Episode 82: Popper’s Ratchet

• Links to this episode: Spotify / 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

[00:00:08]  Bruce: Welcome to The Theory of Anything podcast. Hey, Peter.

[00:00:10]  Unidentified speaker: Hello, Bruce.

[00:00:11]  Peter: How are you doing today?

[00:00:13]  Bruce: Good. Hey, we've got an episode that I'm excited about. As an epistemology geek, I don't know if anybody else is gonna be excited about it, but we've been kind of building up to this episode. And the last one where we talked about easy to vary versus ad hocness, and why I preferred the concept of ad hocness because it was more objective. And in that episode, I gave examples of easy to vary explanations that were nonetheless empirically testable, such as spontaneous generation. And I also showed that there are no hard to vary explanations, not even things like quantum mechanics, which has got to be the most hard to vary explanation or Darwinian evolution. There's none out there that can't be easily varied in the sense that they actually are easily varied. Quantum mechanics is constantly easily varied around trying to explain or explain away many worlds, for example.

[00:01:09]  Bruce: So I said in a future podcast, I'm going to argue that there are no theories so easy to vary that with the proper critical attitude, you can't decide to make it testable and therefore actually test it. And I didn't back in the previous episode explain how that was possible. I'm going to make good on that promise today.

[00:01:30]  Peter: Now this episode is kind of your magnum opus, right? Is that how I'm thinking of it? Are you sure it's just—are you sure you can do this in one episode? Maybe we should make this like a three parter seven. I don't know.

[00:01:43]  Bruce: So I don't know if I'd call it a magnum opus or not, but it is I'm going to reveal something in this episode that I think is one of the single most interesting aspects of Popper that I don't think I have ever heard anyone talk about. Like it is just totally missing from discussions about critical rationalism. I'm going to show it comes straight from Popper. Like it's not something I'm making up, right? But it's an aspect of Popper that I think is super important and in fact critical to understanding critical rationalism. And it just gets no publicity at all. So I'm really excited to reveal that and to say, here's this thing from Popper that you missed that's really exciting. So maybe we can take it from that standpoint, okay? So let me just do a quick review from our previous episodes. So in the last podcast, I challenged the idea that easy to vary, of easy to vary-ness on the grounds that it is subjective.

[00:02:43]  Bruce: So everyone, when they talk about, is this explanation easy to vary or not, they judge for themselves subjectively if their theory counts as easy to vary versus hard to vary. And not surprisingly, everybody thinks their personal theories are hard to vary, and everybody thinks the theories they disagree with are easy to vary. I also challenged the idea that we can always equate a good explanation to be hard to vary, and a bad explanation to be easy to vary. There are other reasons for an explanation to be bad or good, besides Deutsch's, to quote Deutsch, barely connected to the details of the phenomena. Okay, and I'll give examples later. However, I offered Karl Popper's alternative to easy to vary-ness, which is ad hoc-ness. And I really think when it comes down to it, David Deutsch's term, easy to vary, was meant to mean Popper's concept of ad hoc-ness, but he thought it was an easier to understand term.

[00:03:38]  Bruce: And I think, unfortunately, it's a, it is maybe somewhat easier to get intuitive feeling for what it means than the term ad hoc-ness. But it turned out to be a subjective term instead of Popper where ad hoc-ness was a very objective term.

[00:03:53]  Unidentified speaker: Okay.

[00:03:54]  Peter: That's a big aha moment for me there. I can see the connection now between easy to vary-ness and ad hoc-ness.

[00:04:02]  Bruce: Yes.

[00:04:03]  Peter: But easy to vary-ness could be seen as also more specific too, right?

[00:04:08]  Bruce: It can be. So I think the issue is, is that Deutsch defined easy to vary-ness as, that the explanation is barely connected to the details of the phenomena. That's a quote from Beginning of Infinity. But when you really talk to people about easy to vary-ness, they don't treat it only as that. They treat it as something far wider, right? And I think what they really sort of intuitively have in mind is ad hocness.

[00:04:33]  Unidentified speaker: I think that's right.

[00:04:34]  Bruce: And so I think the two concepts were really meant to be the same concept. Now, the opposite of that hard to vary-ness is what Popper calls boldness. It's the idea that your theory is taking risks. And again, Popper's version is more objective, okay? Where Deutsch's might be easier to get an intuition for. So that's why maybe both are useful. Like if you're just trying to give somebody intuition, referring to easy to vary versus hard to vary might be the way to go. But if you're trying to get the person to really understand objectively, what does this mean in practice, you really need to go with the Popper version because it's objective. Or at least this is my argument. So I showed the two concepts were strongly related, that ad hocness in some sense subsumes easy to vary-ness, but in a way where it's objective. It's objective to tell if our theory is ad hoc or not.

[00:05:25]  Bruce: This saves Deutsch's concept of easy to vary-ness by subsuming it. And this is an important thing that you'll often hear me say, where I'll say critical rationalism subsumes induction. When you talk to a lot of critical rationalists, they'll act as if induction is completely worthless as an idea, because Popper disproved it or something along those lines, right? And I don't think that's the right way to look at these ideas. Induction played such a critical role in the Enlightenment, in how we thought about science. And it would be a mistake to downplay the important role that it actually played, right? And I think this idea that critical rationalism subsumes it, or at least attempts to subsume it, largely subsumes it. There may be some question as if it fully subsumes it or not. That's a discussion for another time. I think that this idea helps us understand that a lot of these ideas are okay to have all at the same time, right?

[00:06:21]  Bruce: As long as you understand how they're actually connected. So there's nothing wrong with easy-to-variness, but I do want to suggest that when you really get down to, is my theory easy to vary, the very first thing you should do is turn to the concept of ad hocness and say, is it independently testable? And if it is, then it's... And then determine if it is or it isn't. So the idea, though, is that ad hocness subsumes easy-to-variness into something that is both more expansive and also more testable. Thus, it is a better theory, okay? Just like Popper's critical rationalism is a better theory than induction as a theory, even though induction's got some truth to it. I think when I say induction has some truth to it, I mean the truth to it was exactly insofar as it was an approximation of critical rationalism. So that's the concept of subsuming something.

[00:07:14]  Bruce: So Karl Popper explains what ad hocness is. He says, a good theory is not ad hoc. The idea of ad hocness and its opposite, which may be termed boldness, are very important. 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. This is from Objective Knowledge, pages 15 to 16. By the way, I'm going to be reusing a lot of the same quotes I did from the previous episode, but I'm going to be putting them into a different context this time. This is why ad hocness is, unlike easy-to-variness, not subjective. Your theory either has other independently testable consequences, that is to say it has reach, to use the Deutschian term, or it doesn't. It is up to you to work out what they are. Going on quoting from Popper, we require that a new theory should be independently testable.

[00:08:14]  Bruce: That is to say, apart from explaining all the explicanda, that which we are trying to explain, 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. That was from Conjecture and Refutation, page 327.

[00:08:52]  Peter: And it does sound a lot like Hard to Vary-ness.

[00:08:54]  Bruce: It does. In fact, it sounds so much like it that you can almost see that Deutsch took it from those quotes, right? That he has taken that concept from Popper, straight from Popper, and then put his own words to it that he thought were a little easier to make sense of. Okay, so I argued that Easy to Vary-ness captures three intermingled issues as if they are a single issue, which is probably why it's a little easier to intuit, okay? Because you don't have to break it out so much. So it captures how precisely the theory is stated via axiomatic statements that make it easy to see if you are surreptitiously changing the theory on the fly. And I had a quote from Popper in the previous episode where he talked about that. It also includes how much reach the explanation has due to number one. And I specified here by reach I explicitly mean empirical reach, how testable the theory is due to having consequences that we can check in the real world that weren't part of the problem we were trying to solve, okay?

[00:09:55]  Bruce: And then the third thing that intermingles is whether or not the person defending the theory is willing to ad hoc save the theory from a refuting counter example by introducing untestable auxiliary hypotheses. So similar to number two, your auxiliary hypotheses are also expected to have reach, okay? Now you can see how these three things are clearly all intermingled. Number two is based on number one, and number three is taking it to the next level. Not only does the main theory have to be independently testable, but you're not allowed to save it through an auxiliary hypothesis unless the auxiliary hypothesis is itself independently testable, okay? Or that is to say, not ad hoc. In other words, easy to vary versus hard to vary is partially about the nature of the theory and partially about the critical attitude of the defender of the theory.

[00:10:47]  Bruce: That is to say, their willingness to introduce untestable auxiliary hypotheses to save the theory from refutation. I then argued that while number one and number two explain what counts as ad hocness, number three, the choice to hold yourself to not use ad hoc explanations, was the most important of the three criteria. And I equated this to having a proper critical attitude. Or as Popper put it, I propose that the empirical method shall be characterized as a method that preclude, that excludes precisely those ways of evading falsification. That's from Logic of Scientific Discovery, pages 19 and 20. I argued that the critical attitude, that is to say the choice not to use ad hoc explanations, was the most important factor because all theories have places where the theory is underdeveloped or a bit vague, or where we just can't test it yet.

[00:11:43]  Bruce: So, example, quantum mechanics has this part where you can't test if the other worlds exist or not today. So, there was no such thing as a theory, no matter how specific and testable that it is, that couldn't be easily varied by its defenders, by just simply picking the parts of the theory that were a bit vague, or that there were what I called degrees of freedom, or where it just wasn't testable yet. And by introducing ad hoc auxiliary hypotheses or arbitrarily playing with the degrees of freedom of the theory, if they choose not to have the right critical attitude. And then I said, in a future podcast, I'm going to argue the inverse of this. There is no theory so easy to vary that with the proper critical attitude, you can't decide to axiomatize it in such a way that it becomes testable and has reach. This is what we're going to talk about today now.

[00:12:34]  Bruce: So let's go back to the example that comes from the beginning of infinity that Deutsch uses to explain his concept of easy-to-variness. So from page 21 of Beginning of Infinity, he says, for whenever it is easy to vary an explanation without changing its predictions, one could just as easily vary it to make different predictions if they were needed. On page 22, he says, when theories are easily variable in the sense I have described, experimental testing is almost useless for correcting their errors. I call such theories bad explanations. However, I showed at the time, there was in the previous episode, there was something wrong with this. We took spontaneous generation as an example of an easy-to-vary theory, and we showed that science had in fact tested and falsified that theory, despite it being an easy-to-vary theory. In fact, we struggled to find any examples of easy-to-vary theories that couldn't be tested or falsified.

