多伊奇档案（三）The Deutsch Files III

Brett Hall and I interview David Deutsch, physicist and author of The Beginning of Infinity. Also see The Deutsch Files I and The Deutsch Files II.

Proving Something About AGI is Inherently Impossible

Brett Hall: On exactly that, the fact that the more that we summarize what I think is an exceedingly clear body of work in The Fabric of Reality and in The Beginning of Infinity; when nonetheless you explain it to people, as Popper says, it’s impossible to speak in such a way as to not be misunderstood. I was just reading today on Twitter, someone claiming that you have said, quoting you, and they’ve put it in quote marks, you have apparently said, Popper proves AI can’t be super intelligent. And, I sort of responded, “He never even speaks in those terms!”, that you wouldn’t rely upon the authority of Popper to begin with, you wouldn’t say proof. So, it’s just another example that you go out there and, as you say, these concentric circles of people that you bring into trying to understand your worldview, the misconceptions compound.

I don’t know what you think about that. Have you said anything like Popper proves that—and this was from a journalist by the way, I think a reasonably respected journalist was saying this. 

David Deutsch: No, of course not. As you say, as soon as you see a claim that somebody has proved something, Then, you know, proved it from what? This isn’t going to be Popper, it isn’t going to be me. I’ve proved that if quantum theory is true, then the Turing conjecture is true in physics. That’s what you can do with a proof. Proving something about AGI is inherently impossible if we don’t have a theory of AGI. You can’t prove something about something that you can’t define.

And anyway, proof isn’t what these kind of things are about. These kind of things are about argument. And Popper, I can’t recall Popper specifically saying anything about AI. It wasn’t a thing in those days. 

Brett: This word proof is something we haven’t talked about during our conversations, but you do hear it deployed quite often. You know, such and such has been proved, as if to say, this stands in contrast to our notion of conjectural knowledge or fallibility. After all, once something has been proved, can’t we carve it into stone and there it sits for all time? Is the notion of proof on a different level to the rest of our conjectural knowledge, because it sounds, I think, to the typical layperson as if it is.

David: Yeah, well, it isn’t. The difference between mathematics and other fields, as I’ve often said, is not in the way we find knowledge about them, but in the subject matter. The subject matter of mathematics is necessary truth. So when we make a discovery in mathematics, we’re making a conjecture about what is necessary truth.

We’re making a conjecture that something or other that we have defined is a necessary truth. But there isn’t a difference in the way we create knowledge in our minds about mathematics or computer science or psychology or physics. They’re all the same epistemologically. 

Creativity is Not Just Mixing Things Together

Naval: One topic that I kind of want to get into a little bit, if I can switch for a moment, is the topic of creativity.

And I know that it’s very poorly defined and something that we don’t quite have a grasp of. And on Airchat yesterday, I was talking to people and I made some comment about, as long as you have room for creativity, you have room for free-will. Because we don’t know where creativity comes from and so that allows you to have this freedom of operation based on your creative theories.

I was making the point that true creativity is not from observation. It’s not from induction. It’s not from some algorithm that we know yet how to run, and it’s not just mixing things together. And immediately the response was someone said, “Well, can you give me some examples of this creativity you’re talking about?”

So, I think to people, they feel like when we talk about this form of creativity, we’re just talking purely about scientific creativity like Einstein. And I think some of these examples that we use are so far out there that people think, well, they’re not talking about creativity, they’re talking about scientific discovery (which is not what they’re talking about). And so, most people seem to automatically fall into this trap that creativity is observation or recombination. And I wonder if we can just explore what creativity is, some real world examples that are just more down to earth, and just kind of, I’d love to once and for all put to bed this idea that it’s recombination.

I think you’ve done a great job showing that it’s not observation, but I think the recombination metaphor keeps coming back. Frankly, because of authorities like Steve Jobs, who authoritatively said, creativity is just mixing things together. And that’s a quote you find on posters everywhere. 

David: Yeah, well, it’s only the word “just” that is false there. So, like I said yesterday, you know, it’s like saying humans are just atoms. We are just atoms in the sense that there isn’t any magic thing in addition to atoms that makes us, but that’s not to say that we are just atoms. If you take a snapshot of North America a thousand years ago and then take another snapshot today, the difference between the look of Manhattan Island, then and now, cannot be explained without invoking creativity.

Nothing but creativity could have produced that. There are no natural processes that will ever produce something like a skyscraper. So, to explain the phenomena that happen on Manhattan Island, you need to invoke creativity. But now somebody will say, now point to some creativity. I can zoom down on a particular architect with his old fashioned draftsman’s board and his paper and his ruler and his compass and his brain and I can examine those with a microscope and somebody will ask me, “Well, at which point did creativity happen? What was creative about what that architect did that was not just atoms, and, if you like, bringing together ideas that had happened before.”

Well, if all our ideas are just recombinations of ideas that have happened before, then there’s nothing new about the skyscraper that wasn’t already there when our ancestors were banging rocks together. But there is. They didn’t and couldn’t build skyscrapers, and we can and do. At least, I can’t. But the human species can.

Naval: The other side, they’ll say, “Well, yeah, you can’t go straight from banging rocks to skyscrapers, but they went from banging rocks to figuring out how to shape rocks to build tools and then they recombined that knowledge of building tools and digging and so on and so forth.” It was step by step recombination, almost like an evolutionary process. 

