Episode 61: A Critical Rationalist Defense of Corroboration

• Links to this episode: Spotify / Apple Podcasts
• This transcript was generated with AI using PodcastTranscriptor.
• Unofficial AI-generated transcripts. These may contain mistakes. Please check against the actual podcast.
• Speakers are denoted as color names.

Transcript

[00:00:11]  Blue: Welcome to the third of anything podcast. Hey, Peter. Hey, Bruce. How you doing? Good. So this is a topic that I’ve been wanting to bring up for a while because I’ve had so many people ask me about or argue with me about it, I should say. So it’s corroboration. What is corroboration? How does it fit into Karl Popper’s epistemology? And how do we look at this little piece of positivism, or some people feel that way about it, that’s been inserted into the middle of an otherwise negativist epistemology. Is it wrong? Was he just trying to hold on to positivism a little too much? What’s going on? Why does he make such a big deal about corroboration? Peter, you said that you’ve actually did a little bit of research with chat GPT on this. Why don’t you tell us what you found?

[00:01:05]  Red: Perhaps no big surprises, but I was just curious about what chat GPT would say. And I asked it what Karl Popper thought about corroboration and the difference between justificationism and verificationism and corroboration and to try to get that all real straight in my mind. The chat GPT definitely says that Karl Popper, unlike most philosophers at the time and maybe sense who emphasized verificationism, whereas you verify theories as being true, which I think leads to justificationism, the idea that there are certain theories or ideas that are just justified true belief that that’s true. Popper I think was pushed back against that. At least according to chat GPT, it never says he’s completely against corroboration. He does say it’s an important part of his philosophy, even though the language I think it emphasizes falsification. So we sort of, as I understand it, stress test our theories. It sounds like what you were kind of getting at in our chat before is that perhaps some Popperians are more against corroboration than Karl Popper was even. Is that what I’m understanding? Yes. So

[00:02:52]  Blue: Deutschians in particular seem to have a fairly negative view of it. And I ended up arguing with a number of them over this. And I thought, you know, I need to like put my thoughts together better, put this into a podcast, get the thought out there and then let people criticize it and talk about it. So this

[00:03:15]  Red: could be seen as a Deutschian defense of corroboration.

[00:03:19]  Blue: Yes.

[00:03:20]  Red: Okay.

[00:03:20]  Blue: It’s often seen as a negative as something that I’ll give you an actual quote. So I was actually, what actually led to the idea for this is I had a conversation with Oniket on Twitter and we were arguing about corroboration. And at first like a lot of Deutschians haven’t read the logic of scientific discovery. So initially I thought he was just not aware that Karl Popper had made corroboration a huge part of his theory. So I tried to emphasize that and he came back with, you know, that’s just authoritarianism who cares whether Karl Popper did or didn’t, what he did or didn’t say about corroboration. What I want to know is how does it actually fit into a negativist epistemology? Well, that’s a completely fair question. I mean, the mere fact that Karl Popper said, yes, we need corroboration does not mean we need corroboration. You need an actual explanation as to what corroboration is, why you need it. And I have to confess that I don’t feel like when I’ve read Karl Popper’s books, particularly the logic of scientific discovery, that he spends chapters on this stuff. And even after reading chapters, I’m not sure I could tell you definitively how Karl Popper sees it fitting into his epistemology. So why is corroboration a part of critical nationalism? Why did Popper write several chapters on this? Why did he say it was very important that we test our theories? How do you take this, this idea of corroboration, let’s be honest, corroboration is exactly the same as saying we have a positive instance. We did a test and the test came out positive.

[00:05:10]  Blue: Okay, so from a certain point of view, if you’re looking at the epistemology purely as negativist, you could say, well, we actually learned nothing, right? You really only learn things by having a counter example, having a refutation, as Popper would put it, but as I would put it a counter example, a positive instance should teach you nothing or so goes the argument. Okay.

[00:05:30]  Red: That just doesn’t seem quite right to me the way you put

[00:05:32]  Unknown: it.

[00:05:32]  Red: Well,

[00:05:32]  Blue: it doesn’t seem quite right to me either, right? And that’s why I’ve kind of pushed back on this. And yet I can see what they’re saying. And there’s clearly a fair question being asked here, right? So here is Brett Hall on corroboration. He actually wrote something about it. I think it was on your Facebook site, the Many Worlds of David Deutsch. And we were talking about corroboration. And Brett says, this is a long quote, but I think it’s a good one. I’d ask Popper, were he alive? What problems does use of this concept term word corroboration solve? He wrote about it in Logic of Scientific Discovery, then he doubled down and went further in Objective Knowledge. That’s a book, Objective Knowledge. There’s an entire chapter in there as there is in his much earlier work about corroborating and comparing it to confirmation. I think all of this is a little misguided on his part, just as his overemphasis on probability and logic is too. By the way, I want to call out that statement. We’re not going to discuss that in this podcast, but I think that that statement is problematic. But I know why he’s saying that. It’s also a product of his time. It’s all a product of his time, so to speak. He was engaged in these discussions with the Vienna Circle, who all had an entirely different notion of, well, one would dare say everything in philosophy. So I think he’s actually, in truth, trying to be generous to them and concede some stuff. But today, I don’t think Popperians need the word corroboration. But anyone who wants to use it, I don’t mind.

[00:07:05]  Blue: It’s just that I think we can get away without it, like the word probably, at least in a strict formal sense. But whatever the case, the whole discussion for me collapses on Popper’s own terms when he wrote about explicitly and which I think is far more central to his philosophy than corroboration is. And that central thesis is that debating the meanings of words is not philosophy. It solves no problems. It’s puzzling over what some string of characters arbitrarily designates. Just ask what problem in science is being solved here. Like when Mercury fails to appear, where it should, given Newton’s theory, does this confirmation confirm, corroborate, pick your random string? General relativity? No. It does not do anything of the sort to general relativity. That whole way of thinking is a chimera in Popperian critical rationalist actual philosophy of scientific terms. What we instead say is Newton’s theory is falsified. And so are all other known rivals, of which there are none. The only viable explanation of gravity, and hence the position of planets or moving celestial bodies anywhere anytime, is general relativity done, problem solved. Implication, we don’t need the word corroboration. But he’s not here, so we can’t ask him. I can only guess it’s all the fault of the people he was having debates with. And so he was trying to conjure the language to translate his jarringly different worldview into something that resembled a picture that could begin, they could begin comprehending. But I think we’ve moved on or should have in any rate. So clearly here we’ve got a good example of, and again, does anybody doubt Brett’s Deutchian credentials, right? He’s the biggest, most important podcaster in the Deutchian community.

[00:08:53]  Blue: And I think he’s voicing what a lot of Deutchians feel. They just sort of feel like it just doesn’t make sense to have this concept of corroboration in a negativist philosophy. We don’t need it, we can describe things without it.

[00:09:09]  Red: So I gotta say, I’m already realizing what an interesting issue this is. Truthfully, I was not aware there was such a stark sort of divide between the more hardcore Deutchians and what Hopper apparently said that this is this already dried me into this conversation.

[00:09:33]  Blue: All right, that’s what we’re hoping for. We want to take difficult philosophical questions, and we want to try to make progress against them. So I think here would be my steelman summary of what I think Brett is trying to say. This is my best attempt to make it as strong an argument as I possibly can. Refutations are possible, but verification is literally impossible. Therefore, you can never in any way verify or confirm a theory. There are no positive arguments in favor of a theory, only negative arguments for competing theories. Once a theory has even a single counter observation, it is falsified. Therefore, it doesn’t make sense that a theory could improve or strengthen just because it survived tests that might have refuted it. So test do not change the nature or quality of the theory itself. Such tests do not change the nature or quality of the theory itself. So it seems that Hopper’s corroboration really brings nothing to the table and it’s a relic of his times. I think that is the argument he’s trying to make. And I hope you can see I’m sincerely trying to do it the justice that I think it deserves here. So here’s my thought on that. First, if I’m taking as a given that we can never support, verify or strengthen a theory, then I would agree it’s hard to see how corroboration could matter. If that is my starting stance that it is impossible to support a theory in any sense, it’s impossible to strengthen a theory in any sense, then yes, I guess I would have to say it really doesn’t make sense to have a concept of corroboration. I also want to say that I feel like Brett made a good case here. His

[00:11:13]  Blue: theory, the way he’s wording things comes across quite strong for me. And in fact, but for the fact that I can easily refute it by counter example, I’d probably buy his argument because I think it’s a strong sounding intuitively appealing argument that really kind of pumps my intuitions in the direction that that he’s trying to pump them. So furthermore, I think that the reason why it’s so appealing is because he’s making a fair point that needs some careful consideration. That is to say, while he’s actually technically wrong, I’m going to argue, he’s partially right. There’s very similitude to his argument. So I think there isn’t going to be an answer to this question that’s going to be straightforward, which is why we’re now going to spend the next three hours discussing it because I think it needs some careful nuance. Furthermore, I make no promises that my answer is the right answer, right? This is just me trying to look at this problem and trying to understand it as best I can and saying this is what I’ve come up with so far. I would love other people’s feedback on this. So cameo is not here, but I actually was going to lay this out for cameo. And I know what she’s going to say because we’ve worked together. But let me let me make an example here. Okay, so let’s say we have a release of software and I work in software. She works in software. I work as a developer now. Thank goodness. I used to be a project manager. And so I’ve kind of been on different sides of the keyboard, so to speak, when it comes to software.

[00:12:42]  Blue: And so it’s a big deal to release software out to production because things can go wrong. So in a certain sense, you might say that the software that’s out in production is our current best theory of known theory of how to implement our desired theory of what the software should do. Okay. And let’s say it has a known bug. So we know that software is, if you’re looking at it as a kind of theory as to what it should be doing, we know it’s wrong in some way. We have a definitive example, counter example, where we know it produces wrong results. Okay. Hopefully you’re seeing the analogy here with science. I actually think software is a very, very, very Paparian field. And we can learn a lot about software by applying Karl Popper’s epistemology to software.

[00:13:35]  Red: I like the premise that software is a kind of theory. I mean, that’s kind of what you’re saying, right? It rings true.

[00:13:43]  Blue: Yes. Okay. So now I have a developer and the developer tracks down the bug and explains to me as the project manager, or maybe it’s me. I’m the developer. I track down the bug. And I explained to myself, okay, here is the mistake in the code that caused the bug. And then I explained how I fixed it or how the developer explains to me how they fixed it. Okay. So they say, here is what was wrong. And here is what we’re going to now do. So I’ve now, I’ve now received or have created an explanation for how this change is better than the one that’s in production. Okay. So this is now the old one in production is a falsified refuted theory. And the new one I’ve created is not yet known to be falsified. And I have an explanation for why it’s going to do a better job. Okay. So in paparian terms, you have a current implementation or theory that is known to be false. And you also have a very good solid explanation of what the correct theory should be. So here’s my question for you. Do you need to test the developer’s changes before releasing to production? If so, why? Now, Camille was here. I can tell you what she would say. She’d say, that’s the stupidest question I’ve ever heard. You absolutely never, and I mean, never, and I mean, never let a developer put their code into production without testing. And developers think they can and they want to, if they have a really good explanation in their mind, this is what was wrong. And I’ve now fixed it. They will try to put things into production without testing.

[00:15:16]  Blue: And you have to, as the project manager, slap them hard for doing it. Okay. To really get them to realize, you’re going to be in a lot of trouble if you ever do that again, because it is one of the single most dangerous things you can ever do in software. Now, why is that? Okay. So you might from, if I was holding on to Brett’s explanation that we just read, what possible good could it do to cooperate that it works when you already have a good explanation for why it does? Okay. This is the essence of Brett’s argument intact, but now applied to the domain of software in a domain where it’s well known it’s wrong, that his argument is absolutely known to be wrong. Okay. That you do not do this in software. Okay. Now, why is that? It’s actually a little bit hard to pinpoint at first what’s going on. In fact, it’s so hard. That’s why so many developers sometimes try to make a bug fix in production is because it feels Brett’s arguments in their mind in a nascent form. And it just really feels like, I know what the bug is. I know how to fix it. I’ve got an explanation. I just don’t need to test this. And so they will try to go fix that bug and they’ll try to go put it there and then everything blows up. Sometimes it works, but too often it just blows up in your face. And usually in spectacular ways. Okay. So now, why is that? What’s going on here? Okay.

[00:16:48]  Blue: Why is it that even when I have a good explanation of why this bug fixes the right bug fix, that I still have to corroborate it before I’m going to be willing to put it into production? Why is explanation alone not enough in a case like software? Now, you might argue here that science is different than software. Now, I don’t think that’s true, but I’m going to accept that argument because I’d rather not argue with it. And to be perfectly honest, I don’t need that because I can actually give you an equivalent counter example using science itself.

[00:17:21]  Red: So I guess that’s kind of where my mind was going a little bit that science is maybe concerned with, I guess, maybe sort of deeper explanations than, you know, just whether or not the software program works or not. I don’t know. That’s what I got.

[00:17:36]  Blue: Fair argument, right? I can’t prove to you that these are equivalent cases.

[00:17:42]  Unknown: So

[00:17:42]  Blue: I’m going to now give you what I believe is an equivalent case in science itself. Okay. In fact, I’m going to use the same case that Brett uses. So imagine it is the first day Einstein finished his theory and published it. So the Eddington expedition where they tested his theory and corroborated it hasn’t taken place yet. Now, according to Brett’s argument, we don’t need the editing expedition or at least that’s a completely viable way to read his argument. So I’m going to put it out there as a problem with his argument. So let me explain why I feel that follows naturally from his argument. So we have the perihelion of Mercury. That’s an observation that falsifies Einstein’s theory. We also have the Michelson -Morley experiment without the speed of light. Now, if you’ll recall, they actually were trying to test Ether. It had nothing to do with trying to find a problem with Newton’s theory. And it was really only later that in retrospect, they figured out, oh, actually, this, this is an example of where Einstein’s theory solves this problem and Newton’s theory gets the wrong prediction. But so we have these two well known observations that already exist at this point in time. The point in time is Einstein just finished and published his theory, but it’s not yet corroborated. So from a certain point of view, you might say that before even a single test has been performed, Newton’s theory was already refuted by existing data and Einstein was already the best theory. Since if you really looked into it, you would have found that Einstein’s theory made the right predictions about Mercury, that’s Brett’s point, and also the right predictions about the speed of light in the Michelson -Morley experiment. I

[00:19:25]  Blue: don’t know how to pronounce Michelson or if it’s Michelson, but anyhow, in that experiment. So following Brett’s argument, there’s this implication now that just becomes inescapable if you’re accepting his argument, which is there is never going to be a need to test new theories in science. The whole reason why you come up with a new theory is because you’ve got some sort of counter example that you can’t explain. So there’s always going to be that the moment you make a new theory like Einstein’s, before you corroborate it, it’s already the best theory, according to the point of view that Brett is expressing. Okay, now, is that true? Well, no, it’s not. I mean, if you look at at least how scientists looked at this in real life, they certainly did not see it that way. To them, they thought that Einstein’s theory was wrong. It was very controversial theory. In fact, it was so controversial that he didn’t win the Nobel Prize for it. He won the Nobel Prize for the photoelectric effect, because his general relativity was considered to controversial by most scientists. So the science, so the group of scientists that were out there, they certainly, even though they knew that it could take these existing points of data that Newton’s theory couldn’t properly explain, and they knew that Einstein’s theory could explain those, why didn’t they see it as the best theory? And were they just wrong? So now I could imagine a Deutchean arguing here, that’s exactly correct. So long as old data fits the new theory better, it is already the best theory and its status will not change by doing a new test. This again seems to flow naturally from what Brett was expressing, okay?

[00:21:14]  Blue: So you might get an argument something like this, oh, those silly scientists, oh, but if they only understood true Popperian epistemology, they spent years refusing to see that the moment Einstein’s theory was created, it was already the best theory. But they foolishly instead ignored his theory until the editing expedition took place. Only then did they realize it was the best theory, but that was unnecessary. We already knew it just using old data, okay? Now, let’s talk about this argument for a second. Again, I have done my best to steelman it, okay, to try to lay it out there. Is this a good argument? And if it’s not, try to explain why it’s not. It’s a little hard. It gets you to a really uncomfortable place where that you basically never need to test scientific theories, okay? Clearly that must be wrong. Just by counter example, notice how this is exactly the same as with the software now, okay? I have a clear counter example in science now to Brett’s theory. It clearly must be wrong. We have refuted the theory, but it’s not clear how to solve the problem that’s being presented.

[00:22:23]  Red: I mean, in some ways, I do have a sense that we’re kind of like nitpicking over words in a way. I mean, what is the difference between trying to falsify something and corroborate it? I mean, in a real practical level, they can be seen as maybe sort of two sides of the same coin.

[00:22:45]  Blue: So let me take their point of view for a second. Okay, I’m going to know this isn’t my point of view, but I’m going to try to channel their point of view, okay? So actually, I’m going to argue with you, Peter, that that’s not true. That yes, it’s a test in that sense. They’re both a test. You don’t actually falsify or corroborate. You do the test, and then one of two things happens. Either the test comes out negative and you learn that that theory was false and falsified, or the test comes out positive and you don’t really learn anything. If it’s just a single theory, that’s all there is. So from that point of view, it seems like all the knowledge should come from falsifications and you shouldn’t gain any knowledge from corroborations. So why do we need the concept of corroboration in Popper’s epistemology? Okay, am I convinced you enough that we can go on to admit a problem?