[00:13:28]  Bruce: But what about this myth of Hades and Persephone that Deutsch is using? Can we do that same trick just like we did with spontaneous generation? Could we actually falsify the myth of Hades and Persephone? That would be the key thing I would want to be able to do. Now, this may sound silly, like who cares? It's just obviously a myth. It's so completely as an explanation disconnected from trying to actually explain the seasons that it almost seems like it's worthless to try to falsify it. So, Deutsch goes on to explain why he thinks it's kind of worthless to try to falsify it. So, somewhere between pages 19 and 22 here, he says, But even testable explanatory theories cannot be the crucial ingredient that makes the difference between no progress and progress, for they too have always been common. Consider, for example, the ancient Greek myth of explaining the annual onset of winter.

[00:14:27]  Bruce: Long ago, Hades, god of the underworld, kidnapped and raped Persephone. Goddess of spring. When Persephone's mother Demeter, goddess of the earth and agriculture, negotiated a contract with her daughter's release, which specified that Persephone would marry Hades and eat a magic seed that would compel her to visit him once a year thereafter. Whenever Persephone was away, fulfilling this obligation, Demeter became sad and would command the world to become cold and bleak, so that nothing could grow. That myth, though comprehensively false, does constitute an explanation of seasons. By the way, this is where I just want to point out, I've often had crit rats online say, that explanation that they don't like doesn't explain. It doesn't take much to be an explanation. There's a difference between being an explanation and being a deep explanation.

[00:15:14]  Bruce: And notice that Deutsch is admitting, this does count as a constituted explanation of the seasons. But then he goes on to say, because it's a claim about the reality that brings about our experience of winter, it is also imminently testable. If the cause of winter is Demeter's periodic sadness, then winter must happen everywhere on earth at the same time. Therefore, if the ancient Greeks had known that a warm growing season occurs in Australia at the very moment when, as they believe Demeter is at her sadness, they could have inferred there was something wrong with their explanation of seasons. The reason those myths are so easily variable is that their details are barely connected to the details of phenomena. That freedom to make drastic changes in their mythical explanations of seasons is the fundamental flaw in them. It is the reason that myth-making, in general, is not an effective way to understand the world.

[00:16:11]  Bruce: And that is so whether the myths are testable or not. For whenever it is easy to vary an explanation without changing its predictions, one could just as easily vary it to make different predictions if they were needed. For example, if the ancient Greeks had discovered that the seasons of the northern and southern hemispheres are out of phase, they would have had a choice of countless slight variants of the myth that would be consistent with that observation. One would be that when Demeter is sad, she banishes warmth from her vicinity and it has to go elsewhere, into the northern and southern hemisphere. Similarly, slight variants of Persephone explanation could account just as well for seasons that were marked by green rainbows or seasons that happened once a week or sporadically or not at all. Again, from beginning of infinity between pages 19 and 22.

[00:17:08]  Bruce: So to Deutsch, myths and scientific explanations are fundamentally different sorts of things. While there is definitely truth to this, in our last episode we show that one can have a bad explanation for an entirely different reason not related to this problem of having the explanation be only barely connected to the phenomena. One could be making a circular argument, for example. We talked about this idea from the argument that Dean made that defining loving violence as effectively anyone who is a non-libertarian, you then have this perfectly circular argument that could never be defeated. That would be an example of an ad hoc argument, but it's really hard to say that it's an easy-to-vary argument. Or using an all-purpose argument. We talked about selectively declaring any theory that you don't like as invalid due to, quote, not explaining all the way down, when in fact no theory does that.

[00:18:05]  Bruce: Therefore, if you don't like the theory, you say, well, it doesn't explain and therefore all evidence is just correlations. But if you do like the theory, the theory that in this case wasn't liked was IQ theory, this idea that there's a G factor, a general intelligence factor. Any evidence that is in favor of it, that would be a problem for the refutation of the competing theory, Brett's theory in this case, is immediately dismissed as that's just a correlation. But then when we're talking about something that they do agree with, say, does smoking cause cancer? Oh, yes, that's not just a correlation, that's an explanation. And you can see how this would very quickly become an all-purpose argument that you can basically dismiss anything you don't like and accept anything you do like by using this argument. And again, it doesn't seem like it fits very well with this idea of easy-to-variness where the argument is barely connected to the phenomena.

[00:18:59]  Bruce: The argument here is strongly connected to the phenomena, and yet it's still completely ad hoc. Okay, or just change your definitions on the fly. Declaring disobedience to be key to understanding general intelligence by simply redefining disobedience to mean doing something new and thus equivalent to human level creativity. Or by adding an untestable ad hoc auxiliary hypothesis to avoid refutation. These are all examples of ad hocness that don't fit comfortably into the concept of easy-to-variness unless you stretch that term more and more to be equivalent to ad hocness. In fact, it's fairly rare that an explanation is a bad explanation explicitly because the details are barely connected to the details of the phenomena, like Deutsch claims. This is one of the main reasons that no one ever feels like their personal theories are easy to vary, since their theory is connected to the details of the phenomenon.

[00:19:56]  Bruce: So consider how Dean really and truly thought his choice to claim that José loved violence boiled down to José simply being a non-libertarian. His reasoning was something like democracies or governmental institutions do not reduce violence. José is in favor of democracies and they use institutional violence, which is a kind of illegitimate violence, according to Dean. Thus, José loves violence. I mean, there's an actual logic to this, right? The problem isn't the logic. The problem is that every ounce of these explanations are ad hoc to begin with, and so there's no particular reason to prefer them yet. This is a bad explanation to be sure, but you can't claim it's not connected to the details of the phenomena that he is interested in. Interestingly, Popper held a very different view of the relationship between myth and science. Okay, so let's go over what Popper said, because I really feel like this is important to understand.

[00:20:54]  Bruce: Popper says, this is from Objective Knowledge, page 347, the oldest scientific theories are built on pre-scientific myths. Now, what is new in Greek mythology, this is the same page, what is newly added to all this seems to me to consist not so much in the replacement of myths by something more scientific, scare quotes are his, not mine, as a new attitude towards myths, that these explanations character then begins to change, seems to me to be merely a consequence of this new attitude. The new attitude I have in mind is the critical attitude. Notice how Popper is agreeing with my idea that the critical attitude is the most important part, not the nature of the explanation. In place of a dogmatic handling of the doctrine. That's Objective Knowledge, page 347. On the other hand, it seems to me that the task which science sets itself, that is the explanation of the world, and the main ideas which it uses are taken over by science for myth without any break from the pre-scientific myth-making, Objective Knowledge, page 348.

[00:22:09]  Bruce: Popper instead claims that the real difference between science and myth is not the nature of the theories, as Deutsch was claiming, so much as the critical attitude of the individuals. In place of myth-making, this is again Popper in Objective Knowledge, in place of myth-making, we find the tradition of criticizing theories, which at first themselves are hardly more than myths. It is only in the course of this critical discussion that the observation, that observation or empirical observation is called as a witness. So this is Objective Knowledge, page 348. So to summarize what he just said, he's saying actually the first explanations that eventually became what we think of as scientific explanations today, they all started off exactly the same as myths. There was no difference between myths and scientific explanations. And it was actually the critical attitude of the people discussing them that eventually led to this idea.

[00:23:06]  Bruce: We need to make our theories have empirical consequences. It was their attitude that led to that. And that's what Popper is claiming.

[00:23:13]  Peter: It seems to me you're kind of that Popper and Deutsch both had a similar attitude and they see myths, some almost like some validity in myths as sort of a pre scientific, something that led into science. Is that fair? They had in some ways more similarities than differences in their ideas about myths. But what you're saying is almost like a question of emphasis in how they explain that process. Is that fair?

[00:23:43]  Bruce: So mostly, yes. So I'm getting to what I think is the one big difference. But I think you're right that it's easy to see that Deutsch is clearly in some sense, almost quoting Popper, right? I mean, like a lot of Deutsch's epistemological ideas come straight from Popper. You can now almost find the quotes that he's using that he's then changing the wording on. So I do think you're right that there's a great deal of similarity to their views here. But I do want to call out that there is a difference. And it's going to turn out that this difference, it may seem slight at the moment, but it's going to turn out to be a big difference. And the difference is that Deutsch tries to explain what's wrong with myths, that they are easy to vary, because the phenomena is barely connected, the explanation is barely connected to phenomena. And Popper is not taking that attitude.

[00:24:35]  Bruce: In fact, he's saying there is not a fundamental difference between myth-making and scientific explanation. They're a spectrum. And that what really makes the difference is the critical attitude. Now, let me explain why that's true. Okay, so the key difference here is Deutsch and Popper is, to Deutsch, only certain kinds of theories can be tested. To Popper, once you have the right critical attitude, you can turn myths into valid and testable scientific theories. That's what Popper just said. And that is what's going to turn out to be the one big difference between Deutsch and Popper on this. Okay. At the same time, this is Popper again, Conjecture and Refutation, page 38. At the same time, I realized that such myths may be developed and become testable. That historically speaking, all or very nearly all scientific theories originate from myths.

[00:25:27]  Bruce: And that a myth may contain important anticipations of scientific theories. Examples are, I can't pronounce Greek names, [unclear Greek name; local ASR renders ‘Epamodality’s’; quoted source likely Empedocles’] Theory of Evolution by Trial and Error, or Parmenides' Myth of the Unchanging Block Universe, in which nothing ever happens and which, if we add another dimension, becomes Einstein's Block Universe. So this is why Popper doesn't really have a negative view of myths. He actually sees them as there's no real break between myths and scientific explanation. But here is the problem. What do we mean by critical attitude? So Popper's saying the difference was critical attitude. And is a critical attitude any less subjective than the term easy to vary? Now, every single one of the crit rats that has made arguments that on this show we've taken a look at and would say, oh, that is a bad explanation, particularly in that last episode where I gave examples of that, when we deconstructed them, they definitely, we deconstructed their arguments, we showed that there was problems with their arguments, that they were making bad explanations.

[00:26:35]  Bruce: But all of them thought they had a critical attitude, right? Have you ever met anyone who doesn't think they have a critical attitude? I've never met someone who doesn't very honestly believe that they have a proper critical attitude and that they have arrived at their beliefs, whatever they are, by very carefully thinking it out. And everybody else has the improper critical attitude, or else they would see that they were right, right? I don't care how wild the theory is. I don't care if it's a conspiracy theory. Everybody believes they have a proper critical attitude.

[00:27:09]  Peter: I mean, I guess you could take someone who's extremely wrapped up in something like romantic philosophy, believes in just following your feelings. I mean, that's not so uncommon. But then again, there's sort of a contradiction because how they probably, if you pressed them about how they would have arrived at those those ideas about life, they would probably try to defend it as as through something like a kind of reason that they thought. People always look for reasons, no matter what you.

[00:27:42]  Bruce: Yes.

[00:27:43]  Peter: Yeah.