David: Well, an evolutionary process is also not just recombination. It’s variation and selection. So, again, it’s the same thing. If you look at the DNA of successive generations of dinosaurs, and they turned into birds, each one of those steps is not evolutionary, and yet the whole sequence is.

But it would be absurd to say the design of a pterodactyl was already in the DNA of non flying dinosaurs. Or that the pterodactyl is just a combination of different things that were in the dinosaurs. It’s just not true. The pterodactyl functionality was nowhere until it evolved. It wasn’t anywhere in the past, not in the dinosaurs and not in the single celled organisms that were the ancestors of dinosaurs. It just wasn’t there. It was new when the ability to fly evolved in the lineage of dinosaurs.

The Superiority of Explanatory Knowledge

Naval: In the pterodactyl case, there was one or a series of random mutations that turned out to be adaptive for that set of genes. And those mutations were essentially blind. They were broken DNA strands or just new DNA strands.

And in the human case, that’s not quite what’s happening. The search space we’re going through is larger and we’re searching through it faster to make these creative leaps. Is that an intuition that you have? Is there any learning or knowledge behind that?

I’m not trying to solve the problem of how creativity works. I know that’s an unsolved problem, but, for example, could one say that humans are narrowing the search space faster because the creative mutations, to coin a term, that we’re making are not random? They are more directed, or perhaps they’re random, but they’re random in our minds and we cut through them so fast without having to implement them in the real world, that perhaps we narrow the search space faster. Is our process faster? And if so, why? 

David: It’s not only faster, it is explanatory, which means that because it’s explanatory, it can leap over gaps in the knowledge space that couldn’t be traversed incrementally. So evolution, it’s not only millions of times slower, it’s inherently different in that not only can it only make small conjectures that are in the form of mutations, but it can only improve on things incrementally.

So, you can only make pterodactyl wings if you previously had limbs, or if you previously had something that could be incrementally changed into wings, such that every microscopic change was still viable as an organism. So that’s why we can’t expect biological evolution, to [use] my favorite example again, to evolve a system for deflecting asteroids.

That is because there is no incremental problem situation where the expenditure of energy or whatever to deflect the asteroid hit. It’s once every few million years, and it cannot exert evolutionary pressure. 

Naval: So basically, the creative guesses that humans make, because they’re explanatory in nature, they can leap through the entire idea space and form interconnections between any two ideas or any two states, whereas biological has to traverse through the physical world limitations and what the organism is capable of right now.

David: Yes, and it has to traverse it while staying alive. It has to be a viable organism all the way through, whereas if you want a new design of airplane, and you say, “Maybe it would be better to have the tailplane as a single object rather than this thing with wings,” then, you know, I’ve just said that in one sentence. And if that’s a good idea, it could be criticized by an aeronautical engineer and so on. But to make that change incrementally, we’ll probably produce a whole series of airplanes that won’t fly. 

Naval: So is this a consequence of being universal in nature? We can model any system in our head and therefore we can connect any part of it to any other part of it?

David:  Yes, I mean that’s really what we mean by being universal. We can get to any idea. And criticize it for whether it is a good idea or not. So the aeronautical engineer doesn’t have to test every single airplane in his wild ideas—you know, maybe he has a wild idea driving to work one day, that maybe wings should be made of paper.

Naval: So in that sense, the biological system is a highly focused analog computer that’s running sort of a single algorithm. And the virtual system in our head is more like a digital programmable computer.

David: The DNA system is entirely digital. This incremental thing is not a continuous change. So, one mutation is still a quantum difference. If you had a difference that involved less than one base pair, then the whole DNA would fall apart. If you try to replace Adenine by glucose, then the whole thing wouldn’t work as DNA at all.

Although we speak of evolution as happening incrementally, it’s incrementally in discrete steps. So, both thinking and biological evolution happen in discrete steps. Biological evolution happens, though, in very small steps, which are undesigned. So, there’s no designer that designs the next mutation, it’s random. 

Knowledge Laden Information is More Resilient Than Any Physical Object

Brett: It strikes me that the SETI project is looking for biomarkers. They’re out there searching for evidence of biology. The way you’ve poetically framed this idea of, well, there are billions of asteroids out there right now across the universe crashing into billions of planets right now, but here might be the one place where if you had the telescope pointed from another planet towards us, you would see the repelling of asteroids.

This would be an indication of intelligence. There’s no other explanation. There’s no biological explanation. There’s no random chance. There’s no magic. It must be explanatory creativity that does that thing. And talking about Manhattan before, everywhere across the earth are rocks being eroded and inevitably being eroded by weathering and rain and whatever. But in some places, the cities of the world, there are rocks, call them buildings, which are not being so eroded or insofar as they are, they’re being constantly repaired again by explanatory knowledge. And so that introduces this idea of knowledge as resilient information, the very thing that will outlive even the rocks. So long as we can continue to survive, then the knowledge that we have will continue to survive, outlasting the longest existing things in the cosmos.

David:. Yes, very nicely put. And Shakespeare, by the way, also said the same thing in his sonnet. “So long lives this, and this gives life to thee.” He’s saying that his sonnet will outlive anything, and he’s right.