[00:23:36]  Red: Yes,

[00:23:37]  Blue: this is, it’s a difficult problem. Okay, I’m going to still argue that this whole point of view is still just wrong. Okay, although I’m admitting that there’s some truth to it. It’s got some various millitude. I believe that scientists continually prove again and again that they are better at Popper’s epistemology than any Popperians I’ve ever met. And I’ve brought this up in past episodes where I’ve said, you know what, I would always bet on the scientists. When Popperians say, you know, oh, those silly scientists, they just don’t understand Popper’s epistemology. Occasionally, they’re right. But the vast majority of time the scientists have it ridens the Popperians who have misunderstood Popper’s epistemology.

[00:24:16]  Red: I think that’s partly why I find Popperian epistemology so convincing in a way is that at least it’s an attempt, I think a pretty realistic attempt to explain what is actually happening in science. I think that’s the right way to look at it.

[00:24:35]  Blue: When you’re trying to use Popperian epistemology to attack science as a whole, you’re on shaky ground. Now, so that there is something to be learned by learning Popper’s epistemology. But in a certain very real sense, science as an institution has institutionalized Popper’s epistemology in a way that makes more sense than most critical rationalists understand Popper’s epistemology. Popper was right. He was trying to figure out what is it they’re doing that works. And he did a pretty good job of pulling that out. But he’s a little bit hard to understand sometimes. And you often come away with misunderstandings of Popper that really you think you understand Popper’s epistemology, but you don’t really.

[00:25:19]  Red: So maybe you could say that the average scientist inexplicitly understands quite a bit about Popperian epistemology. They’re experts in it. Explicitly, maybe not so much.

[00:25:33]  Blue: That’s right. Now, they may explain what they’re doing in terms of induction. They may explain what they’re doing in terms of Bayesianism. But what they’re really doing is critical rationalism. No matter what label, they slap on it. OK, I also think this is why the fact that critical rationalists, people who see themselves more as philosophers and not scientists, I think the reason why this happens, the fact that this happens, is why we often find Popperians promoting some really bad explanations online. Right. I mean, we’ve talked about this on the show. Animal feelings never did an episode on that, but we’ve kind of talked about it. Genes can’t affect you. All differences in learning are due to interest, and that’s it. I mean, a lot of these, if you really sit down and you look at them, they’re pretty bad scientific theories, right? Really, really pretty bad, like no empirical content and they’re being immunized. How are we, you know, how is that happening? Well, my argument is it’s because scientists understand Popper’s epistemology quite well, although inexplicitly. OK. And they did before Popper existed, right? Popper was learning from how they actually did things and trying to work that out. He was using them as the prototype. OK. So I’m going to argue that scientists were from a Popperian epistemological standpoint were correct to not assume Newton’s theory was refuted due to the two counter examples that existed at the time. Parahalion of Mercury and the Mickelson -Morley experiment. I’m going to say that they were correct to not pay much heed to Einstein’s theory until it passed a corroborating test. And then only then they were right to only then realize that the two previous counter examples were in fact, refutations of Newton’s theory.

[00:27:19]  Blue: And that’s what happened in real life, by the way, they did not perceive the two existing refutations as refutations until Einstein’s theory passed the Eddington expedition. Famously, Thomas Kuhn claims that the way science works is that you have a paradigm and scientists try to through a sort of orthodoxy protect the paradigm until anomalies grow too much and a crisis takes place. And then you suddenly get these radicals on the edge that start to support a new theory and then slowly it takes over. And he tries to explain the whole thing through through social rules, right? He doesn’t quite explain it in terms of realism and things like that, right? So this is where Kuhn gets a ton, right? But this is where he really kind of goes wrong. This is a really good counter example to Kuhn. There was no crisis over Newton’s theory prior to the corroborating incident of Eddington’s expedition. Then there was a crisis. And then the crisis played out for years afterwards, which is why Einstein did not win the Nobel Prize or solving the greatest mysteries in the universe that we had at the time, right? So and I’m going to argue that they were right to do that, okay? That it actually makes sense from within a Popperian epistemology standpoint that they refused to accept that the counter examples were in fact refutations of Newton’s theory and that they were correct to not pay too much heed to Einstein’s theory until a corroborating instance came across. So now why am I saying that? Now obviously this is going to be hard for me to explain, which is why we’re going to spend a bunch of time with me quoting people and trying to explain this.

[00:28:59]  Blue: But I really think I’ve got a decent argument here. Okay. So now here is Deutsch on the Deumkwein thesis. Okay, he makes a distinction between a thesis and the problem. He agrees the thesis, he disagrees with the problem. But he says, in any experiment designed to test a scientific theory T, the prediction of the result expected under T also depends on other theories, background knowledge, including explanations of what the preparations of the experiment achieve. How the apparatus works and the sources of error. Nothing about the unmet expectations, if you get a counter example, dictates whether T, the theory or any of those background knowledge assumptions was at fault. Therefore, there is no such thing as an experimental result, logically contradicting theory T, nor logically entailing a different credence for T. But as I’ve said, an apparent failure of T’s prediction is merely a problem. So seeking an alternative to T is merely one possible approach to solving it. When we, via arguments or experiment, find an apparent flaw, conflict or inadequacy in our theories, that constitutes a scientific problem. And the theories are problematic, but not necessarily refuted. Now I want to remind people here that I pointed out back in the Popper Without Refutation episodes that Deutsch uses the word refutation differently than Popper. That’s very important to try to make sense of what Deutsch is saying here versus what Popper is saying. Okay, but let me continue. In this view of a scientific theory, a scientific theory is refuted if it is not a good explanation, but also, but has a rival that is a good explanation with the same or more explicanda.

[00:30:40]  Blue: So another consequence is that in the absence of a good rival explanation, an explanatory theory cannot be refuted by experiment. At most it can be made problematic. If only one explanation is known, then an experimental result makes it problematic that can motivate a research program to replace it or to replace some other theory. But so can a theoretical problem, a philosophy problem, a hunch, a wish, or anything. But in any case, the existence of a problem with a theory has little import besides, as I said, informing research programs unless both the new and the old explicanda are well explained by a rival theory. In that case, the problem becomes grounds for considering the problematic theory tentatively refuted. Now, if you’ve been listening to my podcast, this is all not new to you. In fact, I think I’ve actually made this same quote before because I feel like this is Deutsch it is very best, right? He’s really saying something meaningful here that people need to understand in terms of epistemology. He’s explaining why a single observation can never refute a theory. Now, of course, what he means by refute a theory is the theory itself is refuted. To popper a refutation is just a counter example. He doesn’t care whether it refutes the theory or the background knowledge. In his mind, it’s refuting the combination of the theory and the background knowledge. And that’s what he meant by refutation. So both men have it right. They’re just using the word refutation in different ways. Very important that we understand this, right? If you’re trying to hold refutation to have a single meaning, it’s impossible to make sense of one of the two men.

[00:32:11]  Blue: And you’re going to claim that one of them is confused because you have no choice because they’re using the term refutation in different ways. This is by, by the way, if you need a primer on this, this is episodes 41 and 42 from this podcast where I talk about popper without refutation. And I give all the necessary quotes to prove that what I’m saying is correct, that Deutsch and popper do, in fact, use the term refutation in different ways. So because of this, the fact that you can’t actually refute the theory proper with a single observation, scientists are right to not see counter examples as, quote, refuting the theory, where here we mean refute the theory proper until a new theory exists to replace it. Before then, there is too much danger that it’s just a problem with the background knowledge. And of course, that was exactly what scientists were thinking at the time. They had this Nicholson -Morley experiment. They had the perihelion of Mercury. But it was really hard for them to imagine that the problem was with Newton’s theory. It seemed so much more plausible that the problem was with the background knowledge. In fact, as I mentioned, the whole Jupiter or maybe it was Neptune, I always get confused which planet it was. We had almost something identical to the perihelion of Mercury where the orbit wasn’t matching. And it just turned out that we had missed a planet. There was a body of gravity we didn’t know about. So the background knowledge was wrong. There was just no particular reason at this point to believe that the perihelion of Mercury, which remember that is Brett’s example, was actually a refutation at all. This is just how science actually works.

[00:33:46]  Blue: So here’s the problem. It’s too easy to come up with a theory that explains a known counter example. It’s very, very hard to come up with a theory that explains a previously unknown counter example. I’m going to say that again. This is super important to understanding why corroboration is so important. It’s easy, very easy to come up with a theory that explains a known counter example. It’s very, very hard to come up with a theory that explains a previously unknown counter example. If this sounds like easy to varianess versus hard to varianess, you’re right. That’s exactly what we’re talking about here. An explanation that explains the data better and has these two known counter examples, perihelion of Mercury and Michelson -Morley experiment. Einstein’s theory prior to being corroborated by Eddington’s expedition, its real actual status was it might have just been a theory that was made up, that was meant to solve these two problems, known problems. But that’s trivial to come up with. It’s if scientists immediately declared every theory that could do that a best theory, you would have a constant stream of better theories that are better than the current paradox. Okay. So scientists are therefore right to wait for a corroboration of a theory before taking it too seriously. At least as a matter of practicalness, that’s just good practice. Okay. Even in this case, where in retrospect, we know Einstein’s theory was more correct than Newton’s, they can’t have known that at the time. Okay. So the strict pulperian answer might be, so let’s take this question now. Can you support a theory with evidence? Okay. A strict pulperian answer to that question is almost always no because it’s impossible to support any theory. So

[00:35:48]  Blue: in fact, Deutsch specifically says that. He says an important consequence of this exploratory concept of science is that experimental results consist within theory do not constitute support for T. He goes on to say that refutation exists tentatively, but support is non -existent in his paper. Here’s another quote from Kenneth Allen Hopf, H -O -P -F from 2012 that I came across somewhere. Allegedly, every fact is logically consistent with infinitely many mutual incompatible theories. But if a fact is compatible with two different theories and no theories are and those theories are contrary to each other, how does it even make sense to say that the fact supports one of them and not the other? Yet, if you say it supports both, you’re admitting that support points in mutually exclusive directions because the two theories cannot possibly be both true. As Miller and Popper say, support, though it may certainly exist, points in all directions at once and therefore points usefully in no direction. This is kind of the, this is a really well put way of how most pulperians will attack the concept of support. Okay. Now that’s a really good quote and I admit he’s making a really good argument, one that you absolutely should be paying attention to. And what they’re really getting at here is the idea that whereas a refutation definitively points against a theory, a support doesn’t really definitively point in favor of a theory. A support points to any theory that hasn’t been refuted equally or so he’s claiming. Okay.

[00:37:25]  Blue: So then Kenneth goes on to say, based on that, he says the upshot is that whole fields in square, square quotes of what people currently call science are simply academic shams that create the appearance of science for the public. So the public will cost, will continue to cough up tax dollars to support the careers of academics who constantly extol the virtues of the state. That’s the world we actually live in. Because this is me now scientists are a bunch of fools that don’t know epistemology. I’m going to argue that he did really well up until he said that that when he went on to try to use the fact that support can point in multiple directions to try to declare whole fields of science as shams that he was reaching beyond what his point actually allows. Are scientists really the fools that paparians sometimes seem to think they are when they talk about supporting evidence? My answer is no, absolutely not. No, they may be. Now I’m not saying they’re never fools, right? Individual scientists may be fools. In a lot of ways I’m looking at science as a whole, as an institution. And maybe we look at the best of scientists as our exemplars. So how do we square this case? How can there be such a thing as support for theory via evidence if Popper was entirely a negative epistemology? Well, it turns out it’s actually quite easy. Here is a quote from Thomas Kuhn that I think helps here. Counter instances can at best help to create a crisis for or more accurately to reinforce one that is already very much in existence. By themselves they cannot and will not falsify a philosophical theory.

[00:39:13]  Blue: Okay, that’s from the structure of scientific revolutions page 78. Then he goes on after saying that to say, but Karl Popper’s falsification, though it surely occurs, does not happen with or simply because of the emergence of an anomaly or falsifying instance. Instead it is a subsequent and separate process that might equally well be called verification. I would here substitute the word support since it consists in the triumph of a new paradigm over the old one. That’s page 147 of the structure of scientific revolutions. Kuhn calls this a joint support. I’m going to use the word support instead of verification because I think the forward verification is problematic here. A joint support falsification process. Now I want to point out that what Kuhn is talking about here, if you’re looking at what Deutsch just said, he’s correct. He’s maybe not using terms you’re comfortable with like support. Okay, but he’s not saying anything that’s actually epistemologically wrong. Okay, it’s identical to Deutsch’s view. You have to have two theories that are in comparison and the way you actually make a determination in science is you’ve got these two or more competing theories and you look at the evidence and if you got two that evidence, let’s say that we’re talking about Newton versus general relativity, you go and you do the editing expedition and as Brett puts, it falsifies Newton’s theory. That’s how a paparian wants to put it. Well, why not also call that supports the theory? Now, if you’re assuming the word support means I’ve justified this theory as true then clearly that would be wrong but there’s no reason why the word support should mean that. Like really, there’s no reason at all why the word support shouldn’t mean that.

[00:40:57]  Blue: It might just mean I’m in a contest, I’m doing a crucial test and Einstein’s theory made this prediction that Newton’s didn’t, that none of us even knew to try to make because the explanation didn’t make us think about the possibility of stars moving during an eclipse. Okay, so we went out, we did the experiment and we come back and we say it supports Einstein’s theory. Is that really epistemologically wrong? Only if you start with the assumption that support means justified as true but if you’re not starting with that assumption, if you only start with the assumption that the context here is a single test, a contest between two theories, then I don’t see how the word support is any different than the word falsify in this instance. That was the point you originally made now, wasn’t it? I made an argument against you but you were actually correct, Peter. And this is, I think, one of the big things that gets missed is that sometimes we’re talking about doing experiments on a theory and we have no competitor and sometimes we’re talking about a crucial test. If we’re talking about a crucial test, the language of support’s completely valid because you’re just simply talking about which of two theories is the better theory given that the outcome of this experiment. There’s no actual assumption that this theory is now absolutely declared certified as true and most scientists would tell you that they’re not assuming that. So I don’t see any particular problem with the language of support if it’s used in this kind of context where we’re talking about a crucial test. So now at this point, I know a lot of people push back as they just dislike the word support.

[00:42:33]  Blue: Okay, but I’m going to just say, hey, if I decide I’m going to define the word support to mean that we’re in a theory -to -theory comparison, we’re doing an experiment, and it came out in favor and corroborated that theory and by implication refuted the other theory and that’s all I mean by the word support. There is no way to say I’m wrong because what I’m saying, what I’m using the word support as a label to mean is identical to the Popperian -Negativist philosophy at this point.

[00:43:03]  Red: I like how at least chat GPT put it which I assume is somewhat accurate to how Popper thought of it where it said he emphasized that falsification. He didn’t say this is how it works and you have to falsify and corroboration is completely wrong. It was kind of more, well, this is maybe a better way to think of it but both concepts have some validity. I think that just rings true to me for how it actually works in real life.

[00:43:43]  Blue: Yes, yes. I have a quote from Popper and I forgot to put it into my notes but there’s a quote from Popper where he says every single theory that can be falsified, the inverse theory can be verified and vice versa. Now this may sound terribly wrong to your average critical rationalists on Twitter today because they have such a strong reaction to the word verify and to the word support. I call this the Popperian war on words but Popper really didn’t have a problem with this. He understood that there’s nothing special about the word verify versus refute. If you go look at what he actually said in the logic of scientific discovery and again go look at my episodes 41 and 42 where I give the quotes in detail. What he’s really trying to say is it’s better to think of science in terms of its universal laws that we have to refute because those are the ones that are going to allow us to do experiments. It’s not true that you can never verify any theory in any sense. If I word the theory right, and I’m going to get to this in a second with some examples, then I can verify theories and Popper admits this throughout the book, Logic of Scientific Discovery. This is not something that I’m making up. I’m getting this from Popper and this is one of the things that really changed me is when I read the Logic of Scientific Discovery I suddenly realized that the Twitter, Deutsche and version of critical rationalism had misunderstood significant parts of Popper’s epistemology.

[00:45:14]  Blue: So at least amongst the fans, I have found that Deutsche seems to come much, much closer to what Popper actually said than I think a lot of his fans have been able to figure out. So let me give a real life example. Here’s a real study where they say at the beginning we found support for the domain level prediction that domestic skills, attractiveness, and aspects of women’s sexual strategy would be more central to women’s status than men’s status. Okay, so this is obviously a paper that’s about what is effective for women in terms of dating versus for men. Okay, so that they’re saying that domestic skills and attractiveness, and which sexual strategy they use, what do they mean by that is kind of the Madonna horror thing that you probably hear about. I don’t want to go into that because it’s a bit sexist, but they’re saying that those strategies matter more to women than they do to men. Now, when I say that, almost every man and almost every woman I know goes, well, duh, right, this is, I mean, we have to keep in mind and sometimes people make fun of science because it’s telling us what we already knew, but a lot of times we have conflicting ideas. We say birds of a feather flock together, or we say opposites attract, and we’ve got these folk wisdoms that are at odds with each other. So science’s job is to go figure out which of the two folk wisdoms is correct. And then when it figures out which one’s correct, then we go, oh, that was obvious, right? And so the fact that this may seem obvious doesn’t really make it an uninteresting scientific study.