[00:27:44]  Bruce: OK, so I'm going to solve that problem. I suggested the way to solve that problem. The fact that everybody thinks they have a critical attitude. So therefore, when we say you need to have the proper critical attitude, and that's what Popper's epistemology is really, really about, that's entirely subjective. So I'm instead going to define critical attitude as explicitly meaning I follow the no ad hoc rule. And this is going to be critical for me explaining my point. Where ad hoc means I will only offer you explanations that have enough reach that they have testable consequences unrelated to the problem we're trying to solve. Or in other words, that all explanations offered will be formulated in such a way that they take empirical risks. This is what I'm going to define as critical attitude means. Okay, so now let's do an experiment. Let's take the myth of Hades and Persephone and let's add Deutsch's test of Australia.

[00:28:48]  Bruce: So again, we have this idea that Hades and Persephone, we got this myth, it explains the seasons. We discover Australia that it's warm right when Demeter is supposed to be sad. We've now offered this as a test of the theory of Hades and Persephone, and we've now refuted the theory of Hades and Persephone, okay? Now here's the thing. It is definitely true, as per Deutsch, that we could easily save this myth by making a quick change to the theory, just like Deutsch suggests. And the fact that the explanation, a relationship between gods, is so disconnected to the phenomena in question, trying to explain the seasons, this is particularly easy for us to do. But here is my challenge for you. Avoid that counter example of Australia from refuting the theory of Hades and Persephone by offering me a non-ad hoc explanation. That is to say, you must offer me a counter explanation to save the myth that has its own testable consequences.

[00:29:53]  Bruce: And the consequences can't be something we already know about. You have to offer an explanation in such a way that we can all agree there is some previously unforeseen outcome, similar to general relativity with the Eddington Expedition, and then we get to go test it. So Peter, I'm going to give you that challenge. Can you offer me, can you avoid using Australia as a refutation of the myth of Hades and Persephone? Can you avoid it while holding yourself to the requirement that the new explanation you offer must be non-ad hoc, that is to say, independently testable?

[00:30:29]  Peter: Bruce, when you said, when you said, here's my challenge to you, I was hoping you were talking about the listener and not me. I can't think that quickly. Okay, I'm still on my second cup of coffee. Let's just move on.

[00:30:43]  Bruce: So I don't think there's any way to do it, right? It's, you might make a mistake here and think, well, I could make it testable. I could like come up with some myth and I could say, and I'm going to test it by, and then offer some other phenomena that it explains also, or something like that. But recall that according to Popper, it has to be independently testable, and it has to be something, and I just gave the quotes on this, that wasn't foreseen, okay? That if you take the right attitude, that I'm only going to offer those types of explanations, that refutation of Hades and Persephone will stand as a refutation. And you won't be able to come up with an alternative explanation to explain away that refutation. And by merely holding us to the no ad hoc rule, even myths can be tested and refuted. This is why I can't agree with Deutsch, that when theories are easy, this is quoting Deutsch, when theories are easily variable, experimental testing is almost useless for correcting errors.

[00:31:41]  Bruce: Page 22 of Beginning of Infinity. Once you accept the no ad hoc rule, even so-called easy to vary explanations can be put to a test, empirically. This is why, I'm going to argue, Popper believes the real difference between myths and science are not some objective difference between the kind of explanation being offered, but instead boil down to one's critical attitude, or rather, adopting the no ad hoc rule. As well as making your theories axiomatic and logically precise enough that we can work out consequences in the same way we'd work out other logical consequences. And this is how I actually refuted creationism back when I was a creationist. I mentioned that I had refuted creationism, and I didn't explain how. Let me now explain. Despite not knowing about Popper at the time, I have always had a rather critical attitude. I didn't merely offer up problems to Darwinian evolutionists that they couldn't solve, although I did do that.

[00:32:39]  Bruce: And in fact, as I pointed out, some of these problems I offered up to them, they couldn't solve even to today. And the types of explanations they would offer me were very clearly ad hoc saves. And at the time, that seemed like it was kind of a strike against Darwinian evolution in favor of creationism to me. But here's what happened. I also researched what problems Darwinian evolution did solve. And then I chose to take on the burden of seeing if I could come up with an alternative creationist explanation for those same phenomena. So, for example, this may seem like a silly example, but remember, I'm a kid, all right? I need to explain where dinosaur bones came from. So maybe I'd read in a creationist book that they lived before Noah's flood. And then I'd ask, okay, if that's true, then how might we test that? I'd then look up how they dated dinosaur bones, and I'd compare their dates with when Noah's flood was supposed to happen.

[00:33:39]  Bruce: And I'd find that there was hundreds of millions of years gap that I couldn't explain between what the dating was saying and when the Bible said Noah's flood was. Undaunted, I'd then try to explain why the dating was off. But I'd find that no matter what theory I came up with as to why the dating was wrong, I couldn't possibly close a 200 million year gap like that. Eventually I realized that even though Darwinian evolution had serious explanation gaps, and that was what I was finding when I was debating Darwinians, that couldn't be explained. The creationists had at least that much right that they did have explanation gaps. It was nothing compared to how serious were the gaps in creationism as an explanation if one adopted a proper critical attitude towards it. And I would need something like Darwinian evolution to close that explanation gap that creationism had.

[00:34:37]  Bruce: In other words, I took upon myself the burden of coming up with a testable explanation. Now I know many creationists that do not do that and instead give a hand wavy response like, oh, it's complicated or only God knows or something like that, right? What separated me from these other creationists was my critical attitude of accepting the no ad hoc rule. Of course, I didn't know about Popper at the time, so I didn't call it that. But that was the attitude I had naturally.

[00:35:09]  Peter: Now were you ever tempted to sort of pivot to something like intelligent design or, you know, not that there's anything wrong with that necessarily, but I mean, you can kind of there's different takes on the creationism debate, right?

[00:35:23]  Bruce: Yes. Yes. And I looked into each of those. Like this happened over a period of time. And I would have to probably like have a separate podcast to explain each of these. In many ways, they're all they've all got the same problem, right? That ultimately these like intelligent design basically adopts evolution as its explanation, except when it doesn't. And so it's not hard to see that what it's really doing is it's trying to come up with an explanation that borrows the strength of Darwinian evolution as an explanation. And then tries to find counter examples just enough that then it can still claim that there's some something else going on. Right. And as an explanation, it doesn't work. Right. It's I mean, it could be true, but it's a completely, completely untestable explanation because they're always going to say anything you could explain.

[00:36:17]  Bruce: Oh, well, that was Darwinian evolution and anything you can't. Oh, that's a god in the gaps. Right.

[00:36:23]  Peter: Oh, yeah.

[00:36:23]  Bruce: And the gaps kept kept shrinking. If you read Francis Collins book, The Language of God, he's a he's a devout Christian who grew up as an atheist, became a Christian. And then he's famously the guy who is in charge in charge of the Human Genome Project. He's one of the greatest scientists of our time. Right. And he wrote a book written to people who believed in intelligent design and creationism to try to talk them out of it. I actually listened to his book, and that was one of the things I did while I was looking into a lot of these things.

[00:37:01]  Peter: This is as a teenager.

[00:37:03]  Bruce: So when I listened to his book, I would have been in my 20s by then. At that point, I think I more or less was an evolutionist, but I just wanted to get his take on it. And he's basically an evolutionist too, right? So this is why you can't really say that creationism is or isn't easy to vary. To me, it wasn't easy to vary. To other creationists, it might be. Or rather, I chose not to easily vary it, and other creationists chose to easily vary it. But the same could be said of quantum mechanics. Most physicists today treat quantum mechanics the same way that most creationists treat creationism. They simply ignore the no ad hoc rule, and this is talking about many worlds, obviously, and utilize ad hoc explanations to explain away the many worlds' consequences of quantum mechanics. So that's why I do think it's more a matter of critical attitude.

[00:37:58]  Bruce: But you have to understand critical attitude not meaning as subjectively, I think I've been critical, but do you accept the no ad hoc rule? And moreover, do you accept the burden of explanation? People often talk about burden of proof. Usually both sides try to assign the burden of proof to the other side. And I know many crit rats bristle at the term burden of proof because they see proof as anti-fallibilist. By the way, it's not. That's something we would have to do a separate podcast on. But I can at least understand why they would bristle at that term. But I think the proper way to look at this term is more as burden of explanation. Moreover, who has the burden of explanation is always objective. You must come up with only independently testable theories, and you can only save them from refutation via independently testable auxiliary hypotheses.

[00:38:55]  Bruce: This is what is missing in the discussion with most of the discussions I've had with CritRats. I am not a supporter of IQ theory, and I want to make that clear. I really don't think IQ theory is a particularly good theory. I was far more interested when I approached Brett and I approached the defenders of his theory to try to understand where they were coming from and how they were trying to explain it. I wasn't really in favor of IQ theory. And I actually do think Brett's Theory of Intelligence has promise, right? If the defenders of the theory will stop being dogmatic about it and start to treat it like with a proper critical attitude, so they can error-correct it and improve it into a real scientific theory. Because you can take theories like this, and you can turn them into real empirical scientific theories if you have the right critical attitude.

[00:39:44]  Bruce: But their issue today is they do not even slightly accept the burden of explanation that they carry. They need to work out non-ad hoc explanations using their theory that explain why people lose their intelligence due to, say, a severe knock on the head, or a genetic disorder in the case of Down syndrome was the example we used. Or just getting older, the fact that as you get older, that if nothing else, you become senile. You clearly lost your intelligence at that point. But if you don't become senile, people who are older often self-report that they're not as intelligent as they used to be, right? To be able to explain those phenomena in terms of Brett's theory today is completely impossible, right? So they need to take those as serious problems with their theory and take the burden of explanation on themselves, not trying to reassign it to the other person, and say, look, that's a problem I can't solve today, and then try to come up with a non-ad hoc explanation using Brett's theory.

[00:40:43]  Bruce: And what's going to actually happen is, is they'll end up tweaking Brett's theory into something new and better, right? More testable and it explains more.

[00:40:52]  Peter: It seems to go along with the idea of epistemic humility, right? You should always be, you know, talking about an attitude. You should always be open to the assertion that you're, even if the theory you strongly believe in has problems, right? But I'm not sure too many people are, but, you know, on all sides, to be fair.

[00:41:12]  Bruce: On all sides, on all sides. You know, and I need to be careful here. Because I am in the CritRat community, and this podcast is about that, obviously it makes sense for me to take examples from the CritRat community and criticize them. But I do not even for a moment think CritRats are somehow more dogmatic or less rational than really anybody else I've met, right? I mean, every community you're ever going to come across suffers from dogmatic issues.

[00:41:44]  Peter: Very fair point.

[00:41:45]  Bruce: Right, and it just is the way it is, right? You just will not find counter examples to this, right? The one big counter example, and even it, you can argue isn't a counter example, is the scientific community. And even then, if you go look at individual scientists, individual scientists are often very, very dogmatic in ways completely at odds with rationality, reason, or critical rationalism. It's almost more like, as a community, they are more rational, but individually they're not more rational. And I think that's part of the problem is, we're trying to get at, what does it mean to be rational? Why does science work? Why is critical rationalism the correct epistemology? Where does it have problems that we need to correct to get to the correct epistemology? And these are all the questions that I'm interested in and I want to ask. And you do it by looking at the errors, right?

[00:42:38]  Bruce: You look at where things have gone wrong. And this is why I've picked on Brett's Theory of Intelligence quite a bit, right? They simply don't take the burden of explanation seriously today. And instead, they use all purpose dismissals like, well, IQ theory is just fitting data because it doesn't have mechanistic explanations all the way down. Responses like that, when I'm offering you an actual problem that you really need to explain with your theory, they're really trying to shift the burden of explanation back to me when really it's theirs to have. And they are irrational or dogmatic bad explanation responses. It even should be obvious from the outset that there's something wrong with these sorts of responses because it should be obvious that you can't point to the problem of another theory to explain away the problems of your theory. Okay, when they offer such all-purpose explanations to me, I know this means that they don't have the slightest idea how to deal with the problem I'm offering them.

[00:43:39]  Bruce: And so they're moving to a shift of the blame, shift of the burden of explanation to me because they don't have a response, which means their theory has a serious problem that they need to admit to and start working on by trying to conjecture non-ad hoc explanations to deal with those problems. How their theory, how can their theory that intelligence is all about time and interest and that's it be used to explain senility, Down syndrome, or a knock on the head? They need to embrace this as an interesting problem to solve, and it is an interesting problem to solve, right? And they need to do it in a testable way using their theory as the starting point. And this is really what it means to have a critical attitude. You almost fall in love with the problems of your theory. You want to solve the problems of your theory. You don't want to explain them away.

[00:44:33]  Bruce: You don't want to shift the burden to somebody else. You want to grasp that problem and say, it's up to me to explain this, and that's interesting, okay? Now, I'm going to offer you an explanation for why the no-adhoc rule works. But keep in mind, when I say the no-adhoc rule, I'm including the idea that you're making your theories logically precise enough that we can all agree what predictions it should or shouldn't be making using logic, okay? So let's make a few assumptions. First of all, I'm going to assume realism is true. There is a real world out there and it is distinct from us. Now, it's not an accident that Popper was a realist. It's actually key to understanding his epistemology, okay? First, secondly, I'm going to assume explicability. So this I'm taking from Deutsch. The real world follows natural, explicable laws that can be precisely put in terms of logic or mathematical statements.

[00:45:33]  Bruce: Okay, notice the connection there with the church-turing Deutsch thesis. That's not an accident either. Deutsch could see that there was a connection between Popper's epistemology and this idea that Popper relied on that you can put everything into logical statements that you could then use deductive logic on. Okay, and then I'm going to assume the idea of truth. Statements about the world can be true or false or maybe partially true. That is verisimilitude. Okay, now if I accept each of these assumptions, I can now draw the following conclusion. A true theory about the world can always be made logically precise or mathematically precise if you prefer. And in that form, it will have reach, that is to say, independently testable consequences. So that follows directly from those three assumptions now. Now, corollary. A false theory about the world either can't be made logically mathematically precise and still have reach, or if it is, it won't make true predictions about the world.

[00:46:38]  Bruce: The no ad hoc rule, including the idea that we can make out theories, make our theories logically precise, plays off these conclusions. It says in essence, in essence, we could summarize Popper's epistemological view as, okay, if your theory is true, then you should have no objection to making it logically precise and then working out consequences none of us can foresee without the theory. Okay, I'm going to say that again because I feel like this is super important to understanding what I'm trying to get at here. Popper's epistemology could be summarized as saying, in essence, okay, if your theory is true, then you should have no objection to making it logically precise and then working out consequences none of us can foresee without the theory.

[00:47:33]  Peter: I think we need to put that on a meme or something. Bruce's face and the quote.

[00:47:41]  Bruce: All right. Now, why is the no ad hoc rule a good idea? Well, consider Campbell's and Popper's idea of evolutionary epistemology. By the way, for the record, Campbell called it evolutionary epistemology. Popper felt like that term was pretentious. So instead, he called it towards a theory of evolutionary epistemology. But I'm kind of calling them both evolutionary epistemology as a short form. But let me admit Popper didn't necessarily like the term. Okay. So their epistemology was that all inductive achievement actually comes from variation and selection, or if you prefer, conjecture and refutation. Evolution is therefore under this view, a search algorithm. Now, I'm not the first person by far to have noticed that evolution is a search algorithm. In fact, we quoted back in the episode on the problem of open-endedness, this realization from like Leslie Valiant and from people working in the artificial intelligence field, that they understand that the algorithms that they use are evolutionary algorithms, are a kind of search algorithm.

[00:48:51]  Peter: That's a fascinating way to think of it.

[00:48:53]  Bruce: Okay. Now, in fact, let me even just quickly, let me just offer something here. Now, first of all, let me just say that this may only be approximately true, not completely true. And it would be a totally different podcast for me to get into why I have to say that. Okay. But for the moment, we're going to accept it as true or at least as highly having a lot of verisimilitude. If you're doing variation and selection, how could that ever not be a search algorithm? After all, you're trying things. That's a search, right?

[00:49:24]  Peter: But you're not trying everything, of course. But I guess maybe, could you say that about any search algorithm, I suppose?

[00:49:31]  Bruce: Yes. So search algorithms can consist of comprehensive searches that then come with guarantees. This is something they teach you in artificial intelligence school. Or they can be not comprehensive and you're just trying to find a local minima. Okay. So evolutionary algorithms are always search algorithms. Period. End of story. Now, can the reverse also be true? So let's take any search algorithm. Okay. So I'm doing any sort of search. Let's say I'm just doing an A star search to find a shortest path, you know, or I'm trying to just search through every single thing in a stack, trying to find the one needle I want. All of those are examples of going through many variations and then having some sort of selection criteria by which you select the best one. So all search algorithms are variation and selection algorithms, and all variation and selection algorithms are search algorithms.

[00:50:35]  Bruce: The two concepts may well be one and the same. And this is, in essence, what Campbell and Popper were getting at with Campbell's Theory of Evolutionary Epistemology. This is important. This idea that all search algorithms are, in fact, evolutionary algorithms really comes from Campbell, from his 1970s paper. And I challenged it in episode 26. This is why I feel like I need to put a little bit of a caveat on it. Because evolution, but here was Campbell's key point to repeat. We typically understand evolutionary algorithms or evolution in terms of three concepts, variation, selection, but also replicators. And Campbell was the first to notice that the replicators weren't required. That sometimes they're nice to have as part of the variation selection algorithms. They have a purpose. In fact, in machine learning, we actually study what that purpose is.

[00:51:31]  Bruce: And what they do is the replicators then create a sort of proxy for the search space, the fitness landscape that we're trying. You don't need to understand what these terms mean. I'd have to do a separate podcast to explain them. But even without replicators, you can still do variation and selection. And it still works in exactly the same way that evolutionary epistemology works. Because of this, search algorithms and evolutionary algorithms have a deep link, or may even be identical if Campbell is correct. This really helped me make sense of why the no-adhoc rule is so useful. The no-adhoc rule drastically prunes the search tree. Okay, there's this gigantic tree of possible ideas, an infinitely large tree of possible human ideas. And when you say, I'm only going to hear out theories that are independently testable, I'm only going to hear out those theories.

[00:52:29]  Bruce: Even if they're true, I'm not going to hear out your theory if it isn't independently testable. You have drastically reduced the search tree. This is really why the ad hoc rule is such a great thing. Now you might say, well, wait a minute, couldn't that be bad though? And this is something that comes up in machine learning and artificial intelligence. What if your search algorithm prunes out the correct answer? This is really a concern in artificial intelligence. If you make the search algorithm remove the correct answer, you may not know you've removed it, but if you've removed it, then you're going to end up with some sort of approximate answer that's nowhere near as good. So they've got things about guarantees and they're trying to work out. In fact, one of the things I want to argue is that artificial intelligence is actually the study of epistemology.

[00:53:18]  Bruce: But anyhow, that's for another time. So you're only going to consider theories once they have been put in a certain form. That is to say that they are precise enough that we can all agree they have testable consequences beyond the problem we were trying to solve. Because as per our assumptions, all true theories can, at least in principle, be put in such a form, this form that they're independently testable. We are not, according to these assumptions, eliminating any true theory from the search. This is why the no ad hoc rule is so great. It at once drastically reduces the size of the search tree and prunes it, but also never removes a true answer from the search tree. Can you see why this is true? Because I feel like this is one of the most important aspects of Popper's epistemology.

[00:54:15]  Peter: I think so. Where does the concept of, say, a heuristic come into this? Like, say, the chess player that can just sort of rule out 99% of the possible moves and just through some kind of mental process. You see where I'm going with that?

[00:54:36]  Bruce: Yes. So this is something that Campbell talks extensively about in his 1970s paper. He has this idea that search algorithms often utilize a heuristic to prune the search tree because the search tree is often so large that it's intractable. So you have to have some way of pruning the search tree. Again, this is something that comes up in artificial intelligence, like, constantly. Like, the study of artificial intelligence, this is why I wanted to get a master's degree in it, is the study of artificial intelligence is all about how am I going to prune this search tree and how am I going to make this into a tractable problem because it's currently an intractable problem, okay? And heuristics is... One might argue that is what we call... When we do this pruning of the search tree, we call that a heuristic, okay? In any case, heuristic is a way of pruning the search tree.

[00:55:32]  Bruce: And then they have this idea that sometimes a heuristic may prune the search tree without removing the correct answer. They call that an [unclear: immiscible/admissible] heuristic. And sometimes it doesn't. That would be a [unclear: non-immiscible/non-immissible] heuristic. But there's nothing necessarily wrong with removing the correct answer. Like if I'm doing an A-star algorithm, and I use a non-admissible heuristic that prunes out the shortest path, I'm still going to get a path that's really close to the shortest path. And maybe that's good enough, right? So sometimes you want to use a non-admissible heuristic, okay? Now when the chess player is using their gut feel for what a good board looks like based on their past experience, and maybe they can't even explain how they're doing it, what they're doing is they're using a non-admissible heuristic to prune the search tree, okay?

[00:56:20]  Bruce: Which a human player has to do, because a human player just cannot do very much of a search tree compared to a computer. And so humans are very good at this. They have this weird combination. We talked about this in the Problem of Open-Endedness episode. They've got this weird combination of explanation and intuition that they use to be able to play Go at levels that you can't get a computer to play, at least until you get to machine learning where Go is able to start coming up with something similar in terms of intuition that a human player could do. The human players were able to outperform the computer just doing a search. And in fact, if you'll recall, once they actually got AlphaGo to work, it could play at a professional level without doing a search. It would simply make one, it would do a search in that would try one move forward, but that's it.