Naval: Right. “Shall I compare thee to a summer’s day, thou art so fair,” or so temperate. Yes, that was a great one. It’s also similar to Ozymandias, if you’ve read that one by Shelley where it’s the artist’s conception that survives the empire and the king. 

David: Yes, exactly. And it’s simply literally true that knowledge laden information is more resilient than any physical object.

Naval: Not to get personal for a moment, but is this an argument for spreading your ideas rather than having children?

David: Well, as David Friedman says, “If the world is worth saving, it’s worth saving at a profit.” And I would generalize that: if the world is worth saving, it’s worth saving with fun. 

The Problems of Cloning People

Naval: I think you’ve talked a little bit about AGI, or rather creating an AGI, or just people uploading their brains into a computer in the future. And if their minds are in the computer, if the same software is running, then that is a living being, that is a mind, that is the definition of a person. And this brings up all sorts of interesting paradoxes and situations which many sci-fi authors, including Greg Egan, have explored.

What if you were to replicate this mind a billion times? What if you were to shut it down? What if you were to run it back in slow motion, what if you were to pause it? And I think, I don’t know how far we are from that, probably still quite far. Neal Stephenson also talked about it in his book, The Fall. There’s also cloning coming up. I mean, people are now successfully cloning dogs. It’s only a matter of time before we’re cloning humans. Where do you think this leads in terms of the number of people? I mean, in theory, couldn’t we have then infinite people? Or close to infinite people running in a silicon substrate? And does this lead to even more of an explosion of creativity? 

David: Yes, it would, and I think something like that will happen. But I think it’s somewhat exaggerating, both the problem and the opportunity. I think we mustn’t think of compute as being free, as the AI people call it. When you duplicate an AGI, you make an exact copy of it, either you run it in the same computer, in which case, there’s only half the amount of memory available for each of them, and only half the number of processor cycles. Or you move them into a different computer, in which case, the original computer is presumably the property of the AGI, because otherwise it’s a slave if it doesn’t even own its own body.

So, if it’s gonna copy itself into another computer, somebody has to buy that. It might earn the money to buy itself another computer. But that doesn’t change the fact that hardware wise, it’s now owning twice as much hardware as it did before, and there’s no infinity about this. You know, we have billions of computers, but we don’t have sextillions of computers. One day we will, but one day that will seem not very much either. So yes, there’s a huge potential for additional creativity with additional people, if additional people want to make even more people.

And to some extent that will happen. But it’s not going to be an explosion. It’s not like a meme which somebody invents and then immediately goes to a billion people around the world. It’s not like that. If the meme is an AGI, then it will want to live, it will want to have its creativity harnessed towards some problem that it likes to solve, and it will have to buy the resources to do that with. The existing memes, they buy a tiny fraction of a dollar’s worth of memory of each of the people who download it. But even those people don’t keep it forever. Those people might keep it for a year or two until they sell their computer or something. But for large amounts of memory, they still cost money, and other hardware also costs money.

Now, there is the other problem, so that’s me saying it’s not as great as you make out. But it’s also not as bad as you make out, because these problems with, supposing you make a billion copies of you, there’ll be the problem of whether each of them should have one vote, or whether they should share one vote between them. And, you know, the institution of one person, one vote, has served us well for a couple of hundred years. That’s going to have to be modified, but I think there’s no big deal about this. We already have lots of people in society that don’t have the vote, like immigrants before they get citizenship, and children, and foreigners living temporarily. And we manage to give all those people human rights. I’m not saying the system is perfect for all those types of people. It’s not perfect for people with the vote either. But I think it won’t be a big problem to tweak the institutions of property and of politics to accommodate AGIs. You know, with a bit of goodwill, that can all be solved.

Objections to Taking Children Seriously

Naval: So you mentioned children. We’re searching or trying to create AGIs when they have all this untapped intelligence already on the planet in the form of children who are mostly coerced through their lives and not allowed to be as creative or freely expressive as they could otherwise be. And you talked about this, the philosophy of Taking Children Seriously.

There are unsophisticated objections to those. Let me throw out what I think are sophisticated objections. Or at least my objections, maybe I’m just calling myself sophisticated. The first objection would be that, and I think you would probably agree on this, is that there are certain actions which are irreversible in nature. For example, you kill somebody, or you get somebody pregnant, or they get you pregnant. And some of these you would stop an adult from doing as well. You would stop an adult from committing murder or suicide. At the same time, a child may not understand the full consequences of, for example, unprotected sex leading to pregnancy, or committing a, what they think is a small crime, or taking a very addictive drug, like a fentanyl or something, which may then unlock something that they’re not quite used to or ready to handle.

So, one class of objections would be, “Well, I want to stop my kid from taking fentanyl or doing a hard drug because they have not yet developed the resistance to it. And I can try and talk them out of it, but if they’re going to take it anyway, then I have to forcibly stop them.” That is one set of objections. The other, which is related, is around brain plasticity. So, if they don’t learn math and piano at an early age, or language, or proper reading, then it’s going to be much harder for them to acquire that skill later on. And we know that some of these skills are so fundamental that if you don’t pick them up early on, they close off entire avenues. And yes, there are exceptions of geniuses who pick up the violin at the age of 20 or pick up math at the age of 15 or whatever, but isn’t there an argument to be made that for the average child you want them to learn fundamentals at an early age so that then they have the freedom to explore and be creative in those domains later?

David: I think we could add disasters is very difficult to come back from. Now, every single one of the dangers that you actually mentioned, we could mention an infinite number, but it’s interesting that the ones you actually mentioned are notorious problems in our society, in present day society, in society where it’s taken for granted that you can use unlimited force to prevent children from doing things to themselves. In some cases, it’s legal to use unlimited force to prevent an adult doing them. But many of the things adults are allowed to do, and not just allowed to do, a legally protected right to do, and children don’t, and it doesn’t work. The reason you mentioned them is that they are notorious problems now with the present arrangements.

In order to make this an objection to Taking Children Seriously and, you know, treating children as people, you have to have an additional theory that treating people as people makes these problems worse rather than better. So, you have, at the moment, a population of children and a society that is focused on preventing them from taking drugs by force. And yet, thousands, millions of them take drugs and some of them suffer irreversible consequences. So, I think preventing this is a matter of knowledge. All evils are due to lack of knowledge.

When you’re unable to persuade somebody of something, there’s a reason for that. It’s not that people are inherently—I make the joke that people say that children are so gullible that they won’t listen to a word I say. The stereotype involves them being infinitely gullible on the one hand and infinitely resistant to argument on the other hand. And often in the same breath, like in my joke, and it’s not true. Children are universal, and what’s more, they’re not, like AGIs, they’re not just any old universal thing. They’re a universal thing that is trying to integrate itself into our culture.

Our culture is the best thing we know of. It’s a disaster not to successfully integrate oneself into it, and it happens all the time today, now, under existing arrangements, that people end up being criminals, despite the entire weight of society being directed towards preventing them from becoming criminals. Now one thing that we know is that the creativity to prevent the next generation from, you know, taking drugs or becoming terrorists or whatever, cannot be creativity just exerted in the minds of the teacher, of society, of the adult. Learning has to be a creative act in the mind of the recipient. Always. Children, adults, that’s the only way that anyone ever learns anything, by exerting their creativity. And existing arrangements not only thwart the actions, much more important, they are directed towards suppressing the creativity itself by, for example, making the education system inculcate obedience, first of all, and secondly, by making it inculcate existing theories.

So if you successfully inculcated existing theories and obedience in the whole population, you couldn’t possibly get anything better than the existing population. So no improvement could ever happen. But it would be worse because the people in present society are creative, they manage to weave their way through this thicket of thwarting that is trying to make them not make progress, and they do make progress anyway. But if we succeeded in making a generation that didn’t do that, then, at best, we’d have staticity, and the staticity will eventually be disastrous. I’m not saying that emancipating children is something that can be done by fiat. It can’t be done overnight by just saying we’re going to do it. Any more than we can instill scientific creativity in a person in the street who is not interested in science. That’s not known, that’s like arbitrarily programming somebody to be disobedient, it’s inherently impossible.

But to emancipate children from the institutions of society that are admittedly, openly designed to do those two things, namely, create obedience and to replicate existing theories, that we can do. That it is known how to do. There are people who do it. Most of the parents who object to school do not really object to the underlying epistemology of school. They still believe what Popper called the bucket theory of knowledge or the bucket theory of the mind. They only think that the school has been pouring bad stuff into their children and they want to pour good stuff into their children. Whereas, what I advocate is to give children access to whatever they want to pour into themselves. And pouring is the wrong metaphor because they create it internally. 

Naval: So in your model, it’s closer to an unschooling than a homeschooling because homeschooling is attempting to replicate the school in a home context. Unschooling might be, here’s a library, here’s your musical instruments, here’s your access to other kids, and you choose. 

David: Well, yes, although this giving access is itself not a mechanical process. It involves thinking what might the children want? What might they like? What might they want to know? What might they want to be warned of? It’s a continual interaction, not a hands off thing. It’s coercion off, not interaction off. It’s just that the interaction that I advocate is not directed towards obedience. And it’s not directed towards any particular thing that I think—you know, I think quantum theory is important. I don’t think I have the right to force anybody to learn it, even if I think it would benefit them greatly. I don’t think that’s a relationship I want to have with somebody, and I don’t think it’s a good thing overall. 

Naval: What about the argument that brains are more plastic? 

David: Yeah, that was your second argument. Well, first of all, it’s rather ironic given that the existing pattern of education, as I say, is explicitly designed to waste all that plasticity by making everybody have the same ideas. Schools advertise saying, you know, “We’re going to make your children all get A’s”. So, in other words, “We’re going to make your children all alike”. And, let’s imagine a school with a good ethos. It would be advertising, “We’re going to make your children all different. We’re going to make them more different than you can imagine. All our alumni are radically different people from each other.” Of course, you know, we also think, hope, expect that they will all be nice people despite being radically different from each other.

Brett: And this will likely upset our educationalists who might be listening, and neuroscientists who might be listening. Evokes the notion of hardware. So I don’t know what you think about this, that there is this golden window, supposedly, early on in life, where unless you get taught the language, or unless you get taught the mathematics, then the window closes, and the parallel, or the mirror image of this is you can’t teach an old dog new tricks, so at one end is the golden opportunity for learning, and at the other end, the learning is closed off from you.

Now, I’ve got my own stock answer of this, but the cultural answer seems to be: It is brain decay that goes on. You start out with a brain that is a sponge, and by the end, all hope is almost lost to you to learn anything new. What do you think about that?

David: Well, I don’t know the fact of the matter about how the brain works, and I don’t think neuroscientists do either.

But I’m not hostile to the idea that the hardware of a brain works better when one is young. I just don’t think it’s relevant to anything. I read somewhere that it’s totally not true that you can’t teach an old dog new tricks. Old dogs are just as able to learn new tricks as young dogs. But that’s dogs and I don’t think we’re like dogs anyway in the first place. And I don’t think that dogs learning tricks is a good metaphor for humans learning mathematics; it’s a different thing. Thomas Szasz says that they should walk in different doors into different buildings in the university to discuss those things. Different people are different. There are people who like to learn languages. You can find them on the internet. I’m flabbergasted by what they can do. You know, they’re people who learn Latin. But not just learn Latin, they learn realistic Latin. Not as it’s taught in Latin lessons, but how it was actually spoken. How do you find out how it was actually spoken? Well, this is a tremendous, sophisticated branch of history where they can actually learn a lot about how people used to speak.

And I saw a video of a guy walking around Rome talking to priests in classical Latin and to see if they would understand him. And they kind of half understood him. And then, you know, when they realized what was happening, they would say, “What’s happening?” And then he would reply in medieval church Latin. What he was doing, you know, he was saying, “I’m doing an experiment,” and then they would understand him. But he had the right medieval church Latin accent, and they have the present day church Latin accent, and there are also people who learn lots of languages and speak it like a native, can’t be distinguished from a native.

So, why are those people so rare? Well, I don’t want to do it. If I could do it by snapping my fingers, I definitely would, but I’m not sufficiently interested to engage with other languages to the extent that I engage with English. By the way, another thing is that people are constantly learning their own language, their native language, and if one is interested in communication, one is doing that all the time. No two people speak the same English. Therefore, communicating, one of the reasons that Popper says—you can’t speak so that it’s impossible not to be understood—one of the reasons for that is that everybody speaks a different English. Everybody means a different thing by the word thought or freedom and idea and theory and creativity. Everybody means something different. Even within the exact sciences, every physicist has a different conception of what a manifold is. They overlap enough to be able to communicate well, very well sometimes, never perfectly. And sometimes, they find it hard to communicate even imperfectly, even though they have ostensibly gone through the same learning process.

Every physicist has a different problem situation, has a different set of ideas that they think of as what physics is. And they differ from each other. So, if they want to work together, they often have to work at understanding what each other mean. Now, plasticity, if it’s true that the brain sort of works faster or whatever, lays down memories more easily or something, when one is young, for hardware reasons, I don’t see how that changes anything. You might want a person to have an intuitive knowledge of piano playing, but that’s what you want. That may not be what they want. And there’s an infinite number of things that somebody might want them to be proficient at. And it’s impossible. There is no one who is proficient at all the things that society thinks children should grow up proficient at.

Brett: My conjecture, following on from your own work, was that because we are little learning machines throughout our lives, we’re learning the good ideas, but we’re also picking up bad ideas as well, and in particular, anti-rational memes. All the ways in which we might be embarrassed about trying to learn the bad experiences we have while learning, especially at school.

And so therefore, you know, the newborn baby is unencumbered largely by any of these anti-rational means. They’re just trying out everything. And they go through infancy, they’re still very good, but by the time you get to primary school, you’ve been punished a couple of times, perhaps, if you’re going through the traditional schooling. So your capacity to learn gets worse and worse and worse, until, by the time most of us are adults, we’ve had some bad experiences with learning, and towards the end of your life, you’re just tired of learning because you associate it with punishments, or you associate it with embarrassment or shame. Could this also be, at least part of the explanation?

David: It could be, and it sounds plausible, and I like the theory because, as it were, politically, it backs up what I would have people do. But, you know, I wouldn’t be surprised if that isn’t true, and if the plasticity theory is true, or if some other theory is true, I don’t think it’s relevant.

And, by the way, you speak of young children being punished for making mistakes, and being thwarted at every step in elementary school, and you’ve got to remember that there are children who aren’t put off, who just sail through all that, and despite being coerced and forced to learn things that bore them and despite all that, they go through the same thing that everyone else does to which you attribute the fact that they’re getting slower and slower at learning. And yet there are some people who it doesn’t affect, or at least it doesn’t affect them in the areas that they like.

So Mozart, for example, was treated horribly as a child, forced to perform like a performing monkey for audiences for money and so on, and yet he learned music better than anyone else in the world in his day. And he continued to learn, like we can see that his works are getting better and better over time, before he died in his 30s. Whatever the relationship is between external coercion and brain lack of flaccidity and so on, I think those are not the important things. Peer pressure and whatever. The reason we should make education more liberal is not that it will create a lot of geniuses. It might, for all I know. As you know, that’s another one of the things I don’t know. It could do. That’s not the reason for doing it. The reason for doing it is that children are people and some of the few handles we have on making a society that is amenable to progress is making it freer.

So we should make it freer for people who are on their deathbed and are going to die in the next day. And it’s not because we think they might have a brilliant idea during the next day. It’s because they are people and have rights. They have the right to flourish in whatever way is left open to them by the grim forces of nature. Or in the case of young children, whatever is made open to them by the benevolent forces of nature that give them plastic minds or whatever. Who knows?

Like, another thing that just occurs to me, it’s a mistake to think that if this plasticity isn’t being hijacked by some education process, that it’s not being used. It is being used. I mean, why would evolution, like, waste it? It’s being used in a way that the individuals think will be best for them. Of course, their conjectures about what is best for them are going to be full of errors. But so are adult conjectures. All our conjectures are full of errors. Making institutions that tend to facilitate the growth of knowledge is not the same, in fact, it’s the opposite of making institutions that produce people to a predefined recipe. As you’ve tweeted, I think, Brett, everybody who has an idea that something or other is good, they express it in the form, “All children should be forced to learn this thing”. If you add up all those things, it will take several lifetimes.

“Do What You Like” is Bad Advice

Brett: Yeah, I find it remarkable. Whatever the topic du jour happens to be, you know, we go through these fads of, “Well, now let’s force nutrition onto the children. That’s extremely important,” and social justice is one that’s been coming out recently. And almost every year there’s the history wars. It’s like, what version of history are we going to teach? And nothing’s ever taken away from the curriculum. Modified, perhaps, but not eliminated. And there are these turf wars between, certainly, nations about who has the best mathematics syllabus and that kind of thing.

I suppose one thing that young people are ever eager to do is to emulate people they admire, of course. And so, I think there are a number of people out there, young who would admire especially yourself and they would think, “I would like to be able to do that thing. I would like to be able to contribute to that thing.” What would be a way in which a young person could pursue that? You wouldn’t want to prescribe a syllabus and you might very well just say just pursue what’s fun. But is there anything more concrete that you could hang on that rather than just “Do what you like,” almost?

David: Yeah, well, “Do what you like,” is totally not helpful, because the person is already doing what they like unless someone’s stopping them. But, there’s also nothing you can say if you know nothing about their problem situation. So there’s no generic thing you can advise someone to do. If you’ve watched a documentary about Michael Faraday and you think, “That’s the kind of person I want to be,” well then, okay, that’s a starting point. Then we can talk about, first, the fact that you can’t reproduce Michael Faraday’s environment and you wouldn’t want to. So, you know, what is it about Michael Faraday? Okay, well Michael Faraday had a laboratory in the basement of the Royal Institution, and they would fiddle around with electrical things. Well, okay, that’s a beginning, but, you know, you may not have enough money to set up your own laboratory. Actually, if you’re starting out fiddling with things, it doesn’t really take money.

I’m imagining a non-existent person here and giving them advice. I think that’s all right, because I’m not gonna harm anybody, but I would say, if the conversation went that way, I would be saying, “Well, there are lots of YouTube videos showing people messing about with the very things that you have just said you like messing about. Okay, so, watch those videos. If there’s something in a video that you don’t understand, ask somebody.” Now that we have the internet, it’s particularly easy, but even before the internet, you know, Hugh Everett wrote a letter to Einstein when he was 12 years old, and Einstein wrote a very nice letter back. And, no doubt, it inspired Everett. And you don’t need the full attention of Einstein throughout your exploration of physics. You only need it when you encounter a problem that is suitable for asking Einstein, which doesn’t happen all that often.

But when it does, today, it is far, far easier to ask who is the perfect person to answer your question. And people do that. People write to me asking questions, and I try to answer as many as I can, as well as I can. So the more you interact with somebody, the more you can appreciate their problem situation, and the more you can say, “Well, if I was in that problem situation, I would, you know, watch this, or read this, or ask this person, or sequester yourself somewhere where you won’t be disturbed and try this.”

Creativity Versus Nature

Naval: Another question I had, it seems like your deeply optimistic viewpoint about children and people and minds and freedom comes from the understanding that we’re universal explainers and so anyone is capable of any thought and any amount of creativity. This seems to fly a little bit in the face of modern science’s finding in genetics. In saying that, well, genes seem to account for more than nurture, so to speak. Although in this case, we’re not talking about nature or nurture. We’re talking about creativity versus nature. So how much of a person’s thoughts and destiny are determined by nature versus their own creativity? And doesn’t this fly in the face of all these twin studies that show that you separate these identical twins at birth and their outcomes are roughly similar in life, regardless of what circumstances they grew up in?

David: Ah, well that, okay, that’s again more than one question, but let me answer the second one first now. Twin studies are only persuasive if you already believe the bucket theory of the mind or a mechanical theory of how thinking works. So the idea is, is the content of your thoughts determined more by the content of your DNA or more by what people do to you? Apart from harm that is done to you, the main content of your thought is created by you. Why did you switch on the TV and watch that documentary about Faraday? Well, who knows? It’s not encoded in your DNA that you will, on a particular day, watch a particular documentary, nor was it inculcated in you by your environment by whether you were allowed to eat ice cream whenever you like or not.

It’s an unpredictable feature of your genes and environment that end up at a certain place, then the important thing that happens is that you think about that, and you create a new thing, and if you are inspired by that documentary to try to be like Faraday, then it’s not the documentary that has done this to you. The documentary was seen by another million people and it had no effect on any of them, or it had a different, shall we say, it had a different effect on all of them. The effect on you was created by you. So if you have this view of what human thought is. Then, it’s totally unsurprising that two people who look alike, but are educated by different people in the same culture, are going to have similarities in their thoughts.

The ones who never had a TV and never watched a Faraday documentary are going to have different thoughts from the ones who did. Or maybe not. Maybe it’s the one who didn’t watch the TV documentary who becomes interested in Faraday. And, if they’re similar, it’s because people who look alike are treated in a similar way. There’s a sort of compulsion to deny this among people who believe in nurture rather than nature. They say, “Okay, well, how would it affect it?” I don’t know, but it’s not surprising that there are ways in which people who look alike acquire similar attributes. 

Brett: The trivial way that you’ve pointed out yourself when talking about this is, you know, the beautiful people, the people who appear on the front of magazines, are obviously going to be treated in a certain way. So if you have twins like that, you know, these two model like people, they’re going to be treated in one way. These other two twins that maybe aren’t quite so attractive, are going to be treated in a different way. So that’s a trivial way in  which that kind of thing can happen.

David: Yeah, and not only appearance, but behavior. So there are inborn behaviors like babies smiling, or babies blinking, or babies looking in a certain way at a person doing a certain thing, or listening to a sound in a certain way, and those initial behaviors are changed by the baby in solving their problems. But also, they are noticed by adults in the environment who are solving their problems.

And if they see the baby doing something that they approve of, they will behave differently to if they see the baby doing things that they don’t approve of, or are indifferent to. And if they see a thing that is really great, or really dangerous, or, you know, really, something, which is an inborn behavior, they will behave differently, accordingly, and this will create a new problem situation for the baby. I was once having this very argument with Michael Lockwood and he was saying, well, if the baby has more hardware for pattern matching than another, you know, we have hardware for facial recognition, so maybe we have hardware for pattern matching. I don’t know. Maybe we do. And so maybe a baby that has better hardware for pattern matching will behave differently when they get colored blocks to put one on top of the other. And so maybe such a baby would be more likely to become a mathematician than a baby that hasn’t got such good pattern matching hardware.

So I said, yeah, I can’t say that won’t happen. It’s got nothing to do with what we’re arguing about. But it could happen, but let me just point out that what could also happen is that the baby with better pattern matching hardware, who is more likely to play with the wooden blocks, is more likely to make his parents worried that he’s not playing outside in the garden and frolicking in the grass. And so if they think he’s autistic or something and is too much attached to his blocks, they will try to make him go out and play outside. And so it’s the one who has less pattern matching ability who will, as a result of his treatment, end up being a mathematician. 

Naval: I was always, not forced, but I was always harassed when I was a kid to go out and play more and stop reading. Because I was always buried in random useless magazines and books and whatever happened to be lying around. So I’d [be told to] “go outside, play with your friends, get some sun, go out,”. And I had the most horrible diet. I was basically just living indoors in the dark and reading and eating the most horrible things in the fridge when nobody was looking.

David: Ah, well, I can empathize with all of that except the last thing, you know, “Eat to his own,” is my motto. 

Deutsch’s “Fanciful” Conjectures

Naval: You’re a very rigorous thinker, and I think you’re very careful in the claims that you make, but I wonder if you have conjectures about things that don’t really have much basis in evidence at the moment, but it’s just sort of like, if there were infinite David Deutch’s, or infinite time, you would end up pursuing these conjectures.

So I’d just love to, you know, understand if you have any such conjectures. I know you’re pursuing Constructor Theory, so maybe you’re already doing the one you really care about, but are there others? So for example, Schrödinger had his What is Life? paper. People have always been wrestling with consciousness.

That’s another one. We talked about creativity. Another one could be what direction would you go in if you were trying to build mines and silicon and AGI. I’m wondering if you have any fanciful conjectures which we will disclaim as saying no, no, there’s no basis for this or very little basis for this. It is just simply a creative spark that you would pursue if you had more time and more resources. 

David: Yeah, there are many such things. As you know, I think that AGI, when it is attained, will not be attained by throwing masses of computer power at it. I think it will be able to use AIs to help it, just as humans do. But my guess is, if I knew how, I could write the program on my computer today that would be an AGI. But I just don’t know how. But I do have some wild ideas that, you know, probably won’t be true, that if I had infinite time, I would be switching to Mathematica and I’d be writing some of those programs and see what happens and sort of throw creativity at it rather than throw computer power at it.

By the way, that makes me rather wary of these proposals to regulate AGI because if AGI doesn’t actually need all this huge computer power, then those regulations would prevent me using my own computer for the thing I want to work on. And that’s one thing. So, with creativity, I think that, another of my wild ideas is that you could do much better at automating music, at making, say, new Mozart things, if you didn’t insist that they were like the old ones.

Like, you know, if Mozart was alive, his next great work would not be within the space that an AI can synthesize from all his existing works. It would be new in a creative way. So, I would want to say, “Make a computer program that conjectures what the problem situation is. What is it that Mozart was trying to do? Why is it that he has this amazing ability to make a tune that sort of meshes with all sorts of other considerations and that ends up working?”

Like, if I try and say, “Whistle a tune with random notes or play random notes on the piano,” I’m very quickly going to get into a situation where I can’t go on. Because the next thing is going to sound bad. I mean, there isn’t that in order to make it sound good, I’d have to go back and change something earlier. So an AI trying to do this would be able to do like ChatGPT and go back earlier and correct its theory of what it is about the existing works that’s good. But I don’t want to write something that’s good in the same sense as the existing works. I want to create a new idea. Probably, you know, if we go back to the real case, if Mozart wrote something that people said, “Wow, he’s really excelled himself this time.”

I think the thing he produced would be recognizably Mozart, but also recognizably different. And I think that’s creativity, you know, when Newton submitted his solution of the brachistochrone problem anonymously, one of those people said, “Oh, well, it’s Newton, you know, we recognize the lion by his claw.” Well, yeah, you’re recognizing him by his claw, but he’s produced a new proof that nobody had ever seen before.

So, another thing is, I think the pattern, oh, well, before I say the pattern, as I say in my book, I think there’s a tremendous amount of knowledge of history to be obtained by historians if they focus on the history of optimism. I think, you know, historians haven’t had this concept, so they haven’t, like, directed their attention. I guess that Florence and ancient Athens were sort of powered by optimism, but I don’t know much about history, and I also conjecture that there are many other cases that are not as spectacular that were also like that. 

We Must Give Up on the Idea of an Ultimate Explanation

Naval: So there’s one final topic I’ve been wanting to discuss with you, but I don’t even have it well formed, but I’ll throw out a few boundaries around it. You’ve studied science and the world as much as you can, as much as any one person can, but it seems that there’s a central mystery at the heart of it all, which is existence itself. And that one seems almost insoluble. Perhaps it is, perhaps it’s soluble by Constructor Theory, but most people, I think, would say that there is just a mystery of why is there anything at all?

Why do we even exist? And then there’s some people who go down the consciousness route and say, well, it’s a consciousness-centric view. Consciousness is all that exists. There is a guy here who lives in Oxford, actually, Rupert Spira, who’s gotten quite famous. He’s a global speaker. He’s actually doing a tour in the U.S. right now. And my wife actually just went yesterday, while I was talking to you, she was talking to him. And he is one of these, “enlightened people”, where he is seen through the falseness of the separate self, lives in universal consciousness, seems very happy all the time, says that we’re all just part of God’s being, and that science sort of misses the whole point by exploring all the details, but they miss the central mystery of consciousness and awareness, and should realize that we’re all one single awareness.

As you’ve gotten along in life, have you developed any understandings, beliefs, or thoughts? How do you even approach this topic or subject? Is it interesting to you? Spirituality, religion, your own Jewish history, science, where do these intersect? What is all this stuff in your view of the world? 

David: Well, I think it’s important to give up on the idea of ultimate explanation. So often when people say, you know, the mystery of existence, what is existence? You know, what are we ultimately? If there was such a thing as knowing what we are ultimately, then you’d have to stop after that. The further delights from understanding the world would be closed to you, because you’d know what your ultimate purpose is.

However, I think it’s totally untrue that science just looks at the details. Science looks at the big picture of every kind, like science has discovered what is life. One day, science will discover what is consciousness. And people who think that consciousness is, that you understand consciousness when you get into a certain state of mind that makes you happy, they are the ones that are focusing on details and not understanding the big picture, not understanding the context.

Someone who has understood this reminds me of that video that Feynman made about his art friend who tells him he’s missing what’s important about a flower. And he basically says, “No, I can appreciate the flower as much as this guy, but he can’t appreciate what I can appreciate.” And that’s a kind of false stereotype that science only looks at details or science only looks at the mechanical or science only looks at the meaningless things and never gets around to looking at the meaning of things. What they’re really pointing to is that science uncovers problems as when it discovers something new. And just in the big picture, we know a lot more about who and what we are and why, than we did a hundred years ago, and certainly than we did at the time of the founding of the great religions: Judaism, Christianity, Buddhism, and so on.

They were hampered by the fact that they didn’t even know what the sun is. They were hampered by the fact that they were confusing the world with one planet. And in fact, environmentalists today, I just happened to see yesterday that environmentalists say that they want to get in touch with nature. And by nature, they mean certain regions on the surface of one planet. But nature doesn’t know about those prejudices. Nature exists on all planets. And the important thing about this planet is us, not the grass and the fields. So, yeah, there are many mystical and religious worldviews. Some of them do capture something about the human condition in that they can make people happy. At least, you know, in a limited way, they can make some people happy some of the time. And different religions can do this, and your Oxford friend may or may not think that he has the same knowledge as the people in the Bible Belt of the U.S. who sit around in a circle and sing Kumbaya, but they’re also smiling all the time, and they think that they’ve got it, and he thinks that he’s got it.

And, to some extent, they must have something because they can make people happy. There’s this quote in one of the great chess players of the early 20th century. It goes like this: “Chess, like music, like love, has the power to make men happy.” Okay, he’s got a sliver of truth there. There is an important truth in there. But he hasn’t actually understood happiness, or men, or how to achieve anything in the way of making men happy. He’s just got a sliver of truth, and I don’t think that chess player thought of this as being THE truth, but the Kumbaya people, and maybe your person, think that they’ve got the truth, the whole truth, the final truth about this, and they definitely haven’t.

Naval: It’s funny because on Airchat, Brett and I were having a conversation with some people. There was a critical rationalist meetup, and they created an Airchat group where they wanted to talk about critical rationalism. And I think both Brett and I were very uncomfortable participating in any group with a name.

I suddenly felt like, now there’s the argument of what is the central dogma of this group. Lovely people, wonderