutation as his terminological shorthand for his epistemology, and then tried to make refutation be specifically about a logical relationship, rather than about actually refuting theories. People are going to read that term as meaning. I would note that the list I just gave are not small agreements with Deutsch, nor small disagreements with Popper. Considering the strengths of my agreements with Deutsch, which are many, if you are not a long-term listener to my podcast, you are going to be very surprised by what I say next. Popper's original epistemology is in every way superior to Deutsch's adjusted version. I don't even see it as a close call. Popper beats Deutsch pretty much hands down, in my opinion. But I just admitted that each adjustment Deutsch makes is due to a real problem in Popper, right? So why don't I see Deutsch's version as superior? To explain this, let's start with a very simple idea that I'll state like this. Criticism is a poor replacement for testability.

[00:49:50]  Unidentified speaker: I'm going to say that one again.

[00:49:53]  Bruce: My statement is criticism is a poor replacement for testability. Popper came up with the idea of falsification while trying to solve a specific problem. Want to explain why pseudoscientific theories like communism and Freudianism were inferior to real scientific theories. Quote, my problem first perhaps took the simple form, what is wrong with Marxism, psychoanalysis and individual psychology? Why are they so different from other physical theories, from Newton's theory and especially from the theory of relativity?

[00:50:29]  Unidentified speaker: That's from conjecture refutation.

[00:50:31]  Bruce: That led Popper to notice that pseudoscientific theories can't be falsified because they have no real empirical content. This was a bit of a surprise because communism and Freudianism do a really powerful job explaining things, quoting from Popper now, I found those of my friends who were admirers of Marx, Freud and Adler were impressed by a number of points common to these theories and especially by the apparent explanatory power. These theories appeared to be able to explain practically everything that happened within the fields to which they referred.

[00:51:06]  Unidentified speaker: Conjecture refutation, page 45.

[00:51:09]  Bruce: The problem is that a theory that can explain anything, that is any data imaginable, actually explains nothing at all. It may subjectively feel like a lot of explanatory power in the sense that it offers intuitively appealing explanations about observations you see around you. And you can judge the theories you may think, you can judge the theories based on how well it, quote, explains things. It is an illusion. Popper realized that the real power of good explanations is that it constrains what is possible and only narrowly explains specific observations. While making other sorts of observations impossible if the theory holds true. It was a less intuitive, it was less intuitive, but it was, Popper realized, the real power of why scientific explanations were so much better than pseudo-scientific explanations. So Popper realized something big, a good explanation, meaning a scientific explanation rather than a pseudo-scientific explanation was good insofar as it had very specific constraints that it placed on reality. And that put the theory at risk. Popper then realized that a good explanation was testably precise because it had these constraints as to what outcomes were possible. He modeled this using deductive logic. You have a logical universal statement declaring something always true. Such a statement is in contrast to an existential statement, which simply declares one thing true or one or more things true. Popper noticed that you can't really test an existential statement. This still seems intuitively wrong to most people today when you bring it up to them.

[00:52:49]  Bruce: So to understand why most people feel like it's intuitively wrong, consider string theory. String theory predicts certain super partners, but offers no constraints on what energy levels they should appear at. So no matter how much you search for those super partners and fail to find them, you can always claim that the next larger collider will probably find them. And you can keep saying that forever. Yet most people would say that string theory is testable, quote unquote, because it makes a prediction that you can search for and that you might find. So to Popper, this wasn't really testability at all. So Popper realized something important. True scientific testability is when you can perform a test and get an outcome at the end. You knew something after the test was completed that you didn't know before. For this to happen, the test had to be a test of the constraints of the theory. String theory has currently has no constraints. Thus you can't perform what Popper saw as a true scientific test against it. Quote from Popper, every good scientific test, sorry, every good scientific theory is a prohibition.

[00:53:59]  Unidentified speaker: It forbids certain things to happen.

[00:54:01]  Bruce: The more theory forbids, the better it is. Conjecture Refutation, page 48. I think I can see why people refer to a theory like string theory as testable, loosely speaking, even though it has no constraints. Let's not get into an argument over words here. We're not trying to argue what the word testable means. We're trying to express an idea and we're using the word testable.

[00:54:24]  Unidentified speaker: Okay, that's important.

[00:54:25]  Bruce: The key takeaway is that a good scientific explanation is one that prohibits or constrains what is possible, which means logically, if you model it with logic, it is a universal statement, not an existential statement. Consider the theory, at least one unicorn exists. No scientist would consider this a good scientific theory, yet it is testable in precisely the same sense that string theory is, because it makes an existential prediction that you can verify. Since Popper saw such so-called metaphysical theories as inferior theories, he realized that what separates a good explanation from a bad one was whether or not you had to test it via verification or falsification. In Popper's mind, he therefore considered falsifiable explanations as superior. And he saw this as what demarcated scientific, aka at the time good, explanations from something useful but always defective, but from a sometimes useful but always defective verifiable explanation like unicorns exist. I have criticized Popper for trying to equate empirically testable theories to scientific ones. I actually think string theory, while not empirically testable, I think, I would agree with Popper on that, it is still a scientific theory in my mind. Albeit an early research program whose purpose is to explore the idea of string theory until we can refine it into being a true testifiable, falsifiable theory.

[00:56:01]  Bruce: But I have never doubted that the demarcation Popper found was an important one, even if it didn't really demarcate science from non-science, it did at least demarcate truly testable empirical theories from not so testable non-empirical theories. Falsifiability, Popper's term, was therefore equivalent to testability. And both were equivalent to a set of superior theories that we often call empirical theories, or maybe, wrongly, scientific theories. It isn't that an existential theory like String Theory, or for that matter, Unicorns Exist, are necessarily meaningless like the logical positivists thought. It's pretty obvious what those theories mean. And it would be an exciting scientific discovery if we found superpartners. For that matter, it would be an exciting scientific discovery if we found Unicorns. But surely, these existential, aka metaphysical, theories are just not of much theoretical interest. Quote, those theories which are not testable are of no interest to empirical scientists.

[00:57:02]  Unidentified speaker: They may be described as metaphysical.

[00:57:04]  Bruce: That was Conjecture and Refutation, page 346. Recall, this is also my main objection to anarcho-capitalism, which is equivalent to the theory, quote, theory. There exists a future state of governance where all current government services are handled by private companies and we promise it's going to be rad. It's the economic and political equivalent to belief in unicorns. What I have recounted here briefly is the history of the problem that Popper's epistemology was meant to solve. It started with the problem of determining what counts as a bad pseudoscientific explanation and then terminates in Popper's demarcation criteria or a falsification as logical basis for true testability. Or to put it concretely, a good explanation is one that could be tested, or at least this is the initial conclusion Popper came to. A non-testable explanation is not necessarily meaningless or unvaluable per se, but it just isn't of interest to science precisely because it has the defect that it can't be tested. Science is specifically about empirically testable theories because those are the ones of theoretical interest, even because they have theoretical content. Even in cases of scientific theories that can't be tested, string theory, the whole goal is to make it testable. Every scientist knows that until that happens, string theory is not to be considered a true competitor to general relativity or quantum mechanics. Now let's take all that and try to replace testability with criticism. Now you can say the words, right?

[00:58:35]  Bruce: You can say, quote, a good explanation is one that can be criticized, or quote, a bad explanation is one that can't be criticized or is hard to criticize. But those are basically meaningless statements in their current form. No matter how bad a theory is, it can always be criticized in a straightforward sense. Nothing else, you can criticize it for not being testable. Literally anything can count as a criticism, but not just anything counts as a Popperian style falsifiable test. Or theory, and theory can be more or less testable exactly in the sense that it places more or less prohibitions or constraints on reality. But what would make a theory quote more criticizeable or quote less criticizeable, other than obviously that it happens to be more or less testable? What would a quote more criticizeable theory objectively look like compared to a quote less criticizeable theory? This is why criticism is a poor replacement for testability. And this explains why I understand Deutsch's interpretation of Critical Rationalism to be an inferior interpretation to Popper's. It is not sufficient to point out that testability is one type of criticism, and then replace everywhere Popper spoke of testability with criticism, and be done. Do that and you have unsolved the main problem Popper solved. Here is the interesting thing. Deutsch is correct that a test is a kind of criticism, but it's a kind of criticism that is objective, that no one really needs to subjectively decide for themselves if it's valid or not.

[01:00:09]  Bruce: Yes, you can argue over if the test in question can be explained away by offering a non-adhoc auxiliary theory. So yes, it might be invalid in that limited sense. But even then, it remains valid in the sense that you were forced by the experiment to put in the effort to come up with an independently testable non-adhoc auxiliary theory to save your theory from refutation. Criticism has nothing equivalent. CritRats today, following Deutsch's lead rather than Popper's, will simply say, well, not every criticism lands for everyone. Try to say that with a straight face about empirical tests. Well, not every empirical test lands for everyone.

[01:00:54]  Unidentified speaker: It doesn't even make sense.

[01:00:55]  Bruce: Every empirical test has to be explained and can't be so easily dismissed. The best you can do is re-explain it via a different non-ad hoc independently testable theory. But since non-ad hoc explanations are hard to come by, this is a tall order. Whereas when we're talking about criticism, it's a very easy order. It's had for the asking. Whereas with criticism, we all know we rarely even agree on what counts as a valid criticism. What I'm suggesting here is that the moment Deutsch started to make testability simply one kind of criticism, without first noticing what makes testability a very special and superior kind of criticism, that he accidentally took the first step towards unsolving an important problem. How to tell when a theory is better than another theory without having to rely on personal subjective best judgment? Now imagine you're a CritRat that reads Deutsch's books as an introduction to Popper's epistemology. Deutsch's argument that testability is a kind of criticism is clearly correct. You've got no reason to in any way question that. But you don't notice that you just accidentally imbibed not merely that, quote, testability is a kind of criticism, but instead that testability is, quote, merely one possible kind of criticism. If you do that, something crucial just got lost. It becomes unclear to you why testability is actually the gold standard of criticism, why it plays a fundamentally different role from other kinds of criticisms and why those other criticisms simply do not matter nearly as much epistemologically.

[01:02:35]  Bruce: Later, when you finally read Karl Popper himself, you might think to yourself, oh, Karl Popper really would have benefited from reading David Deutsch. He spends all his time emphasizing testability, falsification, and demarcation as if there's something special. But Deutsch already showed that demarcation criteria is just a matter of classification. This confusion has led CritRats to reinvent what refutation means. For Popper, a refutation is synonymous with a falsification, and falsification implies for Popper an empirical test. But you'll often hear CritRats call a theory refuted, even when no empirical test is involved. In fact, it's now common for CritRats to label highly corroborated scientific theories as refuted based on nothing more than vague claims about the practitioner's supposed bad philosophy, often due to the charge of being explanationless science, psychology being a frequent target here. Refuted no longer means to a CritRat, the theory was testable and failed to test, it now means I have a personal criticism that I decided lets me dismiss this theory. Can you see how this small shift, starting with a true statement, so easily becomes the first step towards undermining Popper's actual epistemology? Directly related to this problem is Deutsch's emphasis on explanations. Again, this is not something I would ever argue is technically incorrect. Science is about finding better explanations just like Deutsch says. I even would agree that Popper's attempt to model explanation with deductive logic is at best, as Deutsch puts this, an approximation.

[01:04:10]  Bruce: Let me quote Deutsch again here. Quote, scientific theories are only approximately modeled by propositions, but they are precisely explanations. I recall when I first read this that I felt strongly that I agreed with it.

[01:04:22]  Unidentified speaker: I feel very differently today.

[01:04:24]  Bruce: Even by Deutsch's own admission, we don't really know what an explanation is. Deutsch even argues it's impossible to define what an explanation is. What good does it do to tell me precisely what a scientific theory is if you're precisely telling me it's an undefined and undefinable thing? Popper may only be approximately right. Presumably, that's exactly correct. But that is far better than being precisely correct over something without any definition at all. I can immediately start working with Popper's version. It has immediate real-life consequences that we can check against reality. It's falsifiable, in other words. Insofar as it is only an approximation, I can find where his theory is wrong and error-correct it because it was so explicit. This is just not true for Deutsch's version. Let me make this point concrete using Deutsch's own example of grass-knot curing cancer.

[01:05:21]  Unidentified speaker: I'm going to say cancer, but he actually said the common cold.

[01:05:24]  Bruce: Deutsch's argument is that we can dismiss this theory because it explains nothing, and that good explanations are judged on how well they explain things. But suppose we ran a randomized controlled trial, as a joke maybe, I don't know, and found that cancer or the common cold, cancer went into remission 50 percent more often than if you ate a pound of grass a day. CritRats today, following Deutsch, would argue that this is quote merely a correlative theory, so it quote explains nothing, and they would consider it quote refuted. At first, I agree, there might be skepticism that the test was done poorly, but suppose someone else tests it as well, then someone else, and all of them find that grass actually does cure cancer. In that case, the CritRats are clearly now incorrect. If grass passed this test, and anyone they wanted to go try the test, it would pass for them too. Even without knowing the mechanism, it would count as valid evidence that eating grass cured cancer, and further research would now be warranted. This brings me to the next major problem with Deutsch's formulation. Good explanations do not, at least initially, have to explain very much to count as good explanations. Grass cures cancer, or the common cold, is minimal but sufficient of an explanation to qualify as a scientific one, precisely because it can be tested. Now, yes, this is all silly. Almost no one would test this, precisely because there is no known reason grass might have cancer-curing or cold-curing properties.

[01:06:59]  Bruce: And on that point, I agree with Deutsch, and I'm really not disagreeing with him at all. But if someone did test it and it passed, the theory would be corroborated. Grass would be cultivated for cancer patients, and that response would be completely rational. Put differently, the criticism that grass doesn't cure cancer because there is no explanation why it would, is a reasonable criticism, a very good criticism. But it's inferior to actual testing.

[01:07:28]  Unidentified speaker: Of course it is.

[01:07:30]  Bruce: And surely it isn't equivalent to a refutation, in any case, as modern CritRats want it to be. If grass passed such a test, that criticism would be refuted, because testing is the gold standard of criticism. It is objective and anyone can perform the experiment for themselves. What about Deutsch's claim that, quote, every traditional rule of thumb for making a flint blade or a campfire is testable? Every would-be prophet who claims the sun will go out next Tuesday has a testable theory. First of all, a rule of thumb is a piece of very real knowledge, with very real verisimilitude, at least within a certain set of circumstances. It would never make sense to dismiss such a theory as refuted on the grounds that it was, quote, bad philosophy. Presumably, you'd want to test where it failed to make good predictions and then come up with a deeper, better alternative theory. In other words, you would want to test it and falsify it and then conjecture a better error-corrected version. The prophecy example, really, you may seem like the prophecy example is a better example here.

[01:08:38]  Unidentified speaker: I'm going to argue that it really isn't.

[01:08:40]  Bruce: I think Deutsch is likely hinting at the famous book When Prophecies Fail, 1956 by Leon Festinger and several other authors. In that book, a testable prophecy is made by a UFO cult, specifically that the UFO is going to come pick them all up on a specific day. And when that fails, they all become more deeply devoted to their beliefs because they just change their beliefs to match what happens. It's an almost textbook case of easy-to-variness, right? But this book stopped right before the true ending. What really happened after the book was published is that the weight of the failed prophecy went on to rip the cult apart and ended the cult entirely. Such is the power of real testability. On even a silly but testable prophecy, it cannot be error corrected in a way that say—it can be error corrected in a way that say anarcho-capitalism, or frankly, many popular existing CritRat theories cannot be. It isn't that we should start treating such a prophecy as scientific, and I actually have doubts about Popper's demarcation criteria in this regard precisely for that reason. But we should see its testability as a virtue, and anarcho-capitalism's lack of testability as a vice. Likewise, dismissing a scientific field you dislike as bad philosophy, because it, quote, doesn't explain, is not a minor problem. Every field of science involves statistics, if only because experiments rely on instruments that are inherently error-prone. And no scientific theory explains everything from top to bottom.

[01:10:14]  Bruce: So if a theory can be dismissed for, quote, not explaining, then every theory can be dismissed this way, leaving only personal preferences. And this is exactly what has happened with CritRats today, in my opinion. I have had CritRats tell me that hunger does not influence people to eat, because if it did, evolutionary psychology would be true, which they find unacceptable, and that any test showing this would be, quote, bad philosophy, because it, quote, doesn't explain, it's just correlative.

[01:10:46]  Unidentified speaker: Wait, wait, what?

[01:10:47]  Unidentified speaker: That strikes me as pretty extreme, even by the CritRat standards of some of the most outrageous statements we've heard, that hunger doesn't influence people to eat.

[01:11:01]  Unidentified speaker: Okay.

[01:11:02]  Unidentified speaker: I've literally argued this with multiple CritRats.

[01:11:06]  Unidentified speaker: Okay.

[01:11:07]  Unidentified speaker: And wait, what?

[01:11:08]  Peter: They don't see any relationship between hunger and eating.

[01:11:11]  Unidentified speaker: No.

[01:11:12]  Unidentified speaker: [That's just an X?]

[01:11:14]  Peter: We have an explanation. I'm just trying to think this through. We want to make sure we're steel manning this. The reason we eat when we're hungry, because we have an explanation that eating makes us feel better. But I just don't get that.

[01:11:35]  Unidentified speaker: I don't know.

[01:11:36]  Peter: I just want to make sure I'm understanding that.

[01:11:38]  Bruce: One of the main CritRats I argued this with extensively, I asked him whether smoking caused cancer and he said, of course.

[01:11:46]  Unidentified speaker: Okay.

[01:11:47]  Bruce: So you can do a randomly controlled trial to test for causation with hunger. I pointed this out to him. You can literally do a causation test with a randomly controlled trial with determining if—with people—if people are hungry, if that causes them to, is a causal factor in getting them to eat.

[01:12:06]  Peter: Like every other animal.

[01:12:07]  Unidentified speaker: I mean, Right.

[01:12:08]  Bruce: Like every other animal. I said, you cannot do that test for smoking causes cancer because of ethical concerns. And yet you're going to tell me that smoking causes cancer.

[01:12:18]  Unidentified speaker: You've got no problem with that one.

[01:12:21]  Bruce: And yet you're going to reject that hunger causes, or is a cause of eating. And yep, that's exactly what he was going to tell me. Now, you're asking how he could possibly hold this in his mind. And I wasn't actually going to explain this. The simple truth is, I don't know if I can do a true justice to his position, because honestly, the position is completely and absurdly ridiculous. I think the most steel man version I could give you based on what he said, would be something like this. That every single time I used the word causation, he would reread it to mean deterministically causes one to eat. And then he was trying to actually argue. Now, I made many attempts to clarify.

[01:13:07]  Unidentified speaker: That's not what I meant.

[01:13:08]  Bruce: I'd say, you've misread me. I actually mean it's one influence amongst many. And he'd say, oh, but Gandhi doesn't have to eat, right? And I'd say, no, no, no, no, that's not what I said. And I would, we would go back and forth, and he would continually choose to misread me as meaning that hunger cannot force you to eat. So I kind of think that that's what he really meant, but he didn't want to say it because then he would have to admit it actually was an influence.

[01:13:41]  Peter: So I guess I can almost understand that.

[01:13:43]  Unidentified speaker: So on some meta level, it's really our ideas rather than our bodies causing us to eat. Ultimately.

[01:13:53]  Peter: I can still see that.

[01:13:55]  Unidentified speaker: But if you're going to ask a thousand random people if there's a relationship between hunger and eating, they're going to think the question is so stupid, they can't even get their minds around it.

[01:14:15]  Unidentified speaker: I'm not sure they'd be wrong.

[01:14:18]  Unidentified speaker: And you know what?

[01:14:19]  Bruce: The fact that there are numerous CritRats that will argue that there is no causal relationship between hunger and eating makes us look foolish. And we absolutely need to drop incredibly stupid arguments like that. Even though there may be a certain sense in which he's right, what he should have done is immediately say, yes, of course, it's an influence. Here's the key thing though. It's never actually deterministic.

[01:14:43]  Unidentified speaker: It's still ultimately your ideas.

[01:14:44]  Bruce: He could have confirmed the other viewpoint as correct because it is correct while still getting his own viewpoint in. That would have been the correct critical and he wouldn't have looked like a fool. It's a truly foolish argument. It's embarrassing as a Critical Rationalist, as a CritRat myself, I am embarrassed by arguments like this. They are humiliating that we have a whole community that makes arguments this bad.

[01:15:11]  Unidentified speaker: They're not that hard to correct.

[01:15:13]  Bruce: If his real thing is that he wants to emphasize, look, what I'm really getting at is that ultimately ideas are what determine things. You can say that without having to deny an obvious truth.

[01:15:25]  Unidentified speaker: You can.

[01:15:27]  Bruce: There's just no reason to criticalistically take everything that somebody says and try to force it into the meaning you want so that you can deny it, and you don't have to admit that they've got a point. That's really what, in my opinion, he was doing. I say he, but I literally had a conversation with multiple CritRats, and all of them made the same arguments. Honestly, it's frustrating. It's the reason why I get frustrated with CritRats, that you cannot get them off of these really awful arguments like this. I'm not saying they're wrong. I can see that there's this true idea there. I'm trying to get them to state the true idea and not state a false idea in its place. That's all I'm after.

[01:16:11]  Unidentified speaker: I'm not trying to—and of course, I'm being treated like, oh, you just are claiming it's deterministic.

[01:16:16]  Unidentified speaker: No, I'm not claiming that.

[01:16:18]  Bruce: It's really about ideas. I'm agreeing with you, it's about ideas. I always agree with them on their points, and then go right back to, but it is an influence, isn't it?

[01:16:26]  Unidentified speaker: I just could not get them to say it.

[01:16:29]  Bruce: They were not going to say the words, hunger influences the behavior of eating. They weren't going to say it, even though we all know it does.

[01:16:39]  Peter: Yeah, I guess it seems like an example of where your philosophies can almost move you into positions that are just so absurdly contradictory to just common sense. I mean, one of the things I like about Critical Rationalism is that I think in general, it just coincides with common sense quite well in many respects, but not that one, I guess.

[01:17:12]  Bruce: And of course, all this is ignoring the obvious fact that we all know hunger influenced us to eat. And this person I'm talking to, all these people I've talked to, they've all experienced it. They know how uncomfortable your genes make you if you don't eat, to try to get you to eat and try to realign your interests with theirs. And neither theory, and this is the key point I was trying to make to him, when he claimed that smoking did cause cancer, but hunger didn't influence eating, both of those, the problems he was raising for why he felt he could deny that hunger caused eating.

[01:17:48]  Unidentified speaker: Oh, it's just correlative.

[01:17:51]  Bruce: He had these arguments he was using. It's just bad philosophy. It's just a correlative study. By the way, a randomly controlled trial is not just a correlative study. That is a complete misunderstanding. But that was the argument that they're making, right? And I'm like, okay, smoking causing cancer really is just a correlative study. Like there is no random controlled trial you can do for it. And yet you're prepared to call that one just because it's a physical process and not an ideas-based process. That one you're prepared to buy off on as good philosophy and good science, and you're not prepared to buy off on the other one. It makes no sense. It really makes no sense. Okay, this is the problem in a nutshell with Deutsch's explanationless science argument. It becomes an easy excuse to dismiss any theory you don't like while ignoring tests that contradict your beliefs. In reality, there is no minimum requirement for a good explanation beyond simple causation. That is enough to get the scientific process started by testing the theory. The real flaw, if you understand Critical Rationalism correctly, the real flaw of such shallow explanatory theories is that because they explain so little, it should be easy to propose competing explanations that are also independently testable. But that requires actually providing independently testable alternatives, a requirement that CritRats, at least as of today, reject.

[01:19:13]  Bruce: Instead, all CritRat pet theories are labeled metaphysical, so testing is deemed unnecessary, and they lack Popper's original concept that this makes their theory inferior to a testable competitor. Instead, they judge theories is better or worse, based on which, quote, explains more, really meaning offers a more emotionally satisfying explanation, and which, quote, survives criticism best, really meaning survives my subjective criticism best. So unsurprisingly, it's always their pet theories that come out on top under their epistemology. Directly related to this problem is Deutsch's insistence that you can't define or model what an explanation is. Deutsch seems to have Gödel's Theorem in mind. You can't perfectly model mathematics because then you couldn't model Gödel's Theorem, yet because you can't model Gödel's Theorem using a mathematically formal system. Yet we all recognize Gödel's Theorem as accurate. So by analogy, explanations can't be fully pinned down without precluding new modes of explanation. Honestly, frankly, this whole argument is misguided. In fact, it's actually Roger Penrose's argument, and it's the same misunderstanding as Roger Penrose's argument. It makes explanations seem mysterious and magical. Honestly, CritRats now treat explanations and creativity as if they're mysteriously mysterious and magical and supernatural.

[01:20:39]  Bruce: So consider, for example, that many experiments show that some animals like great apes understand limited explanations, and CritRats, when you bring that up to CritRats and you show, look, there's actual experiments where they put the great ape or the elephant into the setting. It has to have an explanation to know what to do, and it is able to do it. There's no possibility based on the rules of the experiment that it could have learned it through classical conditioning. We've talked about this extensively in past episodes. This is very real experiments that have been done.

[01:21:09]  Unidentified speaker: They're repeatable.

[01:21:11]  Bruce: This is a well-known [under?] thing that certain animals do have explanatory power in a limited degree. Okay, what we might call explanatory power. My experience is that when you pointed out to CritRats, they simply don't label that power as explanations. They feel no need to define what they mean by explanations because Deutsch said it was impossible to define anyway. They personally judge that animals have no explanations as they define that term or rather don't define that term. And that no experiment or theory of explanation would convince them otherwise. In fact, because any actual definition of explanation that is testable would, by definition, be bad philosophy because it's impossible to define what an explanation is, they would reject it on that grounds. You end up in a situation where it's sort of meaningless to say, well, only humans have explanations, but you have no idea what you mean by explanation. Whatever it is, it's not what animals do. That's pretty much where we're at with the argument at this point. I suspect Deutsch has this exactly backwards.

[01:22:15]  Unidentified speaker: It's like claiming you can't model mathematical formal systems, can't create mathematical formal systems, and it would be pointless to even try because of Godel's theorem.

[01:22:26]  Bruce: Now, in fact, that is exactly what Godel's theorem says, is that you can't hem in mathematics through as a single all-encompassing formal system because of Godel's theorem. That doesn't stop us in the slightest from coming up with formal definitions of mathematics. It would be silly to stop formalizing mathematics just because of Godel's theorem. It works as a perfect excuse to dismiss any theory of intelligence that is mathematically precise and forever leaves creativity, knowledge, and explanation, and intelligence as effectively supernatural. If you're going to take the stance, you cannot formalize this idea. It is a supernatural idea for all intents and purposes. This brings us back to perhaps the biggest problem with Deutsch's version of Critical Rationalism, the near fetishization of explanations and explanatory power. But trying to define your epistemology in terms of something as vague as explanation, which you also insist can't be modeled or defined, that really is deeply problematic. Without a precise theory of what an explanation actually is, and Deutsch is denying that such a theory is possible, you might as well just say, go with your gut. In practice, CritRats have weaponized the it's a better explanation card to its maximum damage. Any theory they don't like that relies on probability theory, which is literally all theories, literally all theories, they dismiss as just about correlations, likening it to measuring body humors.

[01:23:54]  Bruce: I cannot tell you how many CritRats have used the body humor analogy with me on all sorts of theories, all sorts of well-corroborated, highly scientific theories. But their own theories are always, quote, explanatory, even when they have no testable content. They're supposedly, quote, good metaphysical theories, of course. A theory that explains something earns credit in their minds, but only if they personally like it. No need for it to actually constrain reality in some meaningful way. Now, Brett Hall's Theory of Intelligence was explanatory in their minds. We've covered this extensively in past podcasts. Sure, it couldn't account for a few things, like senility or severe mental incapacity, but simply offering an explanation that made sense to them was enough to declare it the, quote, best theory and to consider alternatives, quote, refuted, meaning they personally found other theories less compelling. Not that the alternatives failed any tests. Again, I don't doubt that Deutsch is correct that science and knowledge creation is about seeking improved explanations.

[01:24:57]  Unidentified speaker: I'm not arguing against that.

[01:25:00]  Bruce: But defining your epistemology around such vague terms as explanations or creativity is like trying to build a theory of governance, starting with maximum freedom or minimize coercion, terms so abstract that they are almost certain to be misused. Let me quote Popper here from archives of the Open Society and its Enemies revisited. I do believe in freedom and reason, but I do not think that one can construct a simple practical and fruitful theory in these terms. They are too abstract and too prone to be misused. So, likewise, I say, I do believe in finding better explanations, but I do not think one can construct a simple practical and fruitful theory using that term. It is too abstract and too prone to misuse. Popper's approach, modeling explanations as logic, was exactly the right approach because it made his epistemology testable and open to objective criticism. Deutsch's version, by contrast, is easy to vary and nearly immune to criticism. Almost anything can count as a quote better explanation if you lack a precise idea of what an explanation is and you don't value independently testable content. If your goal is to produce hard to vary explanations that are easily testable and thus improvable, a vague term like better explanation and explanatory power are unhelpful. Put simply, if your aim is to maximize explanation and explanatory power, your first step should be to discard the word explanation and reframe the problem so you can formulate theories in a testable way.

[01:26:30]  Bruce: This is exactly what Popper did by modeling explanations logically, and that is what Deutsch lost when he chose not to. This brings me, and this is all true even though modeling it as logic is not ultimately going to prove the best way to do it. There's almost assuredly a better formal way of understanding explanations or something related to explanations that would allow us to understand Popper's epistemology better than trying to model it as logic. But logic was the right way to begin, because it's the way that was available and it was the way that was easily testable and could be error corrected. Now, this brings me to my next concern with Deutsch's interpretation of critical rationalism. The easy-to-vary slash hard-to-vary criteria are problematic.

[01:27:14]  Unidentified speaker: I talked about this in a previous episode.

[01:27:16]  Bruce: On the surface, the easy-to-vary criteria seems correct to me. I perhaps maybe disagree with Deutsch that it is the sole way to judge an explanation as good or bad, or maybe not. I'm a little unsure. I can easily see the easy-to-vary criteria helps explain why some theories are better than others. It makes sense that a good explanation has no spare parts. It needs every part of its explanation. If an explanation like the story of Persephone can be easily adjusted on the fly to account for any data, I can see that that's a very big problem. I would agree that's a very big problem.

[01:27:53]  Unidentified speaker: That would make it a bad explanation.

[01:27:56]  Bruce: But when it comes to the easy-vary criteria, it seems to me that every day is backwards day and I'm the only one that didn't get the memo. Let's consider some concrete examples from past episodes. Think of any of the many theories CritRats have around coercion.

[01:28:12]  Unidentified speaker: What do they mean by that?

[01:28:13]  Bruce: The word coercion implies being forced, but with a decisively negative connotation. Now, even hardcore libertarians, the same ones who claim governments are coercive and oppose governments because they are of their coercion, they have no problem with so-called second use of force. But what is second use of force? It's usually explained as physical violence used in response to someone else first that initiated violence. Let's say someone breaks a contract and keeps money not due to them.

[01:28:47]  Unidentified speaker: They've committed no violence.

[01:28:48]  Bruce: So you can't coercively force them to give back the money? No, says the CritRat Libertarian. Breaking a contract counts as use of force. So this is actually second use of force.

[01:28:58]  Unidentified speaker: See?

[01:28:59]  Bruce: So apparently second use of force means enforcement of laws I agree with, while first use of force means anything I don't agree with. Basically saying that's coercive means I think it's immoral. And saying the government is coercive must then literally just mean, I'm an ANCAP, but I have a [strongly non-held rational?] belief that governments can and should be replaced by markets. It has to be non-rational because there is no way to arrive at that conclusion rationally as of today. Thus, by definition, governments are coercive. Do I have that right? How about when your boss coerces you at work? Maybe yelling at you or threatening to fire you if you don't work the weekend of your Hawaii vacation. Now that's coercive, right? No, says the Libertarian CritRat. You can quit anytime, so that's not coercive at all.

[01:29:47]  Unidentified speaker: Oh, okay.

[01:29:48]  Bruce: But how about when you put pressure on your kid to prevent unwise behavior? That must not count as coercion now since there is no use of force, and even your boss yelling at you doesn't count as coercive. So surely just putting understandable verbal pressure, and that's it, verbal pressure on children shouldn't count as coercive, right? No, says the CritRat. Putting too much pressure on children is coercive. This whole set of theories on coercion seems to rely on the idea that what feels coercive is what you already disagree with. If you don't disagree, it isn't coercive. If this doesn't qualify as an easy-to-vary explanation, I'm honestly not sure what does. Coercion simply is changing meanings on the fly to fit any data you want. No, says the Libertarian CritRat. This is a hard-to-vary explanation. Recall also that in my blog post that I read on this podcast, that I actually sent to CritRats the blog post that I was reading, and I offered how to actually test if their theories were easy to vary or not. Not a single one replied to me. I think about all the times I pointed out to CritRats that their explanations are easy to vary, which is many, many, many, many, many times. I once told Dennis Hackethal that his theories around animals not having feelings based on the fact that they have automatic movements were wholly easy to vary. Humans have automatic movements too, a fact that Dennis does not dispute. Yet he responds that humans are different because they are, quote, creative.

[01:31:19]  Bruce: He says to me, prove animals are creative, meaning animals have human levels of intelligence and he'll change his mind. Until then, he considers this a knockdown argument, even though the argument is, as far as I can tell, entirely circularly being applied to animals. When I pointed out that this was an easy to vary argument, Dennis said, that was my go-to argument.

[01:31:40]  Unidentified speaker: I overused it.

[01:31:41]  Bruce: I was always telling people they had easy to vary arguments. I should stop using that argument because it just wasn't true. Or consider Brett Hall's Theory of Intelligence, which we've discussed many times in the show, which by his own omission cannot be refuted by any imaginable data. Or how CritRats easily vary the criteria for knowledge, to always return to the Two Sources Hypothesis, Episode 131. Or my ANCAP CritRat friend, Episode 132, who advanced wholly untestable praxeology as our best theory of human behavior and the proper basis for understanding economics, because economics is, quote, all metaphysical. To me, this theory is entirely easy to vary. On that, we apparently disagree. He has assured me it is hard to vary explanation. Or take Donald Campbell's and Karl Popper's ideas about universal Darwinism, aka evolutionary epistemology. I would have thought that was hard to vary, yet Vaden dismissed it as easy to vary, and I was in shock, and I still am. I sincerely cannot recall a single instance where I pointed out a theory was easy to vary and a CritRat agreed with me, nor can I recall a single case where someone pointed out to me that a theory I was exploring was easy to vary, and I agreed. Not once, in either direction, unfortunately. It seems to me that the easy to vary, hard to vary criteria is basically meaningless as a practical criteria. But you can't say the same about Karl Popper's criteria for not allowing non-ad hoc theories and judging them by their lack of ad hocness, by their lack of ad hocness or ad hoc they are or not.

[01:33:20]  Bruce: I can't recall a single instance where I pointed out to a CritRat that their theory was not independently testable, i.e. ad hoc, and they did not agree with me. When I pointed out to the defenders of Brett Hall's Theory of Intelligence that their theory has no independently testable consequences, they immediately agreed with me, though they saw this as a strength. Since CritRats today don't see non-testable theories as inferior to testable ones, they have no problem declaring every single theory metaphysical and admitting it has no independently testable consequences. Dennis Hackethal will openly admit his Theory of Animal Intelligence has no independently testable consequences, since he's theorizing that animals will always act exactly as if they feel pain, but just don't happen to feel it. So his solipsistic theory starts with the assumption that it cannot be tested. My ANCAP CritRat friend apparently agreed that praxeology and his economic theories had no independently testable content. We only actually disagreed on whether that was an epistemological flaw and whether the testable competitors were therefore better. Even in episode 100, when Peter asked David Deutsch for evidence, about evidence, that children responded well to certain authoritative parenting styles, as I recall it, Deutsch immediately labeled his theory a moral theory without independently testable consequences, as a way of explaining why he doesn't take such studies seriously.

[01:34:48]  Unidentified speaker: Which I thought was a good response, by the way.

[01:34:50]  Unidentified speaker: I just want to throw that in there.

[01:34:52]  Peter: I really liked how he responded to that.

[01:34:55]  Unidentified speaker: I mean, I'm not saying I'm 100% convinced, but it was thought-provoking.

[01:35:04]  Bruce: So let me say this, though. It's the fact that he gives that answer is exactly what I'm talking about. He was freely admitting it's an ad hoc theory. He wouldn't probably call it that because that's a negative term. He is freely admitting it has no independently testable consequences, right?

[01:35:20]  Unidentified speaker: Yeah, that’s a fair point.

[01:35:22]  Bruce: It's very easy to tell if the theory matches the no ad hoc rule or not. That as a criteria is a definitive criteria that no one ever disagrees over. Whereas the easy-to-vary criteria, no one ever agrees over. It seems to be purely subjective. So whereas no one agrees on what counts as an easy-to-vary theory, everyone seems to agree on what theories are not independently testable. Nothing else, Popper's ad hoc criteria has a virtue. The easy-to-vary criteria does not have. It's testable and it cannot be easily varied. What I find interesting is that I'm half convinced, maybe more than half convinced, that the no ad hoc rule and the easy-to-vary criteria are closely related. They might even be the same criteria. In episode 81, I explored the relationship between ad hocness and easy-to-varyness in detail. I think since then, I've somewhat changed my views, so let me just admit that. I was on the right path, I think, but I don't think I quite had it right yet. I now think Popper's epistemology could accommodate the easy-to-vary criteria as a partial or maybe even full explanation of what ad hocness really is, which is why I'm starting to wonder if they're the same thing. This deserves its own follow-on podcast and would distract from the current topic, so I'm not going to go into that today. Understand, though, that I do not reject the easy-to-vary idea outright. I just reject it as being a good criteria. At least in its current form, it is an awful criteria and it needs to be integrated with the ad hoc criteria which is a truly useful, practically useful criteria.

[01:37:02]  Bruce: The strength of ad hocness is that declaring your explanation non-ad hoc clearly places the onus on you to explain the testable content of your theory. By contrast, declaring a theory merely hard to vary shifts the burden onto the other person to prove that it is easy to vary, which is effectively impossible because easy-to-varyness is itself easy to vary. The burden of explanation shifting depends on how you word it, which is why ad hocness seems to me to work as real-life working criteria and easy-to-varyness does not. Okay, how about Deutsch's claim that refuting solipsism, that is, accepting realism, is an underlying assumption for all scientific experiments? This is correct, isn't it? As a matter of fact, it is correct. Okay, I completely agree. So doesn't that prove that metaphysical theories can be just as valid or even in some cases more basic or fundamental than some scientific theories? The problem I have with this claim is that it's, well, it's false. Realism is absolutely falsifiable and testable. In fact, all forms of solipsism, save one, are also falsifiable and testable. Now, I covered this in episode 99, but let me quote Deutsch himself here. Okay, this is from Fabric of Reality, page 86. James Boswell relates in his Life of Johnson, how he and Johnson were discussing Bishop Berkeley's solpsistic theory of the non-existence of the material world. Boswell remarked that although no one believed the theory, no one could refute it either. Dr. Johnson kicked a large rock and said, as his foot rebounded, I refute it thus.

[01:38:40]  Bruce: I know everyone and their mother knows for sure that realism and solipsism are for sure not testable explanations and are for sure the quintessential examples of a metaphysical explanation that cannot be empirically tested. But isn't the freaking point that Deutsch is making here is that realism makes testable predictions about the independence of the rock and that what will happen, in Deutsch's words, that the rock will kick back at you if you kick it? Why doesn't that count as testable, even empirically testable?

[01:39:20]  Unidentified speaker: Isn't that exactly what we would call empirically testable normally?

[01:39:24]  Bruce: Why do we call this a metaphysical theory? Solipsism avoids this only if you explicitly choose to form a solipsism that exactly mimics realism. There seems like there's a lot of confusion over this. There's solipsism, when people say solipsism, they actually often don't mean solipsism in general. They mean the specific version that exactly mimics realism. Somehow that term has come to be associated with that one form of solipsism. Okay, but in this case, Deutsch argues you might as well call it realism. So Deutsch argues, quote, Thus the solipsist's explanation of the world is in terms of interacting thoughts rather than interacting objects. But those thoughts are real and interact according to the same rules that the realist says governs the interactions of the objects. Thus solipsism, far from being a worldview stripped of its essentials, is actually just realism disguised and weighted down by additional unnecessary assumptions, worthless baggage introduced only to be explained away.

[01:40:23]  Unidentified speaker: Fabric of Reality, page 83.

[01:40:24]  Bruce: But if by solipsism you mean any form of it that does not exactly match realism, then that form of solipsism is an empirically testable theory. Deutsch points out in The Fabric of Reality that, for instance, if you were a prisoner in a VR chess simulator, you would quickly be forced to conjecture the existence of an outside world due to inconsistencies in your explanations. Quote, if the prisoner stuck in a simulation fail after a while to improve upon their existing explanations, they might, of course, give up. But if they do not give up, they will be thinking about those aspects of their environment seem inadequately explained. Thus, if the high technology jailer wanted to be confident that their rendered environment would forever fool their prisoners into thinking that there is no outside world, they would have their work cut out for them. The longer they wanted the illusion to last, the more ingenious the program would have to be. It's not enough that the inmates be prevented from observing the outside. The rendered environment would also have to be such that no explanations of anything inside would ever require one to postulate an outside. The environment, in other words, would have to be self-contained as regards explanations. But I doubt that any part of reality, short of the whole thing, has that property. That's Fabric of Reality, page 139. So all but one form of solipsism would be as testable as realism. You can almost just imagine this.

[01:41:50]  Bruce: If we lived in a simulation and everything outside the solar system was like [didn't really there?] and we're just receiving light waves, you could conceive an experiment to determine you were in the simulator, right? Of course, everyone and their grandma insists that realism and solipsism are untestable, supposedly metaphysical explanations.

[01:42:10]  Unidentified speaker: But they are far from it.

[01:42:12]  Bruce: We judge their goodness and their badness as explanations in exactly the same way Popper laid out, via how much they constrain reality and thus can be tested. So you can't cite solipsism and realism as examples of metaphysical explanations being better or more compelling than testable ones. They are testable in a very real sense, in the sense that matters to Popper, to Popperianism. Interestingly, the so-called untestable solipsism is actually just a form of realism. In that sense, it is as testable as realism. The real issue with untestable solipsism is that this version of solipsism starts with the assumption that it will entirely mimic realism, making it indistinguishable from realism. Similar to Dennis Hackethal's Theory of Animal Feelings, which theorizes that animals act as if they feel pain and offers no theory as to how feelings or not feeling pain should change one's actions. Deutsch is correct to dismiss this form of solipsism as realism in disguise, relying entirely on realism for its explanatory power. And I would add, I feel exactly the same way about Dennis Hackethal's Theory of Animal Feelings. I admit this dismissal is important, and I admit it precedes any discussion of testability. If this is, this is probably what Deutsch's actual point is. And when put in this way, I agree with it, okay? Getting your epistemology right does come before testability becomes meaningful, okay? This is, by the way, it's one of the big Bayesian mistakes. I will have to back up such a bold claim in some future podcasts.

[01:43:45]  Bruce: But one of the big problems of Bayesianism is that they don't really stop and think about what types of explanations are good ones to begin with. And so therefore, they allow for bad explanations that happen to have high testability to be used as competitors and it leads them astray in some cases.

[01:44:05]  Unidentified speaker: Okay.

[01:44:05]  Unidentified speaker: This is a case where they would benefit from understanding Critical Rationalism.

[01:44:09]  Bruce: But the key point, at least the Popperianism that I'm advocating for. But the key point is that we're dismissing such solipsistic theories precisely because we can see that they're being framed to have no independently testable consequences. They violate the no ad hoc rule. None of this is grounds for the conclusion that some untestable metaphysical theories should be given priority over testable ones based on vague ideas about being more explanatory. Independent testability is and continues to be the gold standard. Finally, let's discuss verisimilitude. While David Miller convincingly shows that Popper's notion of verisimilitude is impossible, I confess that I don't see how Deutsch's fix improves anything. Deutsch rewords it as closer to reality instead of closer to truth, which reflects his understandable concern with Popper's formulation. But is there actually any difference between the two formulations? How does saying closer to reality address Miller's problem, that when you have two false theories, one can't be more correct than the other? Doesn't that precisely mean one can't be closer to reality than the other? It seems Deutsch's reformulation suffers from the very same problem as Popper's, with no clear improvement. Now, how about Deutsch's suggestion that we focus on knowledge and problems, rather than verisimilitude? Now, I confess, I don't see how this solves the problem either. Now, I can understand why it might feel intuitively satisfying.

[01:45:46]  Bruce: Saying something like, don't worry about whether you're getting closer to the truth or reality, just solve problems that you have. On the surface, this reformulation might seem to work.

[01:45:54]  Unidentified speaker: But it raises a question.

[01:45:55]  Bruce: Are you advocating for instrumentalism? Perish the thought, says the Deutschian CritRat.

[01:46:01]  Unidentified speaker: That's bad philosophy.

[01:46:03]  Bruce: But when you say that we should, quote, just solve problems and not worry about the truth, I only see two possible interpretations of what you just said. You either mean, who cares if the theory is true? Just use it instrumentally to solve your current problems. Keep all versions of theories around and use the appropriate one for your current problem.

[01:46:23]  Unidentified speaker: If this is what you meant, this is clearly instrumentalism.

[01:46:26]  Bruce: Or you might have meant, take your theories, solve problems, and you'll end up with theories with fewer problems. Presumably, the second was implicitly intended, but it was worded to paper over the fact that it's then identical to verisimilitude. Why is it even possible to create theories with fewer problems, if not for the fact that some theories are closer to the truth than others? If you are claiming it is possible for a theory to have fewer problems, that would seem to be exactly what David Miller's proof shows is impossible. So one formulation of just-solve-problems is equivalent to instrumentalism, and the other is equivalent to verisimilitude.

[01:47:08]  Unidentified speaker: Is there some third option I'm not seeing?

[01:47:10]  Bruce: Like, if there is, I want to know what that is, because that would actually maybe solve the problem here. I'm unable to see it. I don't see how a third option could even logically exist. I don't see how Deutsch's version is an improvement. If anything, Deutsch's version masks the real problem that Popper and Miller identified by being vague about what was intended by the phrase solving problems instead. Popper's version makes it clear there is a serious problem to solve, which is precisely why his formulation is superior to Deutsch's. I think the trick that is being used here is that if you just say concentrate on solving problems instead of getting closer to the truth, you are being just vague enough that you could either mean solve problems instrumentally or it could mean improve your theories by solving problems, essentially getting closer to the truth. By committing to neither interpretation, you may feel like you solved the problem. The instrumentalist interpretation does in fact solve this particular problem, but at the expense of realism, a more significant problem that is not worth the solution. The realist version does not solve the problem at all, but maintains realism. By being vague, you can fool yourself into believing that you solved the problem without becoming an instrumentalist. In reality, you didn't solve the problem, you fooled yourself by papering over the problem.

[01:48:29]  Unidentified speaker: Or at least that's how I see it today.

[01:48:31]  Bruce: I don't see a good alternative. I feel like this should be treated as a real problem, and I think CritRats should take it seriously, that their epistemology has a very real problem that has never been solved. However well-intended Deutsch's adjustment to Popper were intended, and however much they were attempts to solve real problems that needed solving, I'm left with the overwhelming conclusion that something got lost along the way that should not have been lost. I think I would summarize it like this. Popper's original epistemology, rooted in the idea of reducing ad hocness and increasing independent testability, had a strong virtue. It rooted the goodness of an explanation or theory in something we could all check for ourselves. This intersubjectivity meant that the criticism process became objective. Deutsch's adjustments removed that trait, almost across the board. We're left with one subjective judgment on how well does this criticism land for you, and how well do you personally feel this explanation explains things.

[01:49:27]  Unidentified speaker: The net result was predictable.

[01:49:29]  Bruce: CritRat epistemology became entirely subjective. Throughout this episode and across this podcast in general, I've advocated for an interpretation of Popper that adds back the objective element that the Deutsch interpretation lacks. I think the correct interpretation of Critical Rationalism is that you follow the no-adhoc rule to the hilt, even for so-called metaphysical theories. You should give direct preference to theories and explanations that say something about reality and that you can check against reality, even if by reality we only mean mathematical or logical reality, not necessarily physical reality. This is one where I agree with Deutsch. I think reality is bigger than the physical world. Quote, insofar as statements speak about reality, they must be falsifiable. And insofar as they are not falsifiable, they do not speak about reality.

[01:50:15]  Unidentified speaker: That's two fundamental problems, Popper, page 10.

[01:50:18]  Bruce: Note, in context, he was talking about empirical theories, but I feel it more broadly applies. I don't see why it doesn't also apply to, say, mathematical reality or logical reality as well. I've in the past referred to CritRat epistemology as the seek criticism, correct errors interpretation of critical rationalism. Now, there's nothing wrong with seeking criticism, nor correcting errors, of course. So they're right insofar as it goes. But this is generally interpreted as being up to you to decide for yourself what counts as a valid or invalid criticism, or even what counts as explaining more or explaining less. If Popper's epistemology really does boil down to nothing more than you seek out criticism and error correct errors that you happen to personally find compelling. Literally everyone qualifies, at least in their own head, as sincerely doing Popper's epistemology. Every religious person you've met qualifies, every suicide bomber qualifies, every troll on the internet qualifies. This then calls into the question the actual value of Popper's epistemology. I continue to instead advocate for the no ad hoc theory exhaustion interpretation of Popper, where theories that are independently testable are given preference, and you are never allowed to save your theory from refutation unless your new auxiliary theory used to save it is also itself independently testable. Now, this version of Popper forces you to run out of options fast, allowing theories to get effectively falsified in real life, even if you don't have a second theory yet.

[01:51:47]  Bruce: I think this really is, I'm not saying this is an unproblematic thing. There's questions here about what about metaphysical theories. I've advocated for the idea that we simply more loosely interpret the ad hoc rule and the idea of independent tests to include things like checking against mathematical or logical reality. I actually think that these are solvable problems, but I admit that they're problems, that Popper has left a big hole here that he never really filled. But I think that the right way to solve this problem is to not lose that objectiveness that existed in Popper's original theory, and to really see that as central. I think that objectiveness comes from the no ad hoc rule and the idea of independent tests. And giving preference to theories that are independently testable. There's still so much to solve here, really. We're really at the beginning of something exciting. So, Deutsch is expressing and identifying real problems with critical rationalism.

[01:52:47]  Unidentified speaker: I don't doubt that in the slightest.

[01:52:48]  Bruce: And while I don't personally think his solutions work, that means that there are exciting philosophical problems to work on and that are in need of fixing. I have some ideas how to fix these problems, as I kind of just hinted at. But I doubt I can fix all of them, or really any of them fully in a formal way yet. I think there's exciting work to do. I think if you concentrate on the problems themselves, and don't try to gloss over them, don't try to say, that's not a problem, that's not a problem. That you will literally start to see what it is, the true power of Critical Rationalism, of Popper's epistemology, and that you will end up with something that's far superior to what's kind of currently going around as Critical Rationalism in its popular forms. I think you start to see ways in which science could be improved. I think that there's really exciting work to be done here, and I really want to advocate that people start taking the problems of Critical Rationalism more seriously, and not try to explain them away, and that we work on them, and that we improve them, and we actually develop something that's better to replace what we currently have.

[01:53:53]  Unidentified speaker: That's actually the end of my thoughts for today.

[01:53:57]  Peter: Well, Bruce, you're a passionate person, and someday the rest of the world will catch up with you, I hope.

[01:54:13]  Unidentified speaker: Hello again.

[01:54:14]  Peter: If you've made it this far, please consider giving us a nice rating on whatever platform you use, or even making a financial contribution through the link provided in the show notes. As you probably know, we are a podcast loosely tied together by the Popper Deutsch Theory of Knowledge. We believe David Deutsch's Four Strands tie everything together. So we discuss science, knowledge, computation, politics, art, and especially the search for artificial general intelligence. Also, please consider connecting with Bruce on X at BNielson01. Also, please consider joining the Facebook group, The Many Worlds of David Deutsch, where Bruce and I first started connecting.

[01:54:57]  Unidentified speaker: Thank you.