[00:46:56]  Blue: I would also probably note David Deutsch’s criticisms of these sorts of studies. You do have to understand that what we’re measuring is in the context of a certain culture, and that a change of the culture may change the outcomes of this study. I don’t think that makes it an invalid scientific study to be perfectly honest, and I dislike it that a lot of Deutschians claim that it is. Yeah. So let’s look at just the wording here. Okay, I know a lot of Deutschians in particular, not just Deutschians, a lot of modern paparians would really have a cow over the fact that they said we found support. Okay, and they would say immediately, oh no, there is no such thing as support for a theory. That is impossible. Okay, but let me ask, what’s really being said? Okay, is there really a misunderstanding of critical rationalism that they happen to use the word support? Now it seems to me that what they mean in context, and I think you just always have to look at, what’s the actual context? Okay, the context here is that there’s only two logically possible explanations, and they’re trying to choose between those two logically possible explanations. Either these strategies are more effective for women than men, or they’re not. It’s got to be one of the two, right? We’re intentionally narrowing the study up. There might be way more complicated theories we could get into. How much and under what circumstances does attractiveness matter, or something like that? Okay, but it makes sense that for this study, they just want to know over a population, on average, does this strategy work better for women than for men? Does it matter more that women are attractive than men are attractive?

[00:48:40]  Blue: Okay, to use one of their concrete examples. Given that we are in this context only choosing to test between those two competing theories, it does not matter that we call it support, because given that there’s only two to choose between, the word support means that we’re feeding the other. That’s what it means, right?

[00:49:03]  Red: Yeah.

[00:49:03]  Blue: Okay, and this is exactly what Kuhn said. So Kuhn is on, got this, right? He is describing something that is consistent with how science actually operates. So science’s goal is a crucial test between theories. It takes a lot to get there, admittedly, but that’s kind of the gold standard, okay? So it’s done as a theory -to -theory comparison. This is central to understanding Popper’s epistemology, that it is about theory -to -theory comparisons, for the most part, a few exceptions. Epistemology is so wide -ranging that you can never really say it’s only about this, right? Because he reapplies it in all sorts of different circumstances. So they make a choice to limit the number of theories via a level of generalization or abstraction. And then they choose the experiment such that it will choose between the two theories. Given that context, it just does not matter whether we call this refutation or support. It doesn’t. I have some problems with calling it verification, because at least to me, the word verification has a stronger implication. It means that we now know this theory is better than its alternative. Now, maybe in this case, that would be okay. When we talk about, like, say, Newton versus general relativity, we can tell by looking at the equations that general relativity always makes the same predictions as Newton’s, except in the places where general relativity makes better predictions. But that doesn’t necessarily have to be the case. If I’m doing an experiment that tests between the light wave theory of light and the particle theory of light, and such experiments actually existed in real life, I might come up with an experiment that supports one theory and then a different experiment that supports the other theory.

[00:50:56]  Blue: Now, of course, what that actually means is that both theories are refuted, because the word support in this context really just means that I was trying to choose between the two theories based on a single experiment. So I got some concerns with calling that verification, right, because you’re not really verifying the theory. So I would maybe take more the Deuchy and modern Popperian side that if the word verified is being used, I guess I would probably give the benefit of the doubt to the scientists and say, well, they probably actually meant support. But I would admit at that point maybe that wasn’t the best choice of words. Because it may imply something that they didn’t actually mean. But the word support, I think, is pretty innocuous. Now, let me just point out that they could have said to make the Popperians happy, we found a refutation for the domain level prediction that domestic skills attracting us in aspects of women’s sexual strategy would be the same or less central to women’s status than men’s. Okay, sure, now you’re putting things in terms of refutation. But man, it sure is awkward now. Okay, and honestly, it’s not an improvement. It’s not even easier to understand what they mean at this point. And the reason why is because in context there was only two possible theories to begin with. So a refutation of one of the two logically implies support in this context for the other logically. And it’s just easier to put it in terms of a positive outcome in favor of one of the two theories. I would put something else here. Because they chose carefully what the level of abstraction was, there was really only two possible outcomes.

[00:52:41]  Blue: Absolutely one of them had to be more right than the other at the level of analysis they were at. That just logically, and of course this was a choice for how they set up the experiment, etc. Okay, so they did find support from one of the two possible theories. It does no good here to say, oh, but there might be a better theory where they get more specific about why a woman’s attractiveness matters more so than a man. Yes, of course that’s true, but that doesn’t make the slightest bit of difference to the question currently being posed, which did not ask about that. That’s for a future study. So on the one hand, I’m not really denying what Kenneth Allen Hoppe said that support points in multiple directions at once. I’m just pointing out that in context where scientists use this word, there’s only two options. So it can only point in one direction because they’ve designed the question that way. Okay, so that would mean that support can exist if you set up the experiment and the question correctly. Is it too much of a stretch to say that in this question between falsification and verification that he might really be describing an attitude more than anything specific? So a scientist should bring an attitude of wanting to falsify their theories rather than verify them, but it really doesn’t… It’s more of a qualitative statement about the kinds of tests that a scientist would do than something that more quantitative. Okay, let me just say that Popper without a doubt multiple times makes it very clear that in his mind, critical rationalism isn’t so much a methodology as an attitude. I actually slightly disagree with him.

[00:54:37]  Blue: I think that there is a methodology to critical rationalism, not to how you conjecture things but how you go about testing things. But that’s a discussion for another time and let me just say that he’s not entirely wrong either. It sort of depends on what you mean. From that point of view, I think Popper would strongly agree with you that he was talking about an attitude. However, I do think there’s more to it than that. If you go back and you look at Popper without refutation, episodes 41 and 42, Popper clearly had something logical in mind when he talked about refutations. It was really that science had to do experiments and experiments required that you have spatiotemporal limitations. Otherwise, the experiment becomes impossible. Whatever the effect has to happen within a reasonable amount of time, it has to happen within a space that you can actually search. So he realized that just as a matter of practicality, you will only be doing experiments in science, at least, on the universal laws, the constraints that science places on reality. And because of that, he said scientists will concentrate on falsifying their theories, by which he really meant was fighting counter examples. And so on the one hand, yes, you’re completely right. Popper, without a doubt, makes this about a scientist’s attitude. But I do think there’s a deeper reason why he emphasized refutation and falsification than merely that attitude. Now, let me point out, let’s take Kenneth Allen Hopp’s thing about support component multiple directions at once.

[00:56:23]  Blue: Now, I admit that if you have a whole bunch of theories, he was giving the example that maybe you have two theories that all the evidence fits, and then if that’s true, then the outcome of an experiment would support both theories, which means support points towards both theories at the same time. It doesn’t point in a single direction. Now, I just gave you an example where that’s not true because it just so happens that the experiment was set up to answer a question where they had intentionally made it a binary. Now, here’s the question I’ve got for you. What’s the general case for science? Well, here’s what Deutsch says about this. He says, although there are always countless logically consistent options for which theories, for which theory to reject, the number of good explanations, ones that are hard to vary, known for an hexical tandem 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, it is one or zero. Now, this is from page 11 of the Logic of Experimental Test, which is where I’m quoting Deutsch for this podcast, for those who are interested. I want to take that what he just said seriously. He’s saying, and I’m going to make this very clear now, that it is extremely rare that we have three explanations we’re testing between. It could happen. It has happened, but the normal case is zero or one. And if we’re really, really, really lucky, two.

[00:57:53]  Blue: Well, now, if that’s true, if you take what Deutsch is saying seriously, then again, it’s just not true that support points in multiple directions at once, because at most of the time, in terms of actual good explanations that you have available to you, you’ve got at most two. So it doesn’t matter whether you call it support or refutation, if you’re inside of a crucial test, because support points in exactly one direction and refutation points in exactly one direction, because you’ve only got two. Okay. In this case, I’m not saying we’ve set it up that way. I’m saying it’s just so hard to come up with good explanations as competitors that you’re just lucky if you’ve even got two to do a crucial test between. Now, if there were three, and if you came out with a test and it refuted one and supported quote the other two, okay, I’ll grant you calling it support at that point seems a little weird. Okay. I mean, I know what a person’s if they say it, I know what they mean. So I’m not saying they’re wrong, but it would make it would be so much easier to say, well, we managed to refute one of one of the three theories. Okay. But that surely isn’t the normal case. The normal case is we’re testing between two and exactly two theories, according to Deutsch. Okay. So if good explanations are really hard to come by, then realistically, the language support shouldn’t be something that we’ve got such a cow over. Saying the editing expedition supported Einstein’s theory is every bit as meaningful as saying the editing expedition refuted Newtonian physics.

[00:59:31]  Blue: Now, this is going to be one I know I’m going to get pushed back on because of the Popperian war on words. They just don’t like the word support, but I feel like I’ve made a strong case here. In fact, I feel like I’ve made a case that’s fatal to their whole point, right? That it just does not matter in the case of testing two theories, whether we say support or refutation, I know what you mean. And the only sense in which you can say it’s wrong is if you insist that the word support has some sort of essential meaning that implies justification, but I don’t think that’s normal for scientists to think, act or talk that way. So I don’t think there’s any reason to believe that’s the normal meaning of the word support within scientific circles. The other thing that you have to keep in mind is that you just can’t consider theories that don’t exist yet. And in Popperian epistemology, you don’t. You ignore theories that don’t exist yet. So yes, it might be that there’s a theory of gravity that’s even better than Einstein’s theory. And so when that theory comes along, we’ll say, oh, Eddington’s expedition didn’t really support Einstein’s theory because it’s wrong. Of course, that’s not what we meant at the time, right? That’s a misunderstanding of what we meant at the time. And yes, Eddington’s expedition will as much quote support this new theory of quantum gravity because it’ll make the same prediction only better, right? But it seems to me that at this point, we’re really just jumping outside the context of how the word’s actually used.

[01:00:53]  Blue: And really, we’re imposing an essential meaning on it that just isn’t consistent with how people actually use the term. So I will admit, though, that sometimes the words get used wrong. Now, this is kind of my point, though. If you pay attention, at least I feel like I’ve paid attention, it seems to me that the word support gets used in a way consistent with a Popper’s epistemology in the vast majority of cases. And so I have some concern. That’s why I push back on Poparians that will jump down a person’s throat over the use of the word support is because I just don’t believe that if you actually pay attention that scientists in general get this wrong. But they do sometimes get it wrong. You will come across scientists, individual scientists, who when they say this explanation supports this theory, they actually do are trying to imply, we have proven this theory beyond doubt. That does happen. And when that individual case happens, when I’m talking to a scientist or interacting with them, and that’s what they claim, I will at that point trot out Popper’s epistemology and point out that there is no such thing as the term support. If by support you mean justification. You can never certify something is true. And I will at that point attack their use of the word support. But I’m very hesitant to do that until I have good reason to believe that they’re using it in that way. I will try to initially assume the best that they don’t mean it in that way. And I will ask questions to clarify whether they mean it in that way or not.

[01:02:31]  Blue: And what I’ve generally found is that scientists know you can’t prove a theory as true, right? So there’s generally not a problem.

[01:02:39]  Red: Yeah, I can. I think that’s why the Popperian framework just rings so true is that I really think in the real world, if you crest most scientists on this, even if they’ve ostensibly are justificationists, or they’re probably going to readily concede that point. They

[01:02:59]  Blue: will, usually, usually. And I think you’re exactly right, which is why I feel like scientists really are great Popperians, even those that know nothing about Popper’s epistemology. Now, let me ask the question, can you verify a theory? A moment ago, I kind of implied you couldn’t. Even that’s not strictly true. It sort of depends on what you mean by theory in this context. Let me give you an example. And this comes from Popper, by the way. I’m not making this up. In fact, you have to understand that I’m not really that smart. I don’t make up hardly anything, right? Or at least I don’t believe I do. I think a lot of the seemingly heterodoxual ideas that I bring up on this podcast are actually, as far as I can tell, straight out of Popper, right? That this is just what he said. It’s just that a lot of cultural context has made it a little hard to understand what he was saying, and I’m trying to recover that, if that makes any sense. So, okay, so if I were to say, can you verify a theory as correct, I think the standard preparing answer is no, that’s impossible. But again, they aren’t quite correct. So Popper himself gives a clever example of this. He gives the theory that hafnium exists, that the atom hafnium exists. So this is a theory of legitimate interest to science. Science actually does want to know if hafnium exists or not. This is why Popper used this example. And it was verified that hafnium exists according to Popper. How is it verified by finding it, right? They did experiments, they found hafnium atoms, and they verified that the theory hafnium exists was a true theory. Okay. Now,

[01:04:37]  Blue: online critical rationalists often fall into what I call the absolute verification fallacy here. Back in episodes 41 and 42, Popper thought refutation. I define this fallacy as the idea that the asymmetry between refutation and verification is because the word verification always means verified with certainty, whereas the word refutation always means refuted tentatively. Since one can never be certain, verification is impossible, but refutation is possible. Well, now this is just silly, okay? Because even if that were true, which it’s not, if somebody says the word verified, they almost never mean verified with certainty. Okay. If you actually stop and ask them, okay? But even if it were true, then Popper’s epistemology, we could just simply introduce a new use of the term verified called tentative verification. And we could now use the word verification just fine in Popper’s epistemology. If it had been the fact that that was what the asymmetry between verification and falsification was, but it’s not. And Popper makes this extensively clear in his book, The Logic of Scientific Discovery. And again, the real reason was that science cares about doing experiments, and that requires spatiotemporal constraints. And therefore, you’re actually going to end up testing the constraints of that theory sets, which means you’re going to be looking for counter examples. That’s really what Popper said. Now, when you take the example of hafnium exists, what Popper actually had in mind was more about the spatiotemporal nature of experiments. So if what we care about is if hafnium exists or not, of course, you have to verify that theory by actually finding hafnium rather than refuting it because it is being worded as an existential theory. And existential theories are sometimes of interest to science.

[01:06:31]  Blue: Now, Popper, though, he uses this example, and here’s what he argues. He argues that in real life, we may call it the theory that hafnium exists, but really, it is a theory full of a lot more content and constraints than that. If it wasn’t full of constraints, it would be literally impossible to do an experiment to verify the existence of hafnium. Hafnium, the theory of hafnium contains constraints and tells us something about hafnium that allows us to narrow that experiment to something reasonable. So yes, maybe here we could be more technically correct to say we refuted the theory that hafnium doesn’t exist, but that is so dang awkward. And if what I really care about at the moment is does it exist or not, there’s just nothing wrong with wording it in terms of verification. So this is a special case. Admittedly, this is a special case. Popper is intentionally picking it as a potential counter -example to his theory, then trying to show how to integrate it into his theory. This is typical of Popper. He tries to go as hard as he can against his own theory. But what it proves is that verification is a part of science and even a part of empirical science, though it’s more a matter of convenience of language and only in some rare circumstances like this. So this is still compatible with Popper’s refutation -based epistemology. And here’s the reason why, because every verification, if you invert it, is a refutation and vice versa. Remember, I mentioned that Popper said that, right? So in a certain sense, it doesn’t matter if you word it as a verification or refutation. It’s particularly in a case like this where we’re just wondering, does hafnium exist?

[01:08:22]  Blue: It’s okay to word it either way because logically they’re just inverses of each other. Now, that’s not always true. And this is something that Deutsch brings out. It is true that sometimes explanations and their inverses and their inverse is also an explanation. If I’m trying to explain if oranges stops curvy or not, the inverse of that theory is as much an explanation as the theory that oranges stops curvy. But that’s not true for like Newton’s theory, right? And this is what Deutsch brings out, is that Newton’s theory is something so complicated that the inverse of Newton’s theory is not Einstein’s theory. In fact, what the inverse of Newton’s theory really is, is it’s the set of cases that would theoretically be counter examples to Newton’s theory. So because of this, it sort of depends on what the question is we’re asking. In some cases, it may be that the question we’re asking that it just simply does not matter if we word it as a verification or a refutation. But there’s other cases where it does matter. Oh, here’s the quote. I did put it in here. This is from Popper, Realism in the Name of Science, page 182. Whenever we falsify a statement, we therefore automatically verify its negation, where the falsification of a statement can always be interpreted as the verification of its negation. So there it is. Popper really just didn’t have a problem. Also, something else that gets missed. Popper did use the word verification to talk about experiments because an experiment is something you verify. Now, it’s what you’re doing is you’re verifying the outcome of the experiment. It’s phrased as an existential statement that such and such is going to happen and either will or won’t happen.

[01:10:05]  Blue: So you verify the outcome. So Popper would say that verification is a very big part of science. It’s just that it’s how we do experiments. It’s not something we do to the universal laws of theory. Okay, so let’s review. Yes, you can definitely, without a doubt, support a theory through evidence if you’re talking about a crucial test. You just have to take the word support here to mean, in a crucial test, the evidence came out in favor of one theory and against the other, which is a very reasonable way to understand the word support. And I would argue, is what scientists mean by it? Okay, in general. And yes, you can sometimes verify a theory through evidence, though this is more of a linguistic choice. We verified the existence of hafnium by refuting its non -existence. It doesn’t matter which way you phrase it because of the nature of the question we’re asking. Popperians tend to act as if scientists are making extreme claims about evidence -proving theories to be true. It’s not really the scientists that made those claims. It’s the philosophers that did. This is one of the reasons why I’m so much more comfortable with scientists than philosophers. Philosophers make a lot of really, really stupid claims. And scientists are kind of a more pragmatic bunch. Now, there’s exceptions. You’ve got philosopher scientists out there. Some of them are really good, like Deutsch. Some of them are really bad. And so you will find scientists who make some really stupid claims. Of course you would, right? But the language scientists used can be easily, in most cases, map back to Popper’s epistemology of refutation.

[01:11:42]  Blue: And to me, that’s all that really matters, is can I understand this in terms of the correct epistemology? And most of the time I can, so I’m not going to argue with their language. So let’s say a scientist claims that due to some evidence, a new theory is now stronger. Now, you will hear scientists say that, right? The issue here is that to most critical rationalists, you can’t strengthen a theory. A theory is just a theory. And a positive outcome of the theory does not change it in any way, which is true, right? So you cannot strengthen that theory, so argues the critical rationalist, okay? Now, if by strengthen what you mean is, is that the theory somehow changed its nature because of this experiment, then I agree with the critical rationalists online that you cannot strengthen a theory. But here’s what I really want to ask them. Is it necessarily the case that a scientist who claims that some evidence made a theory stronger was claiming that the essential nature of that theory changed or were they saying something that could legitimately be mapped back to Popper’s epistemology, just like the word support can be? So I’m going to argue that the word strengthen can in some really important instances be mapped back to Popper’s epistemology without a problem. So let’s first discuss a relatively uncontroversial example. Now, considering how strongly I’ve had one online critical rationalist after another pushback on me over the idea that you can strengthen the theory, this example I’m about to give you is actually pretty uncontroversial, even for the ones that people have pushed back at me, okay? Here it is. Do we accept Darwinian evolution today? Yes or no, Peter?

[01:13:28]  Red: Unless you want to nitpick about Neo -Darwinism or Lamarcusism or something, sure. That’s exactly the right thing to say. Okay.

[01:13:37]  Blue: So here’s the problem. What did I mean by Darwinian evolution? Did I mean the theory Darwin lays out and has original book, Origin of the Species and Nothing Else? Or wouldn’t it be technically corrected? So let’s say I didn’t mean that. Wouldn’t it be technically correct to say Darwin’s original theory has long been replaced by modern synthesis and Neo -Darwinian evolution that incorporates DNA, genetics, replicators, things like that? From this point of view, we might say Darwin’s theory and refer to it as such, but we really mean the modern theory that has a successor to Darwin’s theory. Now, I’m right at the moment having an argument with Brett on your Facebook page over whether or not it matters, which one we call it. He’s arguing that it does matter and I’m arguing that it doesn’t. I don’t really care if you want to say Darwin’s original theory is refuted and we’ve got a new theory that’s a successor to it that explains the success of Darwin’s original theory. Or if you want to just say, oh, it’s the same theory, but the theory has gotten stronger. I don’t care which way you phrase it. Now, if somebody says, oh, Darwin’s theory has not been refuted, which is how most people would think of it, right? Instead, they would say it’s gotten stronger. Well, what do they mean by that? Today, we may call it Darwin’s theory of evolution, but it’s really a new theory that is far stronger than Darwin’s original theory because the theory has in fact changed. It changed when we integrated DNA into it. It changed into something new when we integrated replicators into it with neo -Darwinian evolution. So what does stronger mean in this context?

[01:15:20]  Blue: Well, the whole argument that you can’t make a theory stronger was based on the assumption that a test doesn’t change the nature of a theory. In this case, we’re explicitly changing the nature of the theory. So there’s nothing wrong with calling it stronger. What stronger would mean in this context is that there’s more empirical content. Darwin’s original theory was barely testable at all. The modern version of Darwin’s theory is extremely testable. It has all sorts of implications, particularly in DNA that are very testable. And this is really what we mean by, this is one possible way we can look at the word stronger that Paparians really don’t have a problem with, that if we mean the theory has become enhanced and it’s now a more empirical theory, then by definition it’s now stronger. There are more ways to refute it. There are more ways to test it. So it’s a stronger theory at this point. If it continues to survive refutation, in other words, be corroborated, then without a doubt, the modern theory of Darwin’s theory is just a stronger theory than the original Darwin’s theory. So scientific theories can and do get stronger over time, by which we might just mean become more detailed and contain more empirical content. So the reason why I’m bringing this up, because obviously this isn’t what when people, when Paparians online say you can’t strengthen a theory, of course what they really mean is unless you change it. And yes, you can strengthen the theory if you change it. Okay, and they’ll all admit that. I’ve never had anyone argue that with me.

[01:16:57]  Blue: Okay, my key point here is let’s not get reactionary towards the word stronger, because there’s at least one sense in which the word stronger can be applied to a theory. If the theory has in fact gained in the amount of empirical content that it has. So remember that testability equals empirical content. Here’s a quote from Popper in Logic of Scientific Discovery. I admit a system as empirical or scientific, only if it is capable of being tested by experience. This is why Popper stressed testability so much in his epistemology, because a theory is only scientific if it is testable in the first place. And the theory is a stronger theory if it becomes more testable. If we add to the theory and it enhances how much empirical content is available to it. And it becomes easier to test. It becomes more ways in which you could theoretically refute it. It is why Popper cared so much about the boundary between empirical scientific theories and metaphysical theories. Here is Popper in the Myth of the Framework, page 88. This criterion of demarcation between empirical and non -empirical theories, I have also called the criteria of falsifiability or the criterion of refutability. Okay, notice that in Popper’s mind demarcation, which most online critical rationalists today say that this is not even central to Popper’s theory, Popper’s side as exactly identical to the concept of falsifiability or refutability. It does not imply that irrefutable theories are false, nor does it imply that they are meaningless, but it does imply that as long as we cannot describe what a possible refutation of a certain theory would look like, that theory may be regarded as lying outside the field of empirical science.

[01:18:46]  Blue: Now, we just did an episode kind of recently, a few episodes back, about the importance of the empirical nature of science. What makes empirical theories special was the name of the episode. This is really where I took this idea from. Again, while I may be expressing what today is seen as heterodoxical ideas within critical rationalism, they’re literally just things that Popper just said in his books. I’m not doing anything fancy here. You cannot claim that demarcation criteria of Popper’s is not central to his epistemology, but then claim that refutability is. At least you can’t make that claim and still understand what Popper meant by those two things, because to him, they are the same thing. So Darwin’s theory of evolution got stronger over time because it went from being barely, if at all empirically testable, to a strongly empirically testable theory, as we learn more. Now, it isn’t hard to see why we prefer a more testable theory over a less testable theory. For all of the pushback I’ve received over cooperation from online critical rationalists, none of them have ever argued to me that we shouldn’t make our theories more testable and more empirical. They may sometimes argue they have a better theory that’s not empirical, which I think is pretty much always wrong when they say that, with maybe a few caveat exceptions. But I think most of them would agree, look, if I had two theories that made more or less the same predictions and one of them was more empirical, I would prefer the more empirical theory as long as it’s not refuted. The example I used was orange’s stop scurvy versus vitamin C stop scurvy.

[01:20:24]  Blue: That’s a more empirical theory because there’s more ways to test it and more ways in which it can be refuted. And then ultimately, vitamin C causes tissue to grow in this sort of way. And here’s the mechanism. And now you’ve got a really highly empirical theory, much stronger theory, as we would use the term. So

[01:20:42]  Red: wait, let me see if I can summarize this. So in the popular mindset, which I think is quite frustrating to Popherians, Popper was saying that only, they think that Popper was saying that only falsifiable theories are important, which is kind of like, since he was arguing with positivists, people who thought only scientific or empirical theories are important, is kind of the opposite of what he thought, which is why it’s, so I think it’s frustrating to Popherians. But then it sounds like what you’re saying is that some Popherians believe that the demarcation criteria between a testable or falsifiable theory and a non -falsifiable theory wasn’t really that important, an important part of his philosophy. Whereas what you’re saying is that it actually, it is important, a little more important than maybe some of the, okay, it’s central, that we should be moving our theories into empirical, trying to make them as empirical as we can.

[01:21:54]  Blue: So let’s go back to what you said about Popper’s epistemology could be, in a certain sense, thought of as an attitude. If you have the attitude, I am trying to move my theories towards being empirical so that it’s easier to come up with experiments and you’re being a paparion. If you’re moving your theories away from the empirical so that it’s harder to refute your theories than you’re doing the opposite of critical rationalism.

[01:22:20]  Red: Okay,

[01:22:20]  Blue: and that’s really the correct way to understand Popper’s epistemology. This is why empiricalness is central and the demarcation is central to Popper’s epistemology because testability is central to Popper’s epistemology. Now, there is, by the way, I don’t know if you remember this, but in the past episode, you actually asked about, you said, I’m not sure if the boundary condition is central to Popper’s epistemology. And at the time I said, well, I actually think it is. And you said, oh, that’s interesting. So now I finally got around to explaining to you why I feel it’s actually central to his epistemology. And of course, clearly he believed it was central. I gave you quotes so you know that this is what he said. He believed it was central to his epistemology. So now there is a direct relationship between testability and Deutsch’s easy to vary criteria. Or I guess I should say criterion. I hate that. I just want to use the word criteria. I’m a layman. I just don’t care. I think

[01:23:14]  Red: that’s the right attitude.

[01:23:16]  Blue: Yeah. So I’ve kind of come to realize that Deutsch’s easy to vary criteria is actually somewhat problematic. And that’s going to have to be a different podcast because I do not have the time or strength to go into that at this point. But I do feel like it was getting to something that was correct, even if it was expressed in a way that ultimately seems to cause a lot of confusion. So Deutsch claims that the easy to vary criteria precedes testability. So because of that, trying to equate testability with easy to vary criteria, if I were to say, well, easy to vary criteria is the same as testability, people will immediately quote Deutsch at me and say, no, he says it precedes testability. And yes, I agree that if you can just easily tweak a theory to fit any test, then there is no point in testing it. And if that’s what Deutsch meant, and I believe that is what Deutsch meant, then I agree with him. But honestly, that is sort of true for every single theory if you aren’t going to accept Popper’s convention of not ad hoc saving your theories. And many Deutschians online today don’t accept Popper’s convention of not ad hoc saving their theories. So there’s a certain sense in which every theory is easy to vary if you have the wrong attitude towards the truth getting back to is Popper’s epistemology an attitude? Yes, it is more than that, but it is also an attitude. And yet despite this fact, so maybe that’s a micro version of why I have some concerns with the easy to vary criteria and why I feel it has led to confusion.

[01:24:55]  Blue: But I do think the easy to vary criteria clearly ties to testability in an important way that a theory that explains more than its rivals should always do so in a testable way, which is equivalent to saying a way where the theory says something about reality. And I think the easy to vary criteria is that as you try to move your theories in that direction, they naturally say something about reality and therefore become testable. So as you make your theory hard to vary, you are moving towards the goal of being testable. Even if it does proceed testability, it’s deeply tied still to testability. And likewise, if you’re going to have the wrong attitude towards your theories, where you’re just going to ad hoc save your theories over and over again, then your theory is no longer really refutable anymore or testable, I should probably say, because no counter examples can exist to it. And this is the same as just saying the theory is easy to vary. So if a theory is not testable because you are saving it through ad hoc saves, that’s really basically the same as saying your version of this theory is easy to vary and therefore it’s not a particularly good theory. It’s a bad explanation. Or at least that’s how I understand what Dwight said in his books. Apparently, everybody disagrees with me, but I really think that’s the correct understanding. Like if that’s not what Dwight said, then I don’t think Dwight is right. Right? I mean, I think if you read Dwight’s the way I just explained, I think he’s right.

[01:26:29]  Blue: I think he exactly matches Popper’s epistemology and it even adds something interesting, the fact that a theory like he uses the example of myths that trying to explain summer versus winter, right? And then he points out that you can easily change the theory to match any outcome. And yet there’s a part of me that still struggles with this because if you really look at the ancient Greeks, they started to criticize the Popper points this out. They started to criticize their theories. And as long as you’ve got the right attitude, it sort of doesn’t matter if the theory is easy to vary or not. As long as you’re willing to let it be refuted by example and then start to realize I need a better theory, then myths are as good a starting place as any in science. Right? I mean, it’s those bad explanations. They’re good starting points for science. And Popper points this out over and over again in his books. And so because of that, well, I accept that a myth is easy to vary and that’s a quality that exists with the myth that doesn’t exist with, say, quantum physics. And therefore, Doge is saying something important here, something really meaningful here. I also would say that even quantum physics is easy to vary if you’ve got the wrong attitude. And likewise, a myth is not so easy to vary if you’ve got the right attitude. So I do think that easy to vary captures something real but in a somewhat awkward way because it intersects with whether you’re prepared to ad hoc save you theory or not. Does that make sense? That’s the miniature of my both criticism and why I also like the easy to vary criteria.

[01:28:07]  Blue: So Popper, he wants to maximize explanation but only as it maximizes testability in the episode about the Deutschian theory of IQ and why empirical theories are special. We talked about how they would make the claim that their theory of IQ was explanatory but they would also make the claim that it could not be tested. Well, that’s like the opposite of what both Popper and Deutsch says we should be doing. That’s what we mean by easy to vary, what we mean by a bad explanation. So Deutsch, and since people are going to now challenge me on this, here’s the quote from Deutsch. An explanation is bad or worse than a rival or variant explanation. Notice it’s a continuum. To the extent that one, it seems not to account for its expocanda or two. It seems to conflict with explanations that are otherwise good or three. It could easily be adapted to account for anything so it explains nothing. Well, that’s the Deutschian theory, online theory of IQ in a nutshell. They’re literally claiming it’s a highly explanatory theory that no experiment can go against. No matter what the outcome of the experiment is, that’s what their theory said. Like that’s literally what Deutsch literally says is what he means by bad explanation.

[01:29:29]  Red: It sounds kind of like an, I don’t know, somewhat like an ad hoc save, honestly.

[01:29:35]  Blue: It is an ad hoc.

[01:29:36]  Red: Maybe a little larger scale, but yeah.

[01:29:39]  Blue: Well, and as I pointed out, their theory is actually more empirical than they think. What they’re actually doing is they’re ad hoc saving it. The actual theory could arguably be considered hard to vary. You just have to decide not to ad hoc save it. Then I gave examples of how you would test it, what those experiments were, what they came out to be, and the theory got refuted. Refuted in this case, not meaning it’s entirely wrong, there’s some verisimilitude, we’ll get to that next, but that there’s something clearly, the theory clearly has a problem that needs to be resolved.

[01:30:12]  Unknown: Okay,

[01:30:12]  Blue: so this theory, this Deutschian theory of IQ that Brett had expressed that we talked about in the previous episode, what was the mistake being made here? Well, he’s trying to move away from the objective nature of science. The objective nature of science is tied up into, as Popper says, its inner subjectivity. Put another way, science is special because it’s the set of theories that we can objectively test because we can do an experiment and everybody gets the same outcome when they do the same experiment. That is, in a nutshell, what defines a scientific empirical theory versus one that isn’t. All right, so based on this then, and now we’re gonna see how this relates to corroboration, corroboration really is just testability plus the test came out positive. So strong theories, according to the way I’m now currently defining it, means more empirical or more testable. Therefore, a corroborated theory must mean it successfully survived tests, which must mean it’s more empirical, which must mean it’s a stronger theory that is not yet refuted. And in fact, this is what corroboration means is that the theory was strong enough to be testable in the first place, and that when we did the test, we were unable to refute it.

[01:31:31]  Red: I’m kind of thinking about the difference between a scientific theory and a philosophical theory though. I mean, when you say that all of our theories should be trying to be more empirical, so more scientific, that makes some sense, I guess. But it makes a sense for certain theories. Yeah, aren’t there a lot of theories that are just in a different category? Yes, there are. You just know that they’re just never gonna get there. So this

[01:31:59]  Blue: one, unfortunately, is beyond the scope of this podcast. So the obvious example here is Hopper’s theory himself, itself, right, critical rationalism. You’re never going to do experiments to try to corroborate critical rationalism.

[01:32:15]  Red: I mean, maybe you could if you looked at, I don’t know, you could think of…

[01:32:21]  Blue: You kind of think of natural experiments, right? Yeah, yeah.

[01:32:24]  Red: Maybe one culture where they don’t believe, where which is this antithetical, the critical rationalism as you can get and say, well, how much progress does this culture make? And then you have one…

[01:32:38]  Blue: I actually think you’re right. So Hopper himself always said his theory was purely metaphysical. And that was why he did not… People would ask him, is your own theory falsifiable? And he’d kind of laugh at him because that meant that he didn’t understand that he was declaring his theory to be metaphysical. And we talked about that in some past episode, somewhere. Here’s the thing though. I’ve also argued Popper wasn’t really right. I mean, yes, that’s what he said. I don’t deny that’s what he said. But you really can offer very natural tests that are counter examples to Popper. And in fact, we’ve done it on this podcast. We’ve said, okay, this is one way, but this is wrong. We have this counter example. Therefore, we’re going to instead understand Popper in this way. And we’ve modified Popper over time, at least in small ways, because we have counter examples to things he said. And isn’t that what we’re doing with Brett’s theory right now? He’s talking about corroboration. It’s not really a testable theory. And no one would expect it to be a testable theory if by testable we mean experiments. So there’s no way to move it towards the realm of doing experiments on it. And yet what we’re doing is, is we’re offering counter examples to it. So I actually think that this is maybe not the way Popper thought. I actually think this is one of the areas where I think Popper was slightly mistaken, that the boundary condition, while it’s important, is actually a fuzzy boundary. And that, like, is math. You can try to find counter examples for math. Are those empirical? Well, no one would claim they were.

[01:34:13]  Blue: And yet they fit Popper’s epistemology just fine as an empirical theory, because you can find counter examples to them. This is something I’ve been meaning to work out and offer up my own kind of slight revision to Popper’s epistemology that I feel solves this problem. But it’s outside the bounds of this podcast, unfortunately, because it’s going to need its own, probably more than one podcast for me to try to work that out and explain what I think I’ve come up with here. So let me just say that you’re right, that there are some theories that will never be testable, and that’s okay. So long as their competitors are equally non -testable. Now, here’s the thing, though. If you’re trying to say something about the nature of reality, then it’s not okay to just try to immunize your theory as non -testable. You really should be having an attitude of, what are the actual implications of this theory? And you should be having an attitude of, how would I go about trying to find counter examples to this theory? And that is the right critical rationalist attitude. Now, maybe the answer does turn out to be, well, for this circumstance, I can’t. So instead, I’m going to rely on criticism. I think everybody agrees that that, in some cases, is just exactly the right answer. Popper talks about it. Deutsch really emphasizes it. Online Deutschians bring it up all the time. And it’s all true, right, that it’s more a matter of attitude. Can you? Should this theory? Is it possible to make this theory testable? If the answer is no, you can’t. Okay, that’s fine. But in the case of the IQ theory, they really are trying to say something meaningful about people.

[01:35:57]  Blue: And it really should have been advanced as a testable theory, and there was a desire not to test it. And that’s really the objection I’ve got here, is that I can declare any bad theory metaphysical. And I’ve literally seen this happen over and over again, that I’ll be discussing with people theories, and I’ll say that’s not a testable theory, and they’ll say I’m declaring it metaphysical. And that’s it. As far as they’re concerned, they’re done. And I just think that’s the wrong attitude, right? Is if I’m going to be allowed to declare any theory metaphysical, and then now you don’t get to test my theory, even though in theory we could have tested the theory, it seems to me like that is literally the opposite of critical rationalism.

[01:36:40]  Red: I guess I am somewhat sympathetic to this attitude that… I’m thinking about psychology and sociology. Yeah, maybe economics is another, but at least with psychology and sociology, the hardcore Deutschians, tell me if I’m wrong, pretty much completely reject these fields. Yeah, they do. And I think, as we’ve talked about at length, I think we both have kind of maybe mixed feelings about that. I really have come to not like psychology and to feel it has… It’s just based on just perceptions, like philosophical errors about what human beings are. But I guess the other side of it is that I can’t quite… Maybe who rejected it is there are elements of I think legitimate science in that field and probably sociology too that aren’t that easy to reject. So I mean, maybe a little bit better attitude is trying to parse out what’s scientific about these theories and what isn’t. But I kind of get the attitude where they just want to say, reject the whole thing too. Yes,

[01:38:04]  Blue: I think psychology is a really good example. Deutsch in one of his talks, he talked about how there’s people who study… I can’t remember what the example he used was, but something in psychology. And he says, this is in the psychology department, but it’s not psychology. Well, psychology is whatever we decided is. I mean, we don’t want to fall into the essentialist error here. So psychology includes some pretty hard sciences. Now, maybe Deutsch would not call that psychology for whatever reason, but it includes things about how we perceive things. And I think it was the perception system that he was talking about. And there are aspects of psychology that are very, very, very deeply empirical and therefore deeply scientific. There’s a reason why we call psychology and sociology soft sciences. That is a very good reason why. It’s because they’re not really strongly empirical theories. There’s some of them are, some parts of psychology are. And here’s the thing though, as we understand the brain better, we are going to turn psychology into a hard science over time. And so the idea that it will forever be a soft science isn’t true either. I mean, imagine if we understood brains so well that we could get right into the software of the brain and recode it to do other things instead. I mean, we could imagine some science fiction -y future where psychology is now a very hard science. Okay, because it’s so well understood. But this is still consistent with Popper’s epistemology. Popper never rejected psychology, sociology, or even history as sciences.

[01:39:37]  Blue: I think what Popper would say is that they’re just low on empirical content at the moment and that they’re trying their best to push towards empirical content as quickly as well as they can. And there’s nothing wrong with that. That would be a completely legitimate thing for them to try to do.

[01:39:55]  Red: Education is a weird kind of, I would say even more extreme example of just a, just seems to be just completely just lacking in any kind of critical rationalist attitude. You know, I’m surrounded by people you hear, oh, well, scientific studies say that you have to give a student nine compliments before every one criticism or something. I mean, something like just a claim. It just sounds just so, it just hasn’t been, you just know that it hasn’t been put to any kind of like critical ration. It hasn’t gotten that, let’s say the Bruce Nielsen critical rationalist treatment, you know, it’s just people just kind of accept it. It is justificationism. They just, oh, that’s a scientific truth. They just did a study. So, but it’s just so, it’s kind of sad. It’s really.

[01:40:52]  Blue: Let me agree with you, but let me actually make a bit of a case against what you just said. If I were to look at pedagogy 20, 30 years ago, and I were to compare it to today, has it

[01:41:06]  Red: improved? I gotta say yes and no, not, I think people’s attitudes towards children have improved. I think school is much less coercive. I mean, if you consider that, you know, and even when we were kids, it would have been very common for a teacher to bring a kid in front of the class and, you know, give him a paddling or something like that. I mean, things are changed tremendously in that sense. I mean, there’s still a lot of bad philosophy in there, I’d say, but yeah.

[01:41:42]  Blue: Let me bring up an example that online Deutschians on Twitter bring up to me nearly constantly. Okay. They would ask me, they’re usually trying to prove some sort of point, they would say, let’s look at education 30 years ago and let’s look at it today. To what degree has education improved because they now realize you can’t just lecture people because there’s no passive learning? Now, I take some exception to this and again, this would require a whole separate podcast to explain why, but you can see they’re kind of right

[01:42:17]  Red: that

[01:42:17]  Blue: we have through the quote science, if I can use that term very loosely here, of pedagogy, come to the realization that active learning is better than passive learning, right?

[01:42:29]  Unknown: And

[01:42:29]  Blue: schools do more active learning now because of that.

[01:42:34]  Red: Yeah. I think that pedagogy has improved because of attitudes about human beings in general and society has improved. It’s not as clear to me that it’s improved through the science following scientific studies. I guess that’s my take.

[01:42:51]  Blue: All right, fair enough. And it’s a little hard to trace there, so I won’t argue with you. But let me just say that perhaps there will someday be a study of pedagogy, science of pedagogy, that is less soft than it is today. And I’m not sure we should necessarily rule that possibility out.

[01:43:09]  Red: But I mean, what you’re really talking about is a science of humans when you get into it. You know, it’s really an uncomplicated equation.

[01:43:19]  Blue: Yes.

[01:43:21]  Red: When you’re talking to people who are so different and have different interests and learning styles and just the more scientific way to think about humans in this field, to me, it just does not ring true on any level. You know, a lot of it is based on behaviorism or Marxism or just weird ideas that just don’t bring true for me at all. So it’s hard for me to see that much validity in it.

[01:43:59]  Blue: All right, fair enough. Okay, so getting back to this now, I just said that a corroborated theory really means that the theory had a certain amount of strength in the sense of it has more empirical content and that it’s not yet refuted. We actually tried to refute it and it’s not yet refuted. This is pretty much just the definition of what it means for a theory to be corroborated. So now one might argue here, I don’t know that I’ve ever seen anyone make this argument, but so this is kind of a straw man I’m putting up just as someone could argue this. One might argue here that we actually care about potential tests rather than actual tests. Remember, tests equals empirical content, right? If you have more empirical content, you have more tests, potential tests. So you might argue that we actually care about potential tests rather than actual tests. Now think about that for a moment and it kind of feels a little silly even just saying it. Okay, but why? What’s going on? Here is now the big secret that I’m going to reveal or at least that I believe I’m going to reveal. I could be wrong, I guess. I want to make a distinction now between the theory itself and our knowledge of the theory. Okay. Can you see that that’s a legitimate distinction?

[01:45:11]  Red: I think so.

[01:45:12]  Blue: Okay. If you had said no, I would have given you the following example. Euclidean geometry, you may know all the axioms, but you don’t know all the implications of the axioms. Okay. So in fact, there is an infinite set of implications. So even if the axioms of Euclidean geometry never change, someone’s going to argue with me here, what about premium geometry? No, just follow my example. I’m just trying to make a point. Even if the axioms of Euclidean geometry never change and so the theory never quote improves in that sense, there is still an infinity of progress that you can continue to make as on Euclidean geometry forever because there’s an infinity of possible implications you still need to work out. This is something that I’ve often argued with, well not often, I have once argued with Sam Kipers on and actually I think he might agree with me at this point. I’m not sure, but he made the claim that if a theory were, if we ever found that one of our theories was true, Popper didn’t, so Deutsch talks about how all our theories are false. He’s talking about scientific theories. If I have a theory that George Washington did or didn’t exist, I’m either right or I’m wrong. So the idea that all our theories are false can’t be correct. But in terms of scientific theories, actual scientific theories, I suspect every single one of our theories in physics and chemistry and biology that all of them are false. Right? So I agree with Deutsch on that. The question we were asking was could there ever be a final theory in physics? Well, there could. Right?

[01:46:45]  Blue: And Sam was arguing to me that, well, if that were to happen, we couldn’t make progress. Well, that’s not true. Even if we had a theory that didn’t change, like Euclidean geometry and its axioms, the content you could continue to make progress against would still be infinite and you would be working that out forever. So you would continue to progress your knowledge in that theory, even if the theory never actually changes again. Having said that, it’s entirely possible that we will never reach an actual true theory in science. In fact, it may even be impossible. Right? I think Deutsch was kind of hinting at that as a possibility. And I could explain to you, using some of Templar’s theories, how that could turn out to be true. But we don’t know, right? It might be possible to have final theories. Also, you can improve final theories in another way. You can integrate them with other theories or you can improve how they’re worded or how well they’re taught. I mean, there’s all sorts of ways that even if you had a final theory, you could continue to progress with it forever. And

[01:47:47]  Red: how are you defining final theory? It seems like there’s a lot of the devils and the details there.

[01:47:53]  Blue: So Euclidean geometry, as a field by itself, it’s a good theory that will never change. It’s true that it may not actually be a reality, but like if studied as a mathematical theory, that theory will never have to change. The axioms will always stay the same and you can continue to study it forever and continue to make progress in the field forever. Another example might be computational theory. It might be computational theory. It could well be that the core of computational theory, which would be the Turing machine, let’s the quantum Turing machine, will in fact turn out to be the final computational machine. It could be that when quantum gravity comes in and when every single physics theory that we have since now, that all of them still ultimately boil down to everything is equivalent to a quantum Turing machine. There would be nothing ridiculous about finding that out. There’s no reason why new physics theories have to produce new computational theories. Now it could be. It could be that we find quantum gravity that computational theory has to change. We don’t know though, but let’s say it didn’t. Then there’s a certain sense in which you can say computational theory is already final and that it’s not going to change again. The class of theories that are P versus NP, our understanding of which ones they go into may change as we start to study and understand the theory better, but in reality, algorithm either is P or NP and there is no other classification and there’s no other computer that can change that. So you could argue rightly that we already have the final theory of computation, if this hypothetical were to take place.

[01:49:29]  Blue: And yet you would continue to make progress in the field forever, understanding it better. So progress and finality don’t seem to actually be connected. Or at least I’m giving you examples of how they aren’t. Whether that actually applies in real life or not, then becomes a separate question.

[01:49:45]  Red: Okay,

[01:49:45]  Blue: but notice though that these examples force us to think about the fact that the theory itself and our knowledge of the theory are two different things. The theory itself for Euclidean geometry is the axioms for computational theory. It’s the quantum Turing machine is the universal computer. And yet our knowledge of that theory can continue to grow exponentially forever because as we start to work out its implications, as we start to discover things that are implications of the theory. So this really shows that there is a distinction to be made between the theory itself and our knowledge of the theory. Okay, you’re with me so far. Because this is probably the most important part of what I’m going to say in terms of me trying to work out my solution to the corroboration problem.

[01:50:37]  Red: Why don’t you just summarize real quick? My mind is coming around here. The theory and the knowledge about that theory.

[01:50:47]  Blue: Okay, so the theory of computation, let’s even narrow it further. The theory of the universal computer that the quantum Turing machine is the universal computer. Let’s say just for the sake of argument that that is a final theory by which I mean that is true and it’s never going to be proven wrong. I can continue to work out implications of that theory forever, discover new algorithms, discover that this algorithm does the same as this one, but this one’s P and this one’s NP. Discover what it means, what the theory implies about what problems are tractable and intractable. Discover what it means in terms of what theories are computable or not. I mean like I could go on making discoveries within this theory forever. Yeah,

[01:51:32]  Red: even within the context of a specific theory, given that maybe any theory really, that we are at the beginning of infinity. There’s an infinity there of knowledge growth for every theory. Okay. That’s an interesting way to put it.

[01:51:49]  Blue: Yeah, unless you’re, I mean scientific theory, I mean I can always find counter examples if I’m allowed to use non -scientific theories. So I could say I’ve got a theory that George Washington exists. The way I’ve chosen to word that there’s probably no infinity of progress to be made. Maybe there is.

[01:52:05]  Red: I mean you could really nitpick something, I don’t know, maybe we’re in a simulation. What does it mean to exist? I guess

[01:52:12]  Blue: so. So, but really when I’m saying yes, there’s an infinity to be learned for each theory, I sort of have scientific theories in mind, right, empirical scientific theories.

[01:52:22]  Red: Okay.

[01:52:23]  Blue: So part of our knowledge, so once now that I’ve gotten you to accept that the theory itself and our knowledge of the theory aren’t the same thing, part of our knowledge about a theory is what its competitors are as that may change over time, even if our theory doesn’t change. So a theory, I argued that a theory can get stronger if the theory changes. I’m now going to argue that the word stronger can also be applied correctly to just the evolutionary nature of theories. So I have Einstein’s theory and it has this competitor Newton’s theory and we do Eddington’s experiment and yes, there’s going to eventually be some theory of quantum gravity that’s going to replace Einstein’s theory. In fact, we already know Einstein’s theory is wrong, but as of today, I don’t have that theory. So in terms of actual theories that exist, the fact that a new theory, so Einstein’s theory could be thought of as having a certain strength today. It’s stronger than Newton’s theory because Newton’s theory is refuted and Einstein’s theory, at least at the time of Eddington expedition, wasn’t refuted. Actually, you could argue it is refuted today. The existence of a new competitor creates a difference in strength. Now this just makes sense in terms of organisms. If I were to say this for organisms, you’d buy it immediately. If I were to say the lion is the strongest of the animals and suddenly a new type of mutant something shows up that can eat lions, then suddenly it’s not as strong. Okay, and you would understand what I mean. I think the same analogy applies very well to science, right?

[01:54:08]  Blue: That the strength of a theory is the word strength of a theory, the phrase strength of a theory, could refer to not the theory becoming more empirical, but how good its competitors are as of today. So if you will accept that the word strong could mean this, then absolutely the strength of a theory can change. Okay, as our knowledge about what its competitors are changes. A theory may get relatively stronger by its current known competitors getting knocked out of the race, or it may get relatively weaker by a new theory showing up that previously we hadn’t been able to think of, that potentially or maybe does make predictions better.

[01:54:56]  Red: Okay, so Bayesianism is true. That’s what you’re arguing, right?

[01:55:01]  Blue: So we’re going to need to do a totally separate podcast on Bayesianism. So let’s say that we had five competing theories. Now we just said this never really happens in real life, but let’s say we did. Okay, this is a thought experiment. I get to do whatever I want in the thought experiment.

[01:55:16]  Red: Okay.

[01:55:17]  Blue: And you used evidence to eliminate three of them. So let’s say that theories A and C are now the only two surviving theories out of the original five A, B, C, D, E, okay? Would it really be a mistake to say those two theories got stronger as a shorthand for we refuted three of the five theories and now there are only two left that we currently know about? That seems to me that’s a super reasonable way to use the word stronger.

[01:55:49]  Red: Sounds pretty sensible.

[01:55:51]  Blue: Now, in a sense, I admit that nothing changed for those two remaining theories, which is the point that every online paparian makes to me. Okay. What really changed was the other three theories that got refuted. But if we stop thinking in terms of theories themselves and instead think in terms of our knowledge about the theories, we just discovered that A and C survived testing while B, D, and E didn’t. Now, for sure, there’s a sense in which God already knew that, but we didn’t until we did the testing. So likewise, if someone discovered a theory F that was previously not known, we might say theory A and C just got weaker. And we know that that meant there’s a new competitor to worry about. Okay. So here the language of strength and weakness refers to the status of the known competitors rather than to the amount of empirical content. So this would be an example of how a theory under Popper’s epistemology can get stronger or weaker without the theory becoming more or less empirical.

[01:56:55]  Red: Okay. That seems very sensible.

[01:56:57]  Unknown: I

[01:56:57]  Red: mean, if you think about it, what I hear you saying is, well, to use the language of explicit and inexplicit again, I mean, within a given theory, there’s a lot of explicit knowledge there, and there’s kind of inexplicit knowledge that goes into that interpretation of that theory that is sort of in a state of flux, I guess. And so in that sense, theory is sort of, it’s not just about refuting or corroborating, it’s about evolution in a

[01:57:33]  Blue: way. Okay. So I’m trying to soften you up for the final blow here by pointing out that the word strength and a weakened could have two different meanings that are completely legitimate within Popper’s epistemology. So I can develop a third one. So there’s something still being missed here in terms of the importance of corroboration. Let’s take MWI, many worlds interpretation of quantum physics as an example. It is not yet the accepted paradigm of QM by most scientists. And I agree with the argument that it should be considered the best version of QM because it’s the only one that even tries to explain QM.

[01:58:08]  Red: But

[01:58:08]  Blue: it’s well known that all other quote interpretations of QM, at least currently, have no testable differences from MWI. They may someday. Deutsch has given examples of tests that would test between them using a AGI that’s built on a quantum computer. But we don’t have the technology to do the tests today.

[01:58:28]  Red: Well, I thought that was really fascinating when he said on,

[01:58:33]  Unknown: I

[01:58:33]  Red: think it was the Ferris podcast, he said that quantum computing was originally proposed as a test for many worlds. That’s accurate, right?

[01:58:47]  Blue: I think it is, yeah. Not many scientists would accept that today, by the way, if there are many worlds. But I think there’s a good argument to be made there. Actually, I don’t think there is a good argument that Deutsch is making there.

[01:59:01]  Red: Well, I think that’s what he, what I got is that that’s what he had in mind when he originally proposed a quantum computer. I don’t think that would

[01:59:09]  Blue: make sense. Yeah, because in the philosophy of what the implications were, that clearly that’s what his interest was.

[01:59:16]  Red: Yeah.

[01:59:16]  Blue: So now, one of the reasons why the other interpretations make the same predictions is MWI is largely due to the fact that they ad hoc in such a way to always make their predictions the same. Bohm in particular is ridiculously guilty of this from what I can tell. However, I’m at least a little bit sympathetic to why scientists are only slowly coming to accept MWI. MWI makes some pretty wild claims. So while I accept that MWI is the best explanation of QM today, I also accept the criticism that you can’t directly test other worlds, that that is a legitimate problem for MWI as of today. Okay. So here’s a thought experiment. Let’s say we built a time machine as per Deutsch’s fabric of reality chapter about time machines, which he says is really just moving to a different universe. And we go back in time and we change things. Now we have access to two worlds, one previously we only had one world. If such an experiment were made, would scientists adopt MWI in droves? Absolutely. Okay. But why isn’t MWI already the best theory? Okay. So on the one hand, it seems obvious to me that MWI is already the best theory. Yet it also seems obvious to me that actually testing one of MWI’s wildest claims and observing it directly should cause us to take MWI more seriously than we do today. And that is what would happen with scientists. But why is, is that just verificationism? Now, David Deutsch answers this question in an interview with Dr. Brakish Patel. It’s a related question, but it’s similar. And he says, what has happened there is that at the moment we have only one explanation that can’t be immediately knocked down.

[02:00:58]  Blue: If we had, if we did that through thought experiment, we might well decide that this will provide the ammunition to knock down even ideas for alternative explanations that have not yet been thought of. I think this is a brilliantly good answer on the part of David Deutsch. He’s saying that the reason why scientists would change their mind if you built a time machine that actually shows you other worlds is because even though MWI is already the best explanation, that we have suddenly corroborated or tested one of its wildest claims and it ruins potential competitors rather than actual competitors. So yes, you got it. Deutsch is saying that corroboration, passing experiments of wild claims has the effect of destroying even potential competitors that aren’t yet fully formed. So let me clarify. Theories go through a life cycle. It’s fairly normal for them to start out as weak in the sense of not testable and low empirical content and then get stronger over time. At any given moment, there are weak alternatives starting to get stronger. An experiment that comes out in favor of a theory, even if that in theory is the only one remaining, will often kill off those slowly strengthening alternatives that don’t yet exist as true theories. For example, there are some fairly weak proposals on how you can still save collapse theory, the collapse theories of QM. If you could actually travel to another world, those would be dead on the vine and no one would keep working with them. But at the moment, it isn’t all that unreasonable to keep trying to see if they can be developed into something stronger in terms of testability, I mean.

[02:02:43]  Blue: By the way, no, this is why I encourage Saadia to continue with her theories, even though I see them as entirely non -empirical and vague at this moment. Now, science is like the halting problem I’ve said. You never know if there is a better theory. You never even get to know for sure that a supposedly refuted theory won’t reemerge as some stronger form. Recall that means more testable. Later and later and thus again get to be a competitor again. This is why we speak of only tentative refutations. But I had to explain my actual criticism of Saadia’s theories in a nutshell, it would be something like this. She advances her theories and she says, oh, look, I’ve got this problem I can show with David Deutch’s theories, which would be quantum physics and Turing -Deutch thesis and things like that. But she acts as if the very fact that she shows a problem with those theories somehow creates merit for her theories, which are non -testable and therefore can’t be refuted. This is not how epistemology works. You only get merit for a theory in science at least, if your theory can clash with experience, re -experiment, and then the actual experiment is actually performed and we find that you are correct. I’m not denying that any of Saadia’s theories, like I’m not saying I know them to be false, I don’t know that. Who knows? Maybe she will eventually turn them into empirical theories and we’ll have something that we can really test at this point. Or if they’re wrong, she’ll never be able to. Now, we already know QM is false due to the fact that it clashes with general relativity.

[02:04:16]  Blue: So that’s why I’m unimpressed when Saadia discovers that there’s a problem of time with quantum theory, because I already knew it was false. Let’s just assume for the sake of argument that she was right though. So what? Quantum mechanics is still the best theory we have. It has never even had a single counter example. If all that mattered to a theory was that it was refuted versus not refuted, which is how the online paparians are looking at this, then I see no reason to not jump to Saadia’s nascent theories over QM, since I know QM is false and I don’t know that Saadia’s are false. There’s a certain rationality to that and that’s what she’s relying on. Even though her theories are actually vague, non -testable theories, they’re in some sense wishes of what a future theory might look like. The reason why we don’t do this is very simple. QM is a highly corroborated theory. It has passed in reality lots and lots of tests, whereas Saadia’s has not only passed no tests, but it’s not even testable. In Popper’s epistemology, it is better to be wrong than untestable. This is something that gets missed a lot. We expect true or true were theories about reality to actually say something about reality. If they don’t, then we’re not that interested in them, at least not as scientific theories. You’ve pointed out it could be of interest to them as philosophical theories maybe. Though we’ll reconsider them if someone manages to find a way to make them empirical theories. There’s a direct link now between the types of theories we’re interested in and their testability. And we don’t want a theory to just ad hoc explain a known problem.

[02:05:50]  Blue: We want it to do so empirically. That it makes brand new bold predictions that we haven’t thought of and then successfully passes those tests. This is Popper’s epistemology in a nutshell. And note how corroboration isn’t some unnecessary part of it. It’s the central to the whole thing working. What we really care about in science isn’t if a theory is true per se, but if it’s truer than all known competitors. This is the Popperian concept of verisimilitude. I wish dutch had not downplayed that concept. I feel like that’s why a lot of these problems come is because he did not include the concept of verisimilitude in his accounting of Popper’s epistemology. There’s a certain sense in which we care about verisimilitude more than truth. Although we do care about both. It depends on what questions we’re asking. The reason I take church -turing dutch seriously is because it follows from QM. Sure, I know QM is wrong, but I have no reason at all as of today to believe that whatever is wrong with QM will in some way affect CTD. Knowing that there’s a problem of time or a problem of gravity tells me nothing about how it would affect CTD. Maybe it will. I don’t know. But if you want to argue to me you should not accept CTD because it follows from QM and QM is refuted theory, then I’m just going to argue back. You should accept CTD because it follows from QM and QM has no true competitors that are in the same class as it today. So it is the truest theory we have available. So it is rational for now to assume its implications are correct.

[02:07:22]  Blue: I’ve now finished explaining why I find Saudia’s theories fairly uninteresting for the most part. Since the strength of a theory is more about what we know about a theory rather than about the theory itself, of course testing a theory changes our knowledge about a theory. It wouldn’t matter a whit if every single theory we ever held in science was false. Deutsch argues that’s the case. That’s why he said maybe all our theories are false. No, we know that both QM and Einstein are wrong, but we don’t know exactly how they are wrong. So it makes sense that we take well -tested implications of a theory more seriously and not so well tested implications with theory less seriously. Put it another way, corroboration is a rough way to figure out how true a theory is in certain areas or implications. It’s a rough measure of verisimilitude. Now, Andrew Crenshaw tells me that Popper and David Miller supposedly tried to formalize this idea. I don’t know if that’s true or not. That sounds very cool. Apparently they failed too. So there doesn’t seem to be any easy way to formalize the idea of verisimilitude mathematically as of today.

[02:08:25]  Red: So it’s almost like he tried to come up with his own version of Bayesianism. Is that what I’m getting?

[02:08:32]  Blue: We’ll have to talk about how this actually differs from Bayesianism because it does. Although what you’re really noticing though is that there are certain aspects of Bayesianism have verisimilitude. Is that really that surprising that Bayesianism has some verisimilitude? Get some things right? Not to me. Okay. I doubt it would be popular if it didn’t get some things right.

[02:08:53]  Red: Yeah.

[02:08:54]  Blue: We test theories to intuitively, amongst other things, we test theories to intuitively measure if the theories’ implications are true for various testable areas. So back to the programming example, of course you should test your code before it goes out no matter how good your explanation is because it’s very difficult to work out the full implications of a theory that contain infinite empirical content as with Euclidean geometry. There’s an infinity of ways that you can go about most programs, depends on the program. There’s usually an infinity of ways in which you can interact with it. So it has an infinite amount of empirical content. It’s impossible to test it all. So what you can do though is you can actually test the code and get immediate feedback from reality to see if the change that you made makes things worse. At least along certain branches of testing that you know are the most important. The ones for my project that might be making sure you can actually check out and buy something because if that’s broken then the whole thing’s broken, right? So testing may not guarantee me the code is entirely correct, it still has bugs, but it does give me confidence that certain highly used paths through the code will work correctly because I verified that they did. And this is now a different way of understanding why corroboration is so important to science. So HoCast 92 question eight for David. Brett argues that in school it’s long been falsely taught that you go out into the world and you observe stuff and you extract knowledge in some way or you derive knowledge from these observations.

[02:10:26]  Blue: He says that instead we should start with conjectures rather than observations and that the purpose of observations on this view is to quote, decide between conjectures. Now the first part of that statement is entirely true that we do not go out into the world and observe stuff and extract knowledge in some way from those observations absent conjecture. But the second part of that statement is it may have some verisimilitude, but it’s not strictly a true statement. The purpose of observations is not solely to decide between conjectures. That is one of their purposes. That’s the gold standard of purposes. It’s the most important purpose, but it’s not the sole purpose. They also help us test the verisimilitude of a theory regardless of if we have a good competitor. This is why it makes sense that scientists would change their mind about many worlds if we actually could travel to another world and actually see them because that would be a test of one of the wildest claims of many worlds that right now we’re incapable of testing. It would create such huge problems for the nascent competitors to many worlds that it would probably just kill them all. Just outright, boom, they’re all done. Now think about that for a second. That’s an example of how a corroborating test can strengthen a theory that has no competitors and is not an example of deciding between conjectures unless you want to say, well, you’re still deciding between nascent conjectures. I guess in that sense, I’d agree, but it’s not strictly speaking a competing theory. These are nascent theories we’re talking about. If we had a way to directly test the existence of alternate universes, we would do that.

[02:12:08]  Blue: And if we found them, it would cause the scientific community to adopt many worlds precisely because such a test would strengthen the theory by killing off nascent competitors to the point where it’s almost impossible to think of how else you would go about it. You can’t even think of good potential competitors anymore. It would kill off the nascent competitors, but it would also verify that that part of the theory, previously untested, has now been tested, therefore adjusting our knowledge about the theory. Okay, now with all this in mind, let’s go back to Einstein. Why did the scientific community not accept his theory as best compared to Newton, even though it fit into existing data better due to the perihelion of Mercury and the light experiment? Well, because it’s trivially easy to fit wrong theories to existing data, but it is not easy to have to make new bold predictions and then corroborate those. So the scientific community was correct to wait for the editing expedition and only have a crisis at that point. Okay, this proves Kuhn right. Right, at least on, this actually, sorry, this proves Kuhn wrong because there was no crisis until Einstein invented one, although it does prove that he’s right in some important ways. Once the evidence came out supporting, quote, in theory, his theory, making a really weird prediction, only then did people realize Newton’s theory was refuted. Individual observations weren’t enough. We needed actual support for a new theory that could cast those predictions into the role of a refutation. Think back to what Deutsch said about, you can’t really refute a theory without a second theory, without a competing theory.

[02:13:48]  Blue: Okay, you literally need a separate theory to be able to refute in the ultimate sense of the theory proper, a theory. You can’t just have observations and it has to be able to explain what’s wrong with the other theory. Okay, and it can’t be just existing data. It has to be a bold new prediction that wasn’t thought of previously, or else for all we know, it was just a new theory that was made to fit the existing data. Okay, that’s impossible to do if it’s a corroborating test, that the fear that, oh, Einstein just fit it to existing data disappears. Right, because it’s a corroborating test. It’s not something Einstein could have just fit the data to anymore. Based on this argument, I’d say critical rational zero, scientists two, one for knowing observations aren’t refutations, and one for knowing that only supporting evidence for a bold new theory can turn observations into a refutation. Now, here’s where things get interesting, Peter. This means the positivists were not entirely wrong. Now, you just brought this up with the Bayseans, right? You can sense there’s a sort of Bayesian character to this. Now, not ultimately, Bayseans make a lot more really wild claims, some of which are just dumb, right? But is it really that surprising that these theories, positivism and Bayesianism have some correct content? Like, why would people believe in these theories if there wasn’t a sort of appeal to them because they kind of get things right in some cases? Right, and I’ve argued this before that induction’s not strictly wrong. It’s just been superseded by critical rationalism, depending on what you mean by induction.

[02:15:33]  Blue: In fact, the inductive way of looking at life in science, it actually worked just fine for years because the institutions built around it were actually doing something slightly different than what they thought it was doing. And what Carl Popper really did is he solved the problem of induction by saying, okay, here’s what’s really going on, and notice that I don’t need any inductive logic to make it work. And that’s the sense in which he refuted induction, right? Is that really he refuted the need for inductive logic and he thereby solved the problem of induction.

[02:16:08]  Red: Can we define positivism for a sec? I used to think it meant the same thing as scientism, but I mean, it kind of does.

[02:16:18]  Blue: So positivism, so I’m not a philosopher and I hate philosophy. So you can’t expect me to be really on the ball with philosophies out there as I don’t study these and I don’t care. There’s positivism and there’s logical positivism. I don’t understand the difference between the two and they seem somewhat similar to me, but both of them had something in common, the word positivism in them. The idea of both theories was that we proved theories true by verifying them. The idea of, so that’s why Popper said he was a negativist instead of a positivist. Now they did have all these justificationist ideals that they were dealing with. So did inductivists, right? And that was really what Popper put down as he showed, you don’t need any of that justificationism. You don’t need certainty. It doesn’t even matter whether you’re certain or not. And because that’s so tightly tied to positivism, it’s really basically impossible to disentangle the true parts of positivism and the false parts of positivism today. But the fact is that there is a somewhat positivistic nature to Popper’s epistemology if you’re talking about verisimilitude rather than truth. And that’s really the point I’m trying to make, right? And I’m arguing that verisimilitude is actually more central to Popper’s epistemology than truth. Truth’s a super important part of his epistemology. You have to understand that really, but really it doesn’t matter if theories are true or not. If truth of the theory mattered, then yeah, maybe I should care about Saudia’s theory more than I care about QM. But I don’t, right? I mean, I know it’s a false theory. It just doesn’t matter because at the end of the day, what I really care about is the verisimilitude of the theory.

[02:18:01]  Blue: The verisimilitude of quantum theory must be super duper high because there’s never been a negative counter -example to it, right? I mean, it’s, and it makes such wild predictions that all come true. So it must be saying something very, very truth -like about reality. Okay, now all of this brings us to Deborah Mail. I’ve mentioned her before and I’ve suggested that her book, How to Get Beyond the Statistics Wars, sorry, Statistical Inference as Severe Testing, How to Get Beyond the Statistics Wars. I’m still trying to read it. I’m not a statistician. I’ve taken stats twice, once at the undergrad level, once at the graduate level for my MBA. And then I took machine learning for a separate, separate masters, which is related to statistics. So I’m not like completely dumb in this department, but I’m also not a professional and I’ve basically just schooled but don’t do it for a living. I just, I struggle with her book. It’s so technical and honestly, the way she writes, I don’t like. I don’t, I think it’s meant for people who are experts in the field and she’s very poetic, but she’s expecting me to have a lot of background knowledge that I don’t. Despite that fact, I’m convinced that what she’s saying in this book is really important that she has actually come up with an improvement on critical rationalism that critical rationalists need to get comfortable with because she has actually improved the theory. Now, I will need to finish the book and I’ll someday do a podcast on it and I will just explain as much as I’m able to get out of the book at this point.

[02:19:45]  Blue: But I want to bring up some things that she talks about in her book because I actually think that her version of this does a better job of explaining what I just explained about corroboration and why it’s important. So one of the things that she points out that a lot of people seem to have missed is that the Eddington expedition, we keep using Einstein’s general theory of relativity as kind of the gold standard. It’s the example that Brett used that I’m using as my jumping off point. It’s the example I keep using. And we often say that Einstein did the right thing. He made this bold prediction and in theory, they could have gone out, they could have done the Eddington expedition and it would have falsified his theory. And therefore he was matching the falsification criteria and therefore it was a scientific theory. None of that is true. Like really and truly, it is just not true. Here’s the actual truth. Based on the level that our instruments we had at the time, we didn’t actually know if they were sensitive enough to show the movement of the stars that was predicted by Einstein’s theory. We hoped they were sensitive enough to show it. But they were blurry, they were doing photo plates according to Mayo and in all honesty, they had to measure them statistically. Now, this is something else because there’s this really huge hostility towards statistics, statistics and probability theory amongst the the Deutschians. But this gold standard of refutation that we use, it was done entirely through probability theory. And if we because it was, if they had not found the curvature, the stars moving like they were, nobody would have thought it refuted Einstein’s theory.

[02:21:41]  Blue: They would have simply said, we don’t currently have evidence to refute Newton’s theory. Now, from this point of view, you might make the mistake of saying, oh my gosh, the positives were positives, positivists were completely right. This was actually a positive theory that we proved by actually finding it. How would you actually fold this example into Hopper’s epistemology? Well, Deborah Mayle explains how and it’s really exactly what you would have expected. In many ways, the theory that the test, the Eddington expedition test, was really about just trying to refute Newton’s theory. If there was a curvature, it refuted Newton’s theory and it left Einstein’s theory as the sole remaining theory. But if it hadn’t found the curvature that they were expecting, then it just simply would have been, well, we’re not really sure. And it would have continued as a competing theory. And at that point, our knowledge of the theory would have been, we don’t have a single corroborating test for Einstein’s theory. And it would have continued to not been taken very seriously until someone came up with a better corroborating test or came with instruments that allowed us to test it better. Now, later, of course, we started using radio telescopes and we now have instruments that can test these theories in a much stronger way. Mayle points out that even with radioscopes, you’re still using probability theory. So the Deutschians that want to erase probability theory out of Einstein’s theory, by the way, that’s such a huge mistake. And Hopper spends chapters explaining why that’s such a huge mistake. But even if you do want to do that, the fact is, is that there probably are no non probabilistic theories.

[02:23:28]  Blue: It’s just that in some cases, our instruments become so precise that we basically don’t have to measure the probability anymore, because it’s the p -value is so small in this case. But this is just basically how science works. You go out and you do this experiment and your instruments are in precise. And so you use probability theory to be able to measure, well, how much of a move was there? What’s the average of this star? And then she points out that the Eddington expedition, that by today’s standards, it probably wouldn’t have been considered as passing strong enough p -value because the instruments were so bad back then. This is why you sometimes hear that the Eddington expedition actually did not support Einstein’s theory. What they really mean is that it was less clear because the photo plates were kind of blurred and it was a little bit hard to tell. And the p -value wasn’t that good at that point. But Mayo argues, actually, this was a decent explanation, especially given at the time. And of course, we’ve repeated the explanation with Rioscope since then. So this is just how science actually works. It’s really not about it. Stop when you think about it. We can’t interact with the theory itself. We can only interact with our knowledge about the theory. I mean, this is so obvious when you stop and think about it. Right? And our knowledge about the theory is constantly changing as we test things, as we corroborate things. Now, because of this, Mayo argues, I’m going to give the actual quote in this case, that we don’t, this is my interpretation. We don’t accept untested portions of the theory. Now, I don’t think she quite says that. So judge for yourself.

[02:25:11]  Blue: But she says, I mean, this is where I have to stop for a second and find the actual quote to make sure I get it right. Here is the actual quote from Mayo. She says, the result of all this, of the analysis with the Eddington expedition, was merely evidence of a small piece of general relativity, an Einstein -like deflection effect. The general relativity theory passed the test, but clearly they couldn’t infer general relativity, general relativity severely because it needs to pass severe tests. Even now, only its severely tested parts are accepted, or at least to probe relativistic gravity. So now, and then she talks about how because of this, general relativity after the Eddington expedition went through a period of stagnation or hibernation after the early eclipse results. No one knew how to link it up with experiment. So this is, by the way, page 128 and page 160 of her book. Now this clashes with something that Deutsch said, or it seems to. I’m going to argue it doesn’t actually clash with something Deutsch said. Deutsch talked about how we should accept the best theories, implications, even the parts we can’t test. Well, this makes a great deal of sense to me. If MWI is our only real explanation of quantum theory, then it makes sense to tentatively accept that there is a multiverse that exists. And I do, right? On the other hand, what Mayo is saying is that it’s understandable that scientists will continue to reject that because that part of the theory isn’t actually tested. We know all theories are false. So it makes all current theories at least. I argued we could have a true theory, but we don’t.

[02:26:56]  Blue: So it makes sense that we would want to test the implications of the theory and that we would see the implications that are untested as a different status than the implications that are tested. Now here’s the brilliant thing. Based on that, and I got to get this quote right, this is page 88 from her book. She talks about a conversation she was having with a fellow traveler over at Starbucks. She doesn’t tell us who the fellow traveler was. She doesn’t even tell us which one of them was her. This is why I find some of her book hard is because she’s so poetic that she sometimes loses me. Traveler one, if merely logical falsifiability suffices for a theory to be scientific, so they’re talking about the boundary condition, then we can’t properly oust astrology from the scientific pathion. Plenty of nutty theories have been falsified. So by definition, they’re scientific. This is a common criticism of Popper. Moreover, scientists aren’t always looking to subject well -crobbered theories to grave risk of falsification. This is Kuhn’s criticism of Popper. Traveler two, I’ve been thinking about this. On your first point, Popper confuses things by making it sound as if he’s asking, when is the theory unscientific? What he is actually asking or should be asking is, when is an inquiry into a theory as an appraisal of claim H unscientific? So she’s shifted this now from, is the theory scientific to, is the inquiry scientific? We want to distinguish emeritus modes of inquiry from those that are bent, which means bad evidence, no test. If the test method enables ad hoc maneuvering, he’s at a very sneaky face -saving devices, then the inquiry, the handling and use of data is unscientific.

[02:28:49]  Blue: Despite being logically falsifiable, theories can be rendered immune from falsification by means of cavilar methods for their testing and adhering to a falsified theory, no matter what, no matter what is poor science. Some areas have so much noise and or flexibility that they can’t or won’t distinguish warranted from unwarranted explanations of failed predictions. Rivals may find flaws in one another’s inquiry or model, but the criticism is not constrained by what’s actually responsible. This is another way inquiries can become unscientific. Now, I do want to point out, she says, adhering to a falsified theory no matter what is poor science. She later does clarify, well, unless you don’t have an alternative. So she’s actually not disagreeing with Deutsch over that. But I think she’s making a fair point. And in fact, I think it’s pretty profound. We keep trying to talk about the boundary condition of theories, and that makes a certain amount of sense. But as I pointed out, it’s really more about the inquiry. So even quantum physics, if you treat it with the wrong attitude, if you make an unscientific inquiry, it becomes unscientific. Likewise, something like a myth, which nobody would call scientific, can actually be quite scientific in the Poparian sense if you got the right attitude towards it. If you’re allowing it to the inquiries to matter and for it to be falsified and to try to make improved theories. There’s a lot more going on here, but I feel like that’s a paradigm shift for me. And I think that that does explain why cooperation actually matters. Because what we’re doing when we’re talking about cooperation is we’re talking about did this test, was it a good scientific inquiry into this theory? And how strong was it?

[02:30:43]  Blue: By which I mean, how much did it actually test the theory? Or is it prone to giving false results? Now let’s go back and let’s talk about some of the things we’ve talked about on this show. And I think I can explain, now that I’ve explained all this, some points I’ve made before that I was a little bit sloppy on because it was the best I could do at the time. I made the statement that IQ theory is scientific, but that Brett’s theory was not, or at least he was choosing not make it scientific because he was choosing not allow it to be testable. But then I said, but IQ theory is really still a very, very weak theory. Now at the time I explained that in terms of not being very empirical. And that’s still true. I think there’s more to it than that. I think what we’re really talking about is that the corroborating instances of IQ theory don’t really test it that well. And that’s therefore the inquiry into it doesn’t tell us very much that we can easily think of ways in which errors may have crept in. I think this is actually why Deutschians hate the theory so much. And I think this is the part of their hatred towards IQ theory that they’re actually getting right. That it’s not a super scientific theory. Yes, you can say it’s because it’s not highly empirical, but in a lot of ways it’s because it’s just really difficult to make good corroborating tests for it that really force us to test the theory deeply, severely. And I think that this is true for a number of things we’ve talked about, right? We haven’t, we’ve talked about animal intelligence.

[02:32:13]  Blue: I think we’ve got a similar problem there. And we haven’t talked about animal feelings that much because we read an episode on it. But I think that there we’ve got the same problem that to some degree, those that are skeptical of animal feelings, even though it is in a certain sense the best theory, it’s just not a strong theory because we’ve never come up with great tests that deeply, deeply tested. Therefore, skepticism of it is at least somewhat warranted. Now, my arguments to the Deutschians on this would be, but you don’t get to call your theory the best theory. That’s really what I’m objecting to. And I know people don’t see me that way because I’ve really come to realize that if I’m criticizing a theory, the default assumption is that I must support the other theory that they know about. And I’ve received numerous emails from people sometimes after listening to an episode or something. Like we were talking about, Miller made a comment about, I was quoting Miller or something and I talked about how he said that there are no good reasons to believe in something. And I kind of took exception to using the term good reasons in that way. But then I went on to say, but I actually get what he’s trying to say by good reasons, he means justified reasons, in which case I agree there are no good reasons. Now, that’s what I said if I recall correctly on the podcast. Despite this, I got an email from someone saying, you just think that there’s a need for justification. And I mean, they laid out exactly the opposite of what I actually said. And it’s not too hard to see why they did.

[02:33:49]  Blue: It’s because I was being critical of something they believed in. And therefore the assumption was, even though I said otherwise, I must be in the other camp.

[02:33:58]  Red: Well, people like to put other people in boxes. And usually they fit pretty well, too.

[02:34:06]  Blue: This is where I want to kind of make a point. If we ever do an episode on Bruce’s new revision of how he looks at Popper’s epistemology, which I intend to do eventually, this is one of the key things I’m going to really point out. And this is, you know, Miller talks about, we don’t believe in any of our theories. And this is maybe a little dumb because in a certain sense we do. But he’s also right, right? Like if you were to ask me, do you believe in many worlds? I really don’t. What I believe, if anything, is that I’ve done a critical analysis of the competing theories. And I found that only one of them was explanatory and the others weren’t. And that is the current state of the critical discussion. And I’ve now made what I feel is an objective analysis of the critical discussion. I don’t really accept, reject, or believe, or not believe any theory. I don’t believe animals feel things or that animals don’t feel things. I simply say, here is the state of the critical discussion. Here are the corroborating instances. This is how well or severely they’ve tested things. And here’s the competitors. At least the ones I know about. And they’re worse. Therefore, this is now currently the best theory, but this isn’t a great theory. Or it is a great theory, depending on the circumstances. And really, that is what critical rationalism is. It’s the assessment of the critical discussion, not really strictly speaking, deciding which theory is true and which theory is not true.

[02:35:42]  Red: In which case you’re the other guy says, you’re nitpicking on just say you believe in it.

[02:35:50]  Blue: Okay, and you know what? That’s fair. And if someone were to say that, I would say, okay, you’re right. But like there’s different levels, right? Like I completely agree with Deutsch that many worlds is the best explanation. Therefore, I think we have good reasons to accept. That is a good reason to accept the multiverse, tentatively, for now. But I also accept Mayo’s point that since it isn’t tested, that is a weaker part of the theory. And that may change if we can come up with the corrobority test for it. Now, those can both be true at the same time because they’re not in contradiction to each other. You try to tell me what my belief is. I’m fine with it. If you want to say that means you believe in the theory, I’ll accept that. It’s just a label. I don’t care, right? But my concentration is on what’s the actual state of the critical discussion at this moment, at least as far as I’m able to figure out. Now, that means I have to constantly reevaluate what the state is every time I discover a new competing theory. One of the things I pointed out with like animals’ feelings, right? Do they actually feel things? Do they have qualia? Initially, I can only look at the theories I know about. So I know about Dennis Hackathall’s theory. And I can assess that objectively. And I know about Bern’s theory and Francis DeWall’s theory. And I can assess that as a scientific theory. How well have they severely tested this versus how well has Dennis? Well, there, there’s no contest. Dennis is absolutely in the non -testable, not well argued. It’s inductive. I mean, like it’s just no contest between those two.

[02:37:35]  Blue: Is that really the best alternative theory though? Is there a better alternative theory that’s out there that I just haven’t discovered that makes the argument that animals don’t feel things? I know, I don’t really know, right? So all I’m doing is, is I’m assessing the current state that I know about. Now I’ve made attempts to go out and to find alternative theories. And I have. I actually, when we do one on animal feelings, I will give you two alternative theories that the Deutschians don’t know about, that are possible ways to go about trying to show animals don’t feel things, or rather, some animals don’t feel things. They’re in lies. The problem is that each of the theories I found suggested that at least some animals do feel things. That’s the right way to go about this, is you’re trying to find the best alternative theories, and then you’re trying to assess how severely tested they are at this point. That is Popper’s epistemology. And that changes over time. That’s why collaborations matter, because they do change the state of your knowledge of the theory almost constantly. There’s a lot more I could probably, well that Deborah Mayo could probably say here, and I’m still trying to work out all the implications of her, what really a small but profound tweak to Popper’s epistemology, by shifting it to be about the inquiry, instead about the theory itself. Clearly there’s some heavy overlap between those two. But I do think that it has implications that most Popperians are not aware of, which is why I’m recommending her book, because I would love to see smarter people than me read this book and see how we can work some of this into critical rationalism.

[02:39:12]  Blue: I think she is clearly a critical rationalist. There are many different brands of critical rationalism. I’ve tended to be a Popperian critical rationalism, but you think about how Popper had numerous students, Lakatos and Freybrand. I can’t even pronounce their names, but there’s a whole bunch of different versions of critical rationalism, and some of them have some true things. If I remember correctly, Lakatos was the first one before Deutsch to point out that you needed a separate theory to truly falsify a theory. So Deutsch may have come up with that on his own separate, for all I know, but he wasn’t the first to come up with it. It was actually one of Popper’s students, and it was something that Popper never really fully got. So we have improved versions of critical rationalism out there. It’s not that weird that Popper got some things wrong, and some of his later students got some things right. And what we’re really trying to do is we’re trying to come up with what’s the actual best version of critical rationalism that exists today. My investigation into this has found that Popper, in some ways, just had so much right to begin with, that it was actually a little difficult to improve on him. And I think that’s why Lakatos famously, in some ways, weakened critical rationalism, even though he got some things right that Popper got wrong, that ultimately he came up with a version that was not as good. And I think most Popperians and critical rationalists would agree with me on that. I’ve not that I’ve read a lot of his books, but just from what I have read, it

[02:40:47]  Blue: seems like there seems to be general feeling that what he came up with, while there were some interesting good improvements that he also weakened many of the points that were already good. And I think it’s hard to improve upon a theory, but I do feel like Deborah Mayo may be onto something that’s a genuine improvement that will increase our knowledge of epistemology, which is why. And by the way, if she’s right, she has a synthesis of Bayesianism and critical rationalism. So you keep asking about that, and I’m still trying to figure out what she’s saying about it, so I’m not sure I can explain it to you. But her improved version of critical rationalism actually explains what parts of Bayesianism get right and what parts get wrong, which I find very interesting, because I’ve been trying to figure that out for myself for a while, and I find it a very difficult subject. So we’ll do a podcast on that, too, at some point. I think most Popperians I know today simply dismiss Bayesianism almost entirely lock -stalking barrel. You pay attention to what Deutsch says, though. Clearly, parts of it are right. He’ll say, well, there’s Bayesian reasoning and there’s Bayesian epistemology, but there’s no hard divide between those. So it’s very difficult to… I’m still trying to work it out myself, and that’s something we’ll save for a future time once I think I’ve studied it more deeply.

[02:42:10]  Red: Yeah, I guess the implications of these words change a lot when you add the ism, like you were saying before. Empirical theories are, of course, great. Empiricism, a little different. I mean, it could be a similar kind of thing with the Bayesianism and just the Bayesian reasoning, which is, I guess, a form of… It’s really a form of induction, right? It’s just kind of saying that induction has some validity, which I kind of think it does, honestly.

[02:42:43]  Blue: It has exactly as much validity as critical rationalism says it does and no more. The relationship is very much like general relativity to Newton’s theory. I would have to explain that in more context in a separate podcast. I’m actually planning to do that. We’ll do a podcast called The Kinds of Induction or something like that, and I’ll explain what I’ve figured out. I believe I figured out up to this point about the relationship between induction and critical rationalism, what parts induction… Popper says this. A lot of people are shocked when I say this. I’ll say, well, actually, Popper said, you can call critical rationalism induction if you want. I’ve had people go through the roof over that, right? It’s like, no, he really does say that, right? Popper had such an open attitude towards these things. He just wasn’t into reacting to words. Popper had no war on words, whereas modern Popperians seem to have significant war on words going on, which I think is unfortunate. Anyhow, that would be my take on corroboration. So to summarize, I think corroboration is actually entirely central to Popper’s epistemology. But I think the way in which it is is really because Popper’s epistemology is more about verisimilitude than truth. It’s about both and both play an important role. But at the end of the day, I care more about the verisimilitude of a theory than the truth of the theory, because all theories are wrong, at least as of today. And therefore, it makes perfect sense that I would care about corroborations, because corroborations allow me to see how severely tested the various implications of a false theory are. Given that the theories are false, it makes sense that some of its implications will be wrong.

[02:44:31]  Blue: Therefore, I want to test them. That just makes sense.

[02:44:37]  Red: Well, I thank you for all your thoughts, Bruce. I’ve learned a lot on my current class on the Bruce Nielsen University. It’s been great as usual.

[02:44:51]  Blue: All right. Thank you. After recording this episode, I felt like there are some things that I hadn’t covered that I wanted to include. So I’m going to do an afterwards here. One of them is that I want to address the argument from David Miller from his book Criticleratialism, a Restatement in Defense about how he handles corroboration and how I in some senses agree with him and in some senses disagree with him. Here’s the quote. What then of corroboration? Why do corroborations matter? According to Watkins, is a question often posed to Popper by his critics but never answered in a satisfactory way. It must be remembered that it is part of falsificationist epistemology that passing a test does nothing to secure, confirm, or in any way brighten the prospects of a theory. Nonetheless, the question can be answered satisfactorily. The answer is that corroborations doesn’t matter, even in the practical realm. It has no epistemological significance at all, as Popper always insisted. But testing matters and has undeniable methodological significance. We want true theories. Testing is important because it is only by subjecting our theories to tests that we have an opportunity of eliminating those that are false. The more severe the tests, the more generous the opportunity. We might put things this way. When a theory fails a test, we learn something but end up knowing nothing since what we knew, our theory, has been eliminated. But when a theory passes a test, when that is to say it is corroborated, we learn nothing since we already knew that what the result of the test was going to be. But we continue to know something. Cooperation is doubtless needed if science is to exist.

[02:46:31]  Blue: For if no theory were ever corroborated, there would be no science, but it makes no contribution to the growth or to the progress of science. Now when I actually first read this passage, I wrote into the margin so totally false. I don’t actually feel that way anymore. And let me explain why. Yes, if you read this as a question about corroboration in any sense at all, and if it matters in any sense at all, then it clearly is a false statement. Because it is just not true that you learn nothing when you corroborate something. Now let me just go back to the example that came from the podcast. If we were to actually invent a time machine and we were to go back in time and we were to actually create a new world in, under many worlds, quantum physics, and now we can actually see that world and verify that it exists by experiment, have we learned nothing? Well, to claim we’ve learned nothing here just isn’t true. And even according to that quote I used from David Deutch, at a minimum, it has drastically changed what we know about our knowledge about the theories. If nothing else, it killed off nascent competitors, collapse theories that were being generated, and showed, look, these aren’t even worth trying to look into, because now you have to take into consideration the fact that we actually have verified the existence of other worlds. So the state of knowledge does change, and corroboration clearly has taught us something. Likewise, it just isn’t true that if we refute a theory that we know nothing because we eliminated the theory. Really, all our theories are already false. We already know quantum physics is false.

[02:48:06]  Blue: If it’s true that at that point we know nothing and we’re not taking verisimilitude into consideration, then this goes back to my argument with Saudia’s, the example with Saudia’s theories. We might as well accept her nascent theories as better than quantum physics, because we already know nothing because we know quantum physics is wrong. It just isn’t true. Quantum physics has all sorts of verisimilitude, truth -likeness, and because we already know quantum physics is false, it makes absolutely perfect sense that we want to test which parts of its implications are true. Some of its implications must be false. That just must be the case. We just don’t know what they are at this point. We haven’t found any that are false at this point, because it’s a highly corroborated theory that has never had a counter -example so far. But because we know it’s a false theory, it makes perfect sense that we want to test the truth -likeness of its implications as much as possible, whatever we’re currently capable of testing with our current knowledge and our current instruments. Corroboration does play a role in how we assess which parts of the implications we’ve severely tested and which parts we haven’t. So I understood that by counter -example I had now falsified what David Miller had said. Now here’s the thing though. When I actually stop and I really pay attention to what he’s saying in context, rather than just pulling it out of context, I think there’s a good case to be made that what he’s really saying is not corroborations do not matter in any sense at all, but that do they matter in terms of whether this theory is true or not?

[02:49:39]  Blue: Now if you look at it in this way, then it seems to me that his statement’s entirely true. I already accept that many worlds quantum physics is the best explanation for quantum and the only explanation for quantum physics. Because of that, there is a prediction that comes out of that, that if we could time travel and go back in time that a new world would open up. So I don’t actually in that sense learn anything. I’d actually just find out that the theory’s prediction was correct just like I thought it would be. This is really what David Miller’s trying to say, I think. Now, here’s where things kind of get interesting though. This is not how he actually words it. You do have to look at the full context of what he’s saying. He actually says why do corroborations matter and then goes on to answer that question. So it’s somewhat understandable that I initially thought that what he was saying was false because I was trying to interpret it in a more absolute sense than he actually intended it. In other words, we could say that the real question he’s answering wasn’t why do corroborations matter, but why do corroborations matter with respect to the truth or falseness of a theory. When it’s rephrased in this way and only when it’s rephrased in this way, the rest of what he says is clearly true. All of this is a good example of what I’m really trying to get at, that there is a difference between the theory itself and our knowledge about the theory. All sorts of things, including corroborations, change the state of the knowledge about our theory. A corroboration kills off nascent competitors.

[02:51:10]  Blue: It makes it much harder to come up with competitors. That is one of the reasons why we need corroboration, and it also tells us something about verisimilitude. So going back to my example of software, if you’re looking at what David Miller said with respect to software, the fact that I run a test and that the test comes out positive actually tells me exactly what I already thought that I had fixed the bug. If you’re only looking at it in terms of is this true or is it not, then yeah, he’s right. However, if you look at it in a different way in terms of verisimilitude, which I think is the way you really should be looking at it when you’re dealing with software or really for that matter with scientific theories, then it comes off looking a lot different. If I can go out and I can fix a bug and then I am able to go through and test from start to finish, say checking out an item in a cart because that’s what my software does, let’s say, then I now know through that corroboration that I’m able to get all the way through the cart, which turns out to be a really useful piece of information. At a minimum, I now know, look, there could, I may have introduced a new bug, there may be a new problem, but at least I know that in terms of certain really important paths I have now severely tested them and I’ve got verification that I’m able to go through those.

[02:52:32]  Blue: So yes, finding out that you can verify the existence of other worlds through use of a time machine absolutely does tell you something that is incredibly important to the state of our knowledge about the scientific theory of quantum physics. This also explains why I really like what Deborah Mayo has to say about this, her attempt to shift it from the question of is this a scientific theory, to is this a scientific inquiry, I really feel like that’s a fairly big paradigm shift, even though it may seem very subtle at first, and that it really teaches us something that’s important here. To understand why I feel this way, let me actually quote something from David Deutsch. So, Deutsch, talking about how do we choose what explanation to adopt says the following. So how does one decide which explanation to adopt? Common sense says one weighs them, or weighs the evidence that their arguments presents. This is an ancient metaphor. Statues of justice have carried scale since antiquity. More recently, inductivism has cast scientific thinking into the same mold, saying that scientific theories are chosen, justified, and believed, and somehow even formed in the first place, according to the weight of the evidence in their favor. Consider this supposed weighing process. Each piece of evidence, including each feeling, prejudice, value, axiom, argument, and so on depends on what weight it has in the person’s mind, would contribute that amount to the person’s preferences between various explanations.

[02:53:57]  Blue: Now, in context, Deutsch goes on to liken this to Eros Theorem, shows how this is actually an incoherent way of looking at it as per Eros Theorem, and points out that we’re really kind of ignores the ability to solve problems, where we can actually create new options as we go. When in full context, I feel like I largely agree with what David Deutsch is trying to say. But you could be forgiven from taking a look at what Deutsch is saying here, and coming away with the idea that there is something wrong with ever talking about the weight of evidence. But in fact, that’s not true. When we refer to the weight of evidence, what we’re actually referring to was how good was the inquiry and how severely did it test the theory, or at least that’s what it should refer to in a scientific context. Now, what do I mean by this? Well, consider that as a hypothetical example, let’s pretend that we have a drug that we’re trying to test. We want to check through testing whether this drug creates the effect, the cause that we intend for it to do, versus, say, a placebo. So we go ahead and we do random control trials, and we come up with some result, and then we say that this result supports or refutes one of the two options. It does or it doesn’t have the effect that we intended. Now, nothing wrong so far. For the sake of argument, let’s say that the result of this ends up supporting our, we say that there’s enough evidence there to support our conjecture that this drug does have the causal effect that we expect it to.

[02:55:31]  Blue: Let’s say somebody comes along and challenges us and they say, well, you know, actually, you didn’t do us sufficiently control on this. For example, it wasn’t double -blind. Now, here’s the question you have to ask. Is this evidence for our hypothesis, or to put this in negativist terms, does it refute the other hypothesis? Well, it turns out there’s no yes or no answer to this question, because what we’ve done is we still have a result. That result might actually be considered evidence for what we’re saying, and that might even be correct. But someone’s now pointed out that there was a flaw in the methodology or into the inquiry, and that really what we’re going to want is we’re going to want to repeat this test with that flaw removed and see what the result is at that point. It would be fair to say, well, this is evidence, but maybe it’s weak evidence because we’ve got concerns with the inquiry. This is really what Debra Mayo’s getting at when she talks about bent, bad evidence, no test. The problem is that this isn’t a dichotomy. You can almost always find some things to criticize about a study, and some of them are more meaningful than others. And so you almost have to put it on a continuum, which is why we then turned it into a weighing metaphor. Now, at some level, I get the argument that we could say, okay, maybe this experiment was so bad that even calling it weak evidence, it would be easier just to say it’s not evidence at all.

[02:56:59]  Blue: There is a subjective call that takes place here, though, because some studies are better than others, and they have more to criticize and less to criticize, and this is how science progresses. Therefore, it makes perfect sense that we might talk about, well, there’s this evidence, but it’s not very strong or something along those lines. It’s a way of trying to assess how good was the inquiry, and therefore how good was the result of the inquiry. Is it stronger or weaker? When put in this way, it’s not actually at odds with what David Deutch is saying, nor is it at odds with what David Miller is saying, but it does cause us to have a new recognition that when a person talks about weight of evidence, they might be talking about something that’s meaningful. A related question might be that of p -values. Let’s say, and this is a hypothetical example with the time to test a medicine, that we came out with a p -value of 0.05. So in other words, there is a 5 % chance that just by chance, we would have got this result, even if the drug was doing nothing. That isn’t as strong as evidence. That’s probably going to be considered by many people to be valid, and we would reject the null hypothesis at this point. But that really isn’t as strong of evidence if the p -value had been 0.01 or something better than that. Again, we see that there are cases where it makes sense to talk about the weight of evidence. Now, I know you’re probably going to want to ask, okay, but how does this relate to Bayesian epistemology?

[02:58:26]  Blue: Isn’t this what Bayesian epistemologies are talking about when they talk about weight of evidence and they change credences and things like that? Well, as I said in the podcast, not everything that Bayseans talk about is wrong in the right context that actually does make sense, but they often apply it to context where it makes no sense at all. Unfortunately, an assessment of Bayesianism is beyond what I can do in this podcast at this point. Now, I might imagine a criticism to me that says something like this, but Bruce, can we take everything you said and put it into negative terms, and wouldn’t that be technically more correct? For example, instead of saying, we built a time machine and we verified the existence of other worlds, and therefore this supports MWI as the correct interpretation of quantum physics, we might instead just rephrase that as, we had an experiment that severely tested the implications of MWI, and it survived the test, and now even nascent competitors to MWI are dead. Now, I do actually agree that you could at least technically take positive language like support or corroboration, and you could figure out some way to reword it into more negativist terms, and it would be somewhat awkward in a lot of cases, and it might require quite a bit of explanation in some cases to be able to fully explain yourself, but I do think that you could at least in principle do this in most cases. I think everything I argue can be put into negativist terms. A corroboration is just a severe test that came out positive after all.

[02:59:58]  Blue: If you want to simply do away with the word corroboration, and you choose towards everything in terms of negativist terms, and you don’t mind how awkward that is going to sound at times, then sure, you can do this. That was why I was trying to show how every single one of these examples could map back to a negativistic way of looking at it. But is it more technically correct to word it in negativist terms? Well, sometimes maybe, but it does not seem that it is always so. The fact is, is that most people seem to come away from a statement like, you learn nothing from a corroboration, and they take that too literally. They think that means that a corroboration teaches you nothing at all. When in reality, corroborating tests may teach you very important things. The fact that you can then reword those things into negativist terms, if somewhat awkwardly, doesn’t mean that a corroborating test taught you nothing. Yes, when we verify that hafnium exists, we also refute the theory that hafnium doesn’t exist. Yes, if we verify the existence of other worlds, the thing we learned was that MWI just survived a severe test and that the implications of that part of the theory are now severely tested, refuting nascent competitors that were trying to capitalize on the fact that that part of QM had never been severely tested. The issue here is that within the context of these examples, wording it positively or negatively, they both mean the same thing, because they’re logically inverses of each other. Or rather, I should say that in the context of a crucial test between two theories, they become logical inverses of each other.

[03:01:35]  Blue: That’s why we shouldn’t fear positively worded terms like strength and support so long as they’re within a context where they can be easily, if awkwardly, rephrased into negativist terms. For when we insist on not using such terms, we quickly start to misunderstand things, like the fact that corroborating tests often do change the state of the critical discussion. Therefore, trying to say you learn nothing and not qualifying what you mean is technically incorrect. So that was three hours of a lot of different thoughts. Let me now pull all of this together and try to make my point more clear. Let’s start again with the example Brett uses of the Eddington expedition, not truly corroborating general relativity, but instead was really only a refutation of Newtonian physics. Brett argues that since all a corroboration in this example really is, is a refutation of Newtonian physics and that it tells us nothing about general relativity, that we don’t need the concept of corroboration. But as we’ve discussed, many assumptions we’ve made here don’t seem to prove out. For example, it isn’t the case that the Eddington expedition had any chance at all of refuting general relativity. Had they failed to find the bending of light general relativity predicted, it would not have refuted general relativity due to the poor quality of the instrumentation available at the time. In fact, this is also true of Newtonian physics. Mayo points out that even if you see the Eddington expedition as falsification of Newtonian physics, which it most certainly is, because of how out of focus the photo plates were back then, they could only probabilistically measure the bend of the light and assign margins of errors or p -values to it.

[03:03:18]  Blue: But the margin for error used at the time, according to Mayo, quote, wouldn’t be considered much of a margin of safety these days. In addition, Eddington tossed out some of the observations that by his own admission matched the Newtonian predictions. Mayo argues that Eddington actually did this for good reasons because, for example, he had seen that the heat of the day had warped the mirrors, things like that. But this led to an accusation that the Eddington expedition was invalid because it had tossed out observations that had come out in favor of Newtonian physics. So let’s make a hypothetical thought experiment. Let’s say that by some weird chance, someone without knowledge of general relativity just happened to decide to measure the stars around the sun during an eclipse and that the result was identical to the Eddington expedition showing that the stars had moved but only probabilistically based on how focused the plates were. Would this have refuted Newtonian physics? Absolutely not. Scientists would have rightly pointed to the fact that the effects had a poor margin for error and that Eddington threw out observations that matched Newtonian predictions. The observations taken in the Eddington expedition alone would not have refuted Newtonian physics at all. In fact, it wouldn’t have even have been seen as an anomaly to Newtonian physics. It would have been assumed to be something wrong with the background knowledge, particularly the experiment itself. So the observation of the Eddington expedition correctly wouldn’t have been considered even a true counter -example or anomaly of Newtonian physics but for the fact that it was taken as a contest that was a crucial experiment between Newton and general relativity. Now let’s strengthen this thought experiment.

[03:05:04]  Blue: Let’s say that in this hypothetical that they used radio astronomical equipment that is far more precise and when they happened to take these observations would this then have refuted Newtonian physics? Well, it would have been seen as an anomaly. That is certainly true at that point. But still no, it would not have been seen as a refutation of Newtonian physics. Why? Because of the reason that Deutsch has explained, because to pinpoint that the theory itself is the source of the counter -example or anomaly requires a second explanation. In this case, general relativity, to explain what is going on. More accurate measure would have been perceived as an anomaly. It would have been seen as no more an anomaly than say Jupiter’s orbit being wrong, which turned out to be a missing planet. Then this is because what Popper calls a refutation is really always a refutation of the theory plus the background knowledge, not of the theory. You need a second theory to pinpoint that the problem was the theory. So let’s take Deutsch seriously here for a moment. You must have a second theory and you can never really refute a theory without a second theory. Let’s say that again. No refutation without a second theory. So when Brett says Newtonian theory is falsified by the observations of the Newtonian expedition, this really isn’t right. Newton’s theory was falsified by a combination of the Newtonian observations plus the existence of general relativity. And the debt to the existence of general relativity goes deeper than even Deutsch is suggesting here. Realistically, we’d never have even made this observation at all, but for the fact that general relativity was making a different prediction from Newton and led us to go try to make this observation.

[03:06:46]  Blue: This is why the fact that the stars moved during the eclipse went unnoticed by humans for apparently thousands of years. Also, consider what would have happened had the observations came out matching neither theory, assuming here we’re using the more accurate radio astronomical devices. It would have probably have doomed general relativity, which needed an experiment in its favor to move from the status of possibly just fitting the data to, oh gosh, this theory might be correct. But it probably wouldn’t have done nothing like that at all to Newton’s theory, as I just explained. It wouldn’t have refuted Newton’s theory. It would have been assumed to be a problem with the background knowledge, similar to Jupiter’s orbit being wrong, according to Newton’s predictions. And the reason is quite simple. It’s because you actually do need a second theory to be able to see that this counter example, even if you know it’s a counter example, is due to Newton’s theory. Namely, you need a second theory that makes a better prediction before you can start inferring that. So when we say that the Eddington expedition corroborated general relativity, the concept of corroboration encompasses both the refutation of Newton, as well as the fact that it was general relativity itself that, one, caused us to collect this observation. Two, interpreted this observation as an anomaly in the first place. And three, allowed us to pinpoint that the source of the anomaly was Newtonian physics and therefore declare it to be false. So I am arguing that the concept of corroboration is a more encompassing concept than the mere idea of refutation, particularly if you’re looking at Popper’s idea of refutation, which is always a refutation of the theory plus the background knowledge.

[03:08:23]  Blue: You simply can’t separate the crisis that the Eddington expedition caused Newtonian physics to experience from the support it gave to general relativity. You cannot simply reword this as a falsification without losing part of what took place. And this is why I believe that the concept of corroboration is a central and necessary part of Karl Popper’s critical rationalism, his theory of knowledge. 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. The first is via our podcast host site, Anchor. Just go to anchor.fm -4 -strands -f -o -u -r -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 4strands.org. There is a donation button there that uses PayPal. Thank you.

---

Links to this episode: Spotify / Apple Podcasts

Generated with AI using PodcastTranscriptor. Unofficial AI-generated transcripts. These may contain mistakes; please verify against the actual podcast.