[00:57:12]  Bruce: That's all it needed to do. And then it could just tell this is going to be the best board position. It could play at professional levels, not necessarily world champion levels, but at professional levels. So a heuristic is really a means of pruning the search tree. So what we would say in is Popper's no ad hoc rule. It's a heuristic. Okay. In terms of how we would call it in terms of AI as a field, but it's an admissible heuristic. It does not remove the correct answer. So it does this huge pruning of the search tree, but it always leaves true answers in the search tree. Now, of course, we're doing this based on those assumptions, realism, that there is such an idea of truth. I can't remember what the third one was. It was explicability, the idea that all explanations can be put into logical terms, computable terms. Logical terms for Popper, but if they're logical, then we can put them into logic circuits.

[00:58:12]  Bruce: Therefore, they're computable. So if you hold on to those three assumptions, then it's guaranteed that the no ad hoc rule will at once drastically reduce the search tree and also leave the true answers there. If it turns out one of those assumptions isn't true, then that would no longer be true. Okay. And let me admit that because, of course, Sadia is going to want to say that her theory doesn't fit into logical terms. What I don't think Sadia would quite get is that's the same as saying Popper's epistemology is wrong. Maybe she's comfortable with that, but these two are connected, right? You can't just split these things apart. Popper's epistemology, Church-Turing-Deutsch thesis, and this idea of realism, this idea of universal explainers and explicability, none of these ideas can be teased apart. They're all kind of all part of one overall theory, if that makes any sense.

[00:59:07]  Peter: And just so we can put this in historical context, Campbell, you keep talking about, this is Donald Campbell, who was a collaborator of Karl Popper in the 70s.

[00:59:18]  Bruce: Yeah, Karl Popper loved his stuff. He was a big fan.

[00:59:21]  Peter: And he's mostly known for this idea of evolutionary epistemology.

[00:59:25]  Bruce: That's right. That wasn't even his main field. That's what I know him for. That's what you hear about him for. But it was like some little side thing he was doing. So I think that often happens, right? You go into one area and then you end up falling into something else.

[00:59:39]  Unidentified speaker: Fair enough.

[00:59:40]  Bruce: And Campbell loved Karl Popper. Like when you read his evolutionary epistemology paper, he's basically just ripping Popper off. Not ripping Popper off. That sounds negative. He's extensively quoting Popper, if that makes any sense, right? So it's probably not too surprising that Popper loved him, right?

[00:59:57]  Peter: I think I've heard of him and his ideas on top-down causation, which is kind of another interesting thing.

[01:00:04]  Bruce: Campbell pops up all over the place. He founded two fields. I don't think intentionally, but he founded two fields. When you are looking at... There's a field called Universal Darwinism that considers Campbell the founder of the field. And there's like a whole field of scientists out there working in this field, right? Publishing papers, things like that, right? And then when I bought a book on evolutionary algorithms by someone who was like an expert in the field, he tries to explain where the field came from, and he actually claims Campbell was one of the main founders of the field, which was a little surprising to me. But he lists out what the first papers were that led to the field, and Campbell's was the earliest. So he doesn't necessarily solely credit Campbell as the founder, but Campbell is the earliest founder of the field.

[01:01:00]  Peter: This is a complete tangent, but I've been meaning wanting to ask you about this. Speaking of the top-down causation, which I think was invented by Donald Campbell, I've been wondering if that's the same thing as another very related idea to Deutsch's ideas about the copper atom and emergence. Yeah, those are related ideas. Do you think they're just different ways of saying the same thing?

[01:01:27]  Bruce: They are different ways of saying the same thing.

[01:01:29]  Peter: So Deutsch would agree with top-down causation?

[01:01:33]  Bruce: If understood in that way, yeah. So I think the problem is that the term top-down causation, it's like free will or it's one of these terms that it comes with so much philosophical baggage, including a lot of really untrue things, that I'm always hesitant to say yes, top-down causation is true because that almost immediately means to some group of people something that isn't true. But I think top-down causation in the Deutsch sense is a totally real thing. And that was what I think Campbell had in mind.

[01:02:08]  Peter: Maybe it would be better if we just replace the word free will with top-down causation.

[01:02:13]  Bruce: Yeah.

[01:02:14]  Peter: I mean, more specific at least. It doesn't sound as cool, I guess, but okay. Sorry, that was a tangent.

[01:02:23]  Bruce: Interesting tangent. Let me back up and repeat it just a little bit. Because as per our assumptions, all true theories can, at least in principle, be put into such a form that they're independently testable, precise enough to be independently testable, we are not, according to the assumptions, actually eliminating a path to any true theory. And we're avoiding wasting time by refusing to deal with any theory that is not yet put into such a form. Meaning that it's precise enough that it has independently testable consequences. Now, interestingly, I don't think Deutsch disagrees with anything I just said. Even though I've been criticizing some of the things he said, I actually think that at some level Deutsch completely agrees with what I'm saying. So Deutsch, in logic of experimental tests, almost says the same thing about how we usually have very few good explanations available.

[01:03:18]  Bruce: So here's the quote from logic of experimental tests, page 11. And although there are always countless logically consistent options for which theory to reject, the number of good explanations known for an explicandom is always small. Things are going very well when there are as many as two, with perhaps the opportunity for a crucial test. More typically, there is one or zero. This is Deutsch saying the same thing. When you require all explanations to follow the no ad hoc rule, to be put into a form that are precise enough that it has independently testable consequences, it is so hard to come up with them that you're doing really great if you've got two. In fact, that is the ultimate gold standard, and it's super hard to get to. You're doing great if you even have one. And this is why this concept of a best explanation really comes down to, it means, I currently have only one independently testable explanation for this phenomenon.

[01:04:26]  Bruce: There may be an infinity of non-testable explanations out there, and they may be the true explanation. But at the moment, I'm going with the one that's independently testable. And the reason why is because if one of those others out in that morass of non-testable theories, if the true explanation is in that morass, it will be possible to eventually figure out how to turn it into an independently testable explanation, and it will eventually become part of the search algorithm. But at the moment, nobody knows how to do that, so I'm not going to consider it any further, because if I do, the problem becomes completely intractable. And this is why science works, and this is Popper's explanation for why science works, and it makes progress so fast compared to the pre-scientific era. So whereas Deutsch defines good theory as hard to vary, which I said was subjective, Popper defines good theory as not ad hoc.

[01:05:26]  Bruce: So here's a quote from Popper, Conjecture and Refutation, page 81. One can show that the methodology of science, I'm going to say that again, the methodology of science and the history of science also become understandable in its details if we assume that the aim of science is to get explanatory theories that are as little ad hoc as possible. A good theory or good explanation is not ad hoc, while a bad theory is. Conjecture and Refutation, page 81. Recall ad hoc means has no empirically testable consequences other than the one problem it was introduced to solve. So while to Deutsch, we are maximizing explanation without usually offering a precise understanding of what we mean by maximizing explanation. Popper offers a very precise criteria for a good explanation. It has independently testable consequences. The no ad hoc rule has some fascinating consequences that I've never seen anyone talk about, and this is the thing I've been wanting to talk about, because just nobody talks about this.

[01:06:33]  Bruce: But they seem to fall directly from Popper's epistemology, the most important of which could be stated like this. You must solve all problems solely by increasing the empirical content of our theories. You are never allowed to solve problems by reducing empirical content of our theories. This follows naturally from the very fact that every problem requires an explanation that if you're following the no-ad hoc rule itself, has its own independently testable consequences. This means that our combined theoretical system will only ever grow in terms of testable consequences, or in other words, it will only grow in terms of empirical content. Popper discusses this idea in Logic of Scientific Discovery, so I didn't make this idea up. I'm getting this straight from having read Popper. He says, as regards auxiliary hypotheses, we propose the rule that only those auxiliary hypotheses are acceptable, whose introduction does not diminish the degree of falsifiability or testability of the system in question, but on the contrary increases it.

[01:07:45]  Bruce: If the degree of falsifiability is increased, then introducing the hypothesis has actually strengthened the theory. I'm hoping that term strengthened the theory, just got every crit rat's hackles up, but this is Popper who's talking. The system now rules out more than it did previously. It prohibits more. We can also put it like this. The introduction of an auxiliary hypothesis should always be regarded as an attempt to construct a new system, and this new system should then always be judged on the issue of whether it would, if adopted, constitute a real advance in our knowledge of the world. For those who claim you cannot strengthen the theory, yes, you can. You can make your theories more empirical. Okay, that's what it means to strengthen the theory. So recall, to use our example of Brett's Theory of Intelligence, the defense would always say that their theory explains intelligence in terms of solely time and interests.

[01:08:49]  Bruce: But Brett understood his theory as maximizing explanation based on this idea that you're supposed to be maximizing explanation. But he found no reason to make his theory independently testable. There were no testable consequences that he would admit to for his theory. But under this idea from Popper that we're only ever increasing the empirical content due to the no ad hoc rule, Brett's theory can be seen as ad hoc because by his own admission it isn't testable and it can't be falsified and it prohibits and rules out nothing. This is no different than communism, which also, quote, maximized explanation in the subjective sense, but not in this objective sense that we're talking about. So to quote Popper, as he put it, a theory which is not refutable by any conceivable event is non-scientific. Irrefutability is not a virtue of a theory, as people often think, but a vice.

[01:09:45]  Bruce: Conjecture and Refutation, page 36. It is only the content of this, it is only in the context of this idea that we are only allowed to solve problems by increasing the empirical content, that we can understand how Popper understood falsifiability or refutability. To many crit-rats that I've talked to, to refute a theory is merely to make a criticism that you personally find convincing. But this isn't what Popper meant by refutability. He specifically understood refutability and falsifiability as being equivalent to the theory having empirical content. Here is what Popper actually says. The criterion of the scientific status of a theory is its falsifiability or refutability or testability. He sees these as synonyms. That's conjecture refutation, page 48. Or, he says on page 36, every genuine test of a theory is an attempt to falsify it or refute it.

[01:10:44]  Bruce: Testability is falsifiability. But there are degrees of testability. Some theories are more testable, more exposed to refutation than others. They take, as it were, greater risks. So to Popper, we compare theories, and we prefer the better explanation by which he specifically means the more testable theory with the higher amount of empirical content. And on logic of scientific discovery, page 105, he says, Thus I regard the comparison of empirical content of two statements as equivalent to the comparison of their degrees of falsifiability. This makes our methodological rule that those theories should be given preference. Say that again, those theories should be given preference, which can be most severely tested. He says that's equivalent to the rule favoring theories with the highest possible empirical content. Okay, there's just no doubt this is what Popper meant, right?

[01:11:45]  Bruce: He's so clear on this. So I'm a little unsure. This is something that we should cover in a separate podcast. How this idea that refutation just meant, I made a criticism that I personally find convincing. When Popper was so specifically talking about, your theory must have independently testable consequences, and it must have actually survived those tests. And it must have empirical content, and the more empirical content it has, the better. And we will always prefer the theory with the highest empirical content. I mean, there's just no doubt. I just quoted him. That is clearly what he meant.

[01:12:27]  Peter: Out of curiosity, how many people have you convinced of this?

[01:12:32]  Bruce: Zero. Zero.

[01:12:35]  Peter: You've convinced me, but it might be more like brainwashing since I've listened to you about 20 hours on it. But anyone else?

[01:12:44]  Bruce: No, I haven't convinced anybody as of yet. So I think what they would say is that's an important criteria. I actually have to go find the quotes. But I've had them say, oh, that's an important criteria, but it's not definitive, like you're saying. It certainly was definitive what Popper said it was definitive. You can disagree with Popper, I guess, and maybe that's exactly what they're doing. But there is definitely not this idea that critical rationalism is explicitly about making your theories increasingly empirical. I've never come across anyone that actually... I shouldn't say that. In terms of discussions on Twitter that I've never come across anyone who agrees with me on that, I can give you very specific quotes of critical rationalists that are famous, like, say, Jonathan Rauch, where he explicitly says that. So there's tons of big names out there that agree with me on this.

[01:13:37]  Bruce: And in fact, probably I in part believe this because I've read their books and I know that's what they said. Like, probably I got the ideas from Jonathan Rauch and from others, right? And from reading Popper, like Popper says these things, I read Logic of Scientific Discovery and went, oh, this is what Popper is saying, right? If you haven't read Logic of Scientific Discovery, I think there's very little hope that you'll actually understand what Popper's epistemology was, which is why I think you should start with that book, even though it's the hardest book.

[01:14:06]  Peter: And that's the first one too, right? The first book he wrote, like in the 30s or something?

[01:14:11]  Bruce: It's the only real book he wrote, right? I think a lot of his other books, I might be wrong on this, but I think a lot of his other books are like collections of essays or speeches or something like that. They're usually not comprehensive books, if that makes any sense.

[01:14:25]  Peter: Well, Open Society and Its Enemies is kind of a...

[01:14:27]  Unidentified speaker: Oh, good point.

[01:14:29]  Unidentified speaker: That is absolutely a three-volume book.

[01:14:31]  Peter: Basically, each volume is book length, I think.

[01:14:33]  Bruce: Yeah, it's two volumes, but each one's book length. So yeah, you're right. That's actually another example. I guess I wasn't counting that one because it wasn't explicitly about his epistemology. Obviously, it's deeply rooted in his epistemology. And that's what he talks about a lot throughout the books. But I think the only book he intentionally wrote about his epistemology was Logic of Scientific Discovery. And the other collection of books are usually... He has the book about his biography, which is actually about epistemology. But in that, it was a book he wrote as a book, right? But it wasn't like a book on epistemology.

[01:15:12]  Peter: It reads more like essays, honestly.

[01:15:14]  Bruce: Right. So I take it back. Of course, he wrote other books. That's not the only book. But it's the only book where he intentionally said, I'm going to sit down and I'm going to explain from start to finish what my epistemology is. Let me just back up a little. The last quote from Popper was that we should give preference to the most severely tested and that that is equivalent, and this is a quote from Popper, equivalent to the rule favoring theories with the highest possible empirical content.

[01:15:45]  Unidentified speaker: Okay.

[01:15:46]  Bruce: Logic of Scientific Discovery, page 105. Popper saw this as directly related to adopting, quote, the decision for laying down suitable rules for the empirical method is to adopt such rules as will ensure the testability of scientific statements. Remember, testability means falsifiability, which is to say their falsifiability. I should have just continued reading the quote. Logic of Scientific Discovery, page 27. So Popper is pretty harsh on the idea that you can just trot out a metaphysical theory as a competitor to an empirical theory, like crit rats often do. This is why it's not okay to claim that you refuted IQ theory using Brett's Theory of Intelligence. Brett's Theory of Intelligence needs to first be turned into, at least as empirical a theory, as the theory of general of G factor, right, of IQ theory. I'm not saying IQ theory is a great theory.

[01:16:41]  Bruce: I haven't read, there's a great article that I've kind of skimmed, but I haven't read in detail, but I'm planning to do in a podcast at some point by Nicholas Taleb, where he says IQ theory is mostly pseudoscience. And kind of skimming through it, like I really like what I've seen in the article. So I've been wanting to dig into it more deeply and do it since we've said so many things accidentally in favor of IQ theory, because we're saying criticisms of Brett's theory of intelligence. The simple truth is, is that IQ theory is a really sucky theory. It has empirical content and it has survived corroborations. That makes it in some sense better than Brett's theory because Brett's theory is not even trying to do that. But that's about all I can say for it. I mean, it's probably a terrible theory in almost every other conceivable way. What we're really looking for is a theory that is at least as testable as IQ theory, but says something different than IQ theory.

[01:17:40]  Bruce: And I think Brett's theory could be turned into that, right? In fact, I could almost tell you right now what the tweak you need to make is to Brett's theory, to make it a competitor, is you probably have to admit that there is such a thing as a damaged brain, and that if you have a damaged brain, then you aren't going to be able to learn as fast. And now you need to work out how that is a factor in addition to time and interest. Now, once you've introduced that, I can see why they're so resistant to introduce that idea. It almost feels like you've just given in to the idea of G factor, but you haven't, right? You could make the case that once you reach a certain point where it's functioning correctly, that IQ could measure a G factor up to the point [unclear wording: ‘where you've now of regular intelligence’], and that everyone of regular intelligence from that point forward, it really is just time and interest.

[01:18:36]  Bruce: Now, I don't know if that theory is true. Like, I'm not even advocating that theory. But that theory now encompasses what the empirical content from IQ theory without giving up on the idea that time and interest are, if they're not the only factors, they're not, they are still the most important factors. And you could almost from here tease out consequences and turn this into a testable theory. Since it's not my theory, I'm not going to be the one to do it, right? I mean, the burden of explanation is on Brett, not me, because this is the theory he's advancing. He needs to figure out how to solve this problem, and it's going to require him to introduce some sort of other factor than the two he wants, okay? But it doesn't necessarily require that he embrace IQ theory. I don't know if that makes sense or not, right? He could come up with something that's an alternative to it.

[01:19:28]  Bruce: And by the way, everything I just said right now, that's totally ad hoc at this point until it has its own testable consequences. Let's be clear about that. But I'm just giving an example of how you could go about this. So this idea that we're only allowed to increase the amount of empirical content, let's give this idea a name. This is the idea that I just quoted from Popper. He clearly says it. I just make sure that there's no doubt. I got this idea from Popper, Reading Logic of Scientific Discovery. But this idea is the one I've never heard people talk about. Okay, let's give it a name so we can talk about it really, really easily. The name I'm going to coin for it, this is my term, is Popper's Ratchet. Popper's Ratchet is this idea that you are only allowed to solve problems by making your theories more precise and increasing the empirical content of your theories, never by making them vaguer or adding degrees of freedom or reducing the empirical content of your theories.

[01:20:25]  Bruce: This is what I'm going to call Popper's Ratchet. That's my term. That's why you've never heard it before. But the idea is 100% Popper. Okay, I made the term up, not the idea. Understanding Popper's Ratchet is key to understanding why discovery of the scientific method exploded into this incredibly fast progress. And it's also key to understanding the difference between dogmatism and a critical attitude. Indeed, it literally defines what it means to have a critical attitude versus being dogmatic. How do we increase the empirical content of a theory? Well, Popper, and I had the quotes from the previous podcast where I quoted Popper. So let me just kind of summarize what Popper said. The key thing is to make your theories more universal so that they are easier to falsify, make them so that they are more prohibitive, if you will, or rather easier to imagine falsifying cases while it's actually impossible to find them in real life.

[01:21:23]  Bruce: Now, this is something that I don't think Popper ever quite explained well. He often talked about falsifiability, but a true theory can't be falsified. So in a certain sense, a true theory could never have any falsifiability. What Popper really has in mind is the ease with which you could imagine a counterexample. So think about like Newton's theories, okay? But for the sake of argument, let's say that Newton's theory had never been falsified, and we never had a single counterexample. For thousands of years, there were no counterexamples to Newton's, thousands of years, hundreds of years, Newton's theories had zero counterexamples, okay? And so people just thought it was just the truth. But it was super easy to imagine what a counterexample would look like. You could just imagine dropping that rock, and the rock falls, and it hits the ground, and it doesn't follow the equations that Newton gave you, okay?

[01:22:18]  Bruce: Falsifiability is the ease with which we can easily work out what a counterexample would look like. It's not actually the ability to find a counterexample, okay? Does that make sense?

[01:22:32]  Peter: Yes, it does. And Newton died in 1727, by the way. I thought he was a little earlier than that, but yeah, so a couple hundred years at least.

[01:22:40]  Bruce: A couple hundred years, yeah. So the key thing here is, if you make your theories more universal, then it makes them easier to falsify. He says, the degree of corroboration of a theory which has a higher degree of universality, this is Popper, can thus be greater than that of a theory which has a lower degree of universality and therefore a lower degree of falsifiability. Related to that is making our theories more precise. And Popper, this is from Logic of Scientific Discovery, page 268. In a similar way, theories of a higher degree of precision can be better corroborated than less precise ones. The second way we could do this is we could remove degrees of freedom. You can make a theory more precise by making it axiomatic and thus easy to tell if it's changed or not. This is what Popper had said. Plus easier to apply logic to. So Logic of Scientific Discovery, page 50.

[01:23:36]  Bruce: Popper says, For a severe test of a theoretical system, or in other words, an explanation, presupposes that it, the theory, is at the time sufficiently definitive and final in form to make it impossible for new assumptions to be smuggled in. In other words, the system must be formed sufficiently clearly and definitively to make every new assumption easily recognizable for what it is, a modification and therefore a revision of the system. This is why precision is so important to understanding Popper's epistemology. It's actually a way of increasing the empirical content of your theory. You make it more precise, make it more universal, but also if you make it more precise, the number of things that you can say, oh, that theory now implies this, and you can go test the theory, goes up, your empirical content grows. Or put it another way, Popper's Ratchet implies that we're always reducing the degrees of freedom of a theory rather than making our theories vaguer.

[01:24:41]  Bruce: Note that if your theory can't be put into precise logical terms, Popper's epistemology could not apply to it. This is actually my main criticism of [unclear name: source ‘Sadeus’; local ASR ‘Sadea’/‘Sadia’] theories. When she says she's got these theories of physics, but they can't be put into computable forms, it's the exact same thing as saying there's just no way to apply Popper's Ratchet to it, and therefore it's entirely outside of Popper's epistemology. That's the part I actually reject. This is, and again, this is how Turing-Deutsch thesis and Popper's epistemology interact, and in fact are one and the same thing. So why is Popper's Ratchet not talked about? I'm going to suggest that there are at least some psychological reasons for why that is. Popper's Ratchet is the secret to why science makes fast progress, but it is not what human beings enjoy doing in their arguments.

[01:25:35]  Bruce: Human beings want to reason using their intuitions. Peter, you and I, we'll talk about things and we'll often say that rings true to me, right? That's using your intuitions. Something feels true. It kind of rings true, even if you can't tell exactly why. Popper's Ratchet takes a lot of the fun out of it. Our intuitions feel correct to us. It's very hard to believe that they aren't just truth. So here is an example that I thought was brilliant. This comes from Hans Rawlings, Rosling, from his book Factfulness, which by the way, if you're a fan of David Deutsch, you will love the book Factfulness.

[01:26:13]  Peter: Oh, yes, yes, I read it. Kind of an essential in the optimism canon.

[01:26:18]  Bruce: Yes, he basically shows that people are overwhelmingly pessimistic and to the point where they have no idea what the truth actually is in terms of the growth of positive things in the world. If you go out and you give them tests, they will actually do worse than a monkey who's just throwing darts because they have such a pessimistic bent that they automatically pick the wrong answer.

[01:26:42]  Peter: It's like the negativity bias, you might say. You're more worried about the cheetah eating your baby than the wonderful campfire you've just made or something.

[01:26:54]  Bruce: So Rosling gives an example that in the moment I heard it, I thought, oh, I got to use this in a podcast. He says, one evening in 1974, I was shopping for bread at a supermarket in a small Swedish town when I suddenly discovered a baby in a life-threatening situation. An untrained eye couldn't see the danger, but fresh out of medical school, I heard my alarm bells go off. I walked over to the stroller as quickly as I could, and I lifted the baby, who was asleep on his back, and I turned him over and put him on his tummy. Now, are you familiar with this at all?

[01:27:30]  Peter: Oh, yeah, we went through that with our babies. I feel like maybe people have come away from that a little bit.

[01:27:36]  Bruce: No, it's the opposite today. And this is the point he's making, that you want your baby sleeping on its back, not its tummy, because there's good evidence that if your baby's sleeping on its tummy, that the incidence of SIDS goes up.

[01:27:51]  Peter: Yeah, that's right.

[01:27:53]  Bruce: So he's actually doing the exact opposite, because back in 1974, they thought it was exactly the opposite. So he goes on to explain that back then, we thought it was dangerous to let babies sleep on their backs because it was well known that during the wars, like World War II and such, soldiers on their backs tended to die if they were unconscious. They would, while unconscious, vomit and drown in their own vomit. So we learned to put them on their stomachs so that the vomit would come out of their mouth and they wouldn't die. So the assumption was that this should apply to babies given that babies also vomit while they're unconscious. Okay.

[01:28:32]  Peter: What I remember now from when I researching this was that I used to worry about our babies doing that, but you have to be like, you know, I mean, when Jimi Hendrix or someone, all these rock stars die of choking on their own vomit, they're heavily medicated.

[01:28:54]  Bruce: Yes.

[01:28:56]  Peter: So when a baby would never, you know, unless you're giving your baby drugs and alcohol, it's always going to wake up when it vomits.

[01:29:04]  Bruce: That's right.

[01:29:05]  Peter: Right.

[01:29:06]  Bruce: And that is the correct answer. So he actually goes on to explain that. He points out that it seemed like it was just pure logic. If soldiers die due to vomit when they're on their backs, babies should too. It didn't feel like a conjecture. It felt like they were just working out an obvious logical outcome of an explanation. But actually, it's far more dangerous for babies to sleep on their tummies, because that leads to SIDS. So he goes on to say, The mental clumsiness of a generalization like this is often difficult to spot. The chain of logic seems correct. When seemingly impregnable logic is combined with good intentions, it becomes nearly impossible to spot the generalization error. Even though the data showed that sudden infant deaths went up, not down, when you put them on their tummies, it wasn't until 1985 that a group of pediatricians in Hong Kong actually suggested that the prone position might be the cause.

[01:30:05]  Bruce: Even then, doctors in Europe didn't pay attention. It took Swedish authorities another seven years to accept their mistake and reverse the policy. Now he explains why. Unconscious soldiers were dying on their backs when they vomited. Sleeping babies, unlike unconscious soldiers, have fully functioning reflexes and turn to their sides if they vomit while on their backs. But on their tummies, maybe some babies are not yet strong enough to tilt their heads up to keep their airways open. He then goes on to say that last part is a conjecture. They're not actually sure if that's the reason or not yet. But it may be the reason why sudden infant death syndrome goes up when you put babies in their tummies, is that some babies may be vomiting when they're on their tummies. They can't move their heads enough, and so the airway stays closed. This has been, in a nutshell, the real problem of induction or generalizing.

[01:30:56]  Bruce: All our generalizations are actually just conjectures. This is actually what Popper is saying, right? That they're conjectures. No matter how much they feel like pure logic, they need to be actually tested to see if we made a mistake somewhere in that chain of pure logic. Popper's Ratchet also reveals how often our favorite ideas are really just meaning memes and not the best theories that we thought they were. It forces one to admit that most of the time, we just don't know what the best theory is because none of the competing theories are actually currently testable, or all of them are equally testable, so there's no clear winner. Some work in some cases and some work in other cases. It forces one to take a far more fallibilist stance, except in the rare cases where best theories, that is to say, there's only one good explanation, that is to say, there's only one explanation that is independently testable, empirically, emerges.

[01:31:54]  Bruce: People seem to have an almost primal urge to believe they've thought things through and that they've worked out the truth. So we're almost constantly in danger of becoming dogmatic. So that's actually the end of the first part of my explanation. I wanted to throw an appendices on there though, that this is a little unrelated, but it's tangential, but I think it's related enough. I want to throw it out there. So as an appendices, one might ask, okay Bruce, even if I buy Popper's Ratchet, how would we ever get conjectures in the first place? I don't know if someone would ask this or not. I asked this. And we talked about this a little bit in the last episode. When Einstein comes up with his conjecture that Newton's theory is wrong and that instead we need to be thinking of some sort of new theory that treats gravity not as a force, at that point his explanation is entirely non-testable.

[01:32:51]  Bruce: And so from a certain point of view, you might say, well it doesn't make sense. We would never even make the conjecture. And it's going to take him eight years. It literally is going to take him eight years to turn it into a testable theory. So for eight years, is Einstein being irrational, preferring his theory over Newton's theory? Okay, so that's the way, that's the Steelman version of the issue I'm trying to raise. Here's my answer to that question. I'm going to answer it with a question. And the question is, did Popper say there was or wasn't a scientific method? This is why I recently had a Twitter tweet where I asked for where did Popper say that there wasn't a scientific method? And David Deutsch himself responded to the tweet and gave me the reference so that I could use it for this podcast, by the way. So thank you, David Deutsch, for helping me out here.

[01:33:42]  Bruce: You hear all the time that Popper said there was no scientific method. And here's the actual quote that gets quoted where he says this. This, the course, he's like teaching a course in the scientific method. He says, the course is in scientific method. It is called Introduction to Scientific Method. And the first thing I want to say as an introduction is that this is a subject that does not exist. Now, you may have noticed, as I emphasized a little, that in the quotes I just made from Popper, explaining what Popper's Ratchet was, that he without a doubt said there was a scientific method. So we've kind of got these two things. One where you've got definitive quotes from Popper where he says there isn't a scientific method, and one's from Popper where he definitively says there is a scientific method. Now, how do we reconcile this? And it turns out that the way we reconcile this is the answer to the question I just asked.

[01:34:41]  Bruce: Specifically, Popper said that part of the scientific method was this idea that we have rules or conventions, such as the no ad hoc rule, which is the basis for Popper's Ratchet, that we follow because they make it so that theories can be falsified, and that if you don't follow them, then no theory can be falsified. And that is what Popper said the scientific method was. It was this group of rules or conventions that you followed. So in Logic of Scientific Discovery, here's the actual quote. I tried to determine empirical—to define empirical science with the help of the criterion of falsifiability, but I was obliged to admit the justice of certain objections. I promised a methodological supplement to my definition. Just as chess might be defined by the rules proper to it, so empirical science may be defined by means of its methodological rules.

[01:35:37]  Bruce: That's page 32 of Logic of Scientific Discovery. First, a supreme rule is laid down, this is Popper again, which serves as a kind of norm for deciding upon the remaining rules, which is thus a rule of a higher type. It is the rule that says that other rules of scientific procedures must be designed in such a way that they do not protect any statement in science against falsification. Page 33 of Logic of Scientific Discovery. So apparently Popper actually explained what the scientific method was, and he claimed it doesn't exist. So how are we going to reconcile this? The answer is, not surprisingly, it depends on what you mean by scientific method. So let's take a look at the full context of what Popper said when he said there wasn't a scientific method. Here's the actual quotes in context. Now let us begin our course, the course in scientific method.

[01:36:33]  Bruce: It is called Introduction to Scientific Method. The first thing I want to say as an introduction is that it is a subject that does not exist. Now the fact that the subject does not exist may strike some as strange. The usual idea of the scientific method is that there is a certain technique or procedure for making scientific discoveries and that you can learn this technique and then apply it in science to make discoveries. According to the usual view, a man either knows scientific method and makes scientific discoveries or he does not know scientific method and makes no discoveries. Unless, of course, he should happen to make some by chance. If this is what is meant by scientific method, then the subject certainly does not exist, for there simply are no techniques or procedures for making discoveries in science. Contrary to the usual view, this is still Popper, there are many examples of scientists and many very great scientists who have only once in their lives made a major discovery.

[01:37:37]  Bruce: These are scientists who, even though they continued to be the forefront of science and were able to criticize other people's discoveries in a most excellent way, and even though they had problems they would have loved to solve and surely tried to solve, were never again able to make a major discovery. Now this would be inexplicable if being a great scientist consisted in being an expert in scientific method. In other words, Popper is, this is me now, contextually arguing that there is a common misunderstanding of what the scientific method is as some magic automated procedure that you follow and you'll inevitably make scientific discoveries. And he's saying, this is not correct. There is no automatic procedure to make discoveries. The scientific method in this sense does not exist. The real scientific method consists first of a creative conjecture phase that is entirely non-methodical.

[01:38:37]  Bruce: But once you make the conjecture, there is a definitive method to how to go about handling that conjecture in a scientific way. And that is the set of conventions we've been talking about, particularly the no ad hoc rule. You follow that and you quickly find that it's impossible. These conventions such as the no ad hoc rule that you follow, that you should follow Popper's arguing, or you'll quickly find it's impossible to falsify your theories and error correction will grind to a halt. You'll become satisfied with the pure logic you think your conjecture came from, and it will feel not like a conjecture, but like you reason to the truth. This is why in the same lecture, Popper goes on to say, It, meaning the scientific method, exists only in the sense that we can study the ways of criticism. Now, how does this answer our question about how to deal with conjectures?

[01:39:33]  Bruce: Once you realize this, this does, in my opinion, answer the question, how do we come up with new theories in the first place if you can't consider non-testable theories due to the no ad hoc Rule? The short answer is you're allowed to make your own personal risk to pursue any conjecture you want for any reason you want. But that is completely different from having reason to a correct solution. On the contrary, during the conjecture phase, there is really no reason for anyone to take your conjecture seriously until you've made it independently testable to prove that it is not an ad hoc theory. I've mentioned before that in my disagreements with the people I've talked to, I've used a lot, obviously, Saudia, Brett, and Dennis as people I've disagreed with over this, that in many cases they have reason to something they feel quite certain about.

[01:40:25]  Bruce: They don't know how to turn their theories into independently testable theories, but it just really feels like they've reasoned to something that's true. And then they can't seem to understand when I say something like, yes, that's an interesting conjecture, feel free to keep pursuing it. Let me know if you find a way to make it testable. And then otherwise I don't show much else interest in the theory. It's because this is their personal conjecture. It's not yet been fixed or turned into an independently testable theory. So it's not yet a theory that people should start adopting as if it's a best theory. To them, it feels like they found really good criticisms of the prevailing theory. Church-Turing-Deutsch thesis for Saadia, G-Factor for Brett, animal zombies for Dennis, and they think they've reasoned to the only possible alternative theory, so they can't understand why I'm not embracing their theories as if they're true.

[01:41:24]  Bruce: But the simple truth is that we often feel this way about our conjectures. We self-identify with our conjectures, and we tend to accidentally become dogmatic about them. But the correct, fallible position is neither to reject these conjectures, nor embrace them as correct, but instead to ask the person to make them independently testable. To summarize, then, Popper's Ratchet is one of the least talked about parts of Popper's epistemology. Most people don't even realize it is a crucial part of his epistemology. Even people that have extensively studied Popper's writings don't seem to have any knowledge of it, or if they do, they see it as optional. Popper's Ratchet means you are only allowed to solve problems by making your theories more precise and increasing the empirical content of your theories, never by making them vaguer, adding degrees of freedom, or reducing the empirical content of our theories.

[01:42:17]  Bruce: I say our theories here because, of course, it may be that you're increasing the empirical content of an auxiliary theory, not the theory proper, but either is fine. It's the overall level of empirical content that goes up. It doesn't have to be the theory itself. It may be the theory itself, but it doesn't have to be.

[01:42:33]  Peter: Okay, so to summarize in my own way, we should just be, always be trying to make our theories more empirical.

[01:42:42]  Bruce: Yes.

[01:42:42]  Peter: That's kind of the gist of it. That's where the direction, the ratchets should be turning.

[01:42:48]  Bruce: That's correct.

[01:42:50]  Peter: And is it just science, or are we talking about moral theories, political theories?

[01:42:55]  Bruce: No, that's a good question. And it's obviously going to be outside the bounds of this podcast episode for me to answer it.

[01:43:02]  Peter: Yeah.

[01:43:02]  Bruce: And so I've mentioned this several times, and I've been building up to this, where I've said, what's the difference between a bad explanation and a good metaphysical theory? And of course, I asked you that, and you couldn't answer that question, right? And I think most people couldn't answer that question because they've never thought to ask it, okay? In some sense, that's the question we need to answer, okay? That there must be some sort of objective difference between a bad explanation and a metaphysical theory, or a philosophical theory, okay? What is that difference? So let me just say at this point that that is a question that's a little harder to answer. And so what I'm presenting today would be for any theory that is trying to explain real life consequences, okay? So it should apply to Brett's Theory of Intelligence. For Brett to declare his theory philosophical or metaphysical is cheating because he is telling us something about how intelligence in the brain actually works.

[01:44:08]  Bruce: Therefore, we have every expectation for him to turn it into an empirical theory. But there are other theories out there where that wouldn't be true. For example, Popper's Theory itself. There's no way to do an experiment to determine which epistemology is the right epistemology. Therefore, it's unclear how you could ever turn Popper's Theory into an empirical theory, at least in the strict sense, right? Because of that, we see an immediate difference there. It wasn't Brett himself, but when the defenders of Brett's theory, I would give them counter examples and they would declare it a philosophical theory. At that point, you knew they were cheating, right? You knew that they were dogmatically defending their theory rather than taking it seriously as a theory. And the reason why is because the theory itself is trying to tell you something about the world in a way that absolutely could be testable, therefore needs to be testable.

[01:45:06]  Bruce: And this is going to be as much as I can answer right now. I think that I can give you a deeper answer, but it's going to require a totally separate podcast.

[01:45:15]  Peter: So you're saying they're they're they pivot to the assertion that empirical ideas don't don't matter to metaphysical theories.

[01:45:25]  Bruce: That's what they pivot to.

[01:45:26]  Peter: Yeah.

[01:45:27]  Bruce: So it gets used as a way of trying to immunize what should have been an empirical scientific theory. You simply declare it to be philosophical. I again, this wasn't the only time this has happened. Before I ever even started to bring it up on the podcast, I had had numerous crit rats suddenly tell me this is a philosophical theory when I would bring up counter examples to the theory.

[01:45:51]  Peter: But, you know, at the same time, I'm not sure if I'd want to apply Popper's Ratchet to everything in life. That might be a bit neurotic or something. I mean, does my does my dog love me?

[01:46:06]  Bruce: So, you know, I think we should do an episode specifically about what are the limits of Popper's Ratchet or more maybe more more like how would you adapt Popper's Ratchet to apply to a non empirical theory? And I think there's a there's some really like incomplete but really good progress that can be made by trying to answer that question. And I even got some ideas I want to like suggest, but we don't have time today. I'll have to do that in a future podcast. So even a question like, does my dog love me? The first question you would have to ask is, is there some reason why you definitively need an answer to this question? Is it important to you? Is it important that you know if it's true or not? If it is, then yeah, we probably ought to be applying Popper's Ratchet to it in some way. And now what does that mean in this context? Well, I'm not sure.

[01:47:02]  Bruce: We'd have to now stop and think about it. Okay, how would you apply the same concept as Popper's Ratchet to a metaphysical theory? And that's the question we would want to now ask, because obviously it can't be that the amount of empirical content goes up because metaphysical theories by definition don't have empirical content. So it creates a really interesting question that I actually think is worth asking and trying to dig into. Let me just say, though, that I think a lot of these questions, like, does my dog love me? Maybe I don't even care what the answer is. For all intents and purposes, I want to believe my dog loves me. I don't own a dog, by the way. But my sister owns a dog and it's the family dog, so that's my dog. And I believe my dog loves me.

[01:47:51]  Peter: So there is some validity to the idea that if it feels good, let's just go with that.

[01:47:57]  Bruce: There is. It's not really a pressing question in any other way. If I were to apply Popper's Ratchet, I believe I would come to the same conclusion, by the way. I do believe that the theory that my dog loves me, if I get serious about it as an empirical theory, that it can be turned... We've already mentioned this at a high level in other podcasts, like Animal Grief, things like that. They've actually attempted to test these things. And they've come up with clever ways to test it. And so far, the theory that your dog loves you is the best corroborated theory, right? Even if you do take it seriously as a scientific theory, it has so far survived corroboration, attempts to falsify it. So I think it is a best theory empirically. It's an example of where, if you actually really needed to know, go read what the science says on the subject, and that's probably what you would want to do, to find out if your dog loves you or not.

[01:48:52]  Bruce: And you'll find that right now, we think that your dog does love you. That's what the current theories say, right? Okay. And then you could talk about how good are these theories? How empirical are these theories? Are they strong theories? And I think the answer is no, they're not, right? And that's kind of what the state of the current theory is at this point, that they're moderately testable, but not strongly testable yet. So in a future podcast, let's discuss the key, let's discuss folk epistemology. I raised this in a past podcast in the episode on the problem of open-endedness. I would claim that by default, human beings have a certain epistemology that they follow, that I would call folk epistemology, and that you just naturally, intuitively do it. So one of the key questions we want to ask is, what's the difference between folk epistemology and critical rationalism?

[01:49:48]  Bruce: I'm going to argue that the key difference is Popper's Ratchet, or the no ad hoc rule. Popper's Ratchet being a consequence of the no ad hoc rule. So because folk epistemology isn't the same as dogmatism, but folk epistemology, its main problem is that it allows for dogmatism. It's an epistemology that can get to the truth, but can also get bogged down in falsehoods. Whereas Popper's Ratchet, Popper's epistemology doesn't. So I'm going to argue that that's why I'm going to argue that Popper's Ratchet is a key difference between being dogmatic and being rational. Now let me remind you though also being dogmatic is not all bad. Episodes 51 and 52 we talked about cases where being dogmatic was actually quite helpful to society. That we've got these dogmatic people out there that force us to really think about certain problems. And think about vague theories about how to solve those problems that then eventually get turned into something that's stronger, right?

[01:50:52]  Bruce: That's more empirical. I'm also going to argue that Popper's Ratchet is what it really means to reason. That is to reason is to severely test our ideas no matter how much we believe them personally. In fact, you might even, to kind of borrow language from Deborah Mayo here, you might take a phrase like reasoning as severe testing. That to reason is to severely test.

[01:51:18]  Peter: But Bruce, this has been another wonderful conversation, at least to me. And I'm still not bored with listening to you talk. I hope our audience feels the same way. That was nice. And I hope you have a wonderful day.

[01:51:38]  Unidentified speaker: All right.

[01:51:39]  Bruce: Well, yes. Thank you, Peter. The Theory of Anything podcast could use your help. We have a small but loyal audience, and we'd like to get the word out about the podcast to others so others can enjoy it as well. To the best of our knowledge, we're the only podcast that covers all four strands of David Deutsch's philosophy as well as other interesting subjects. If you're enjoying this podcast, please give us a five-star rating on Apple podcasts. This can usually be done right inside your podcast player, or you can Google The Theory of Anything podcast Apple or something like that. Some players have their own rating system, and giving us a five-star rating on any rating system would be helpful. If you enjoy a particular episode, please consider tweeting about us or linking to us on Facebook or other social media to help get the word out. If you are interested in financially supporting the podcast, we have two ways to do that.

[01:52:29]  Bruce: The first is via our podcast host site, Anchor. Just go to anchor.fm slash four dash strands, F-O-U-R dash S-T-R-A-N-D-S. There's a support button available that allows you to do reoccurring donations. If you want to make a one-time donation, go to our blog, which is four strands dot org. There is a donation button there that uses PayPal. Thank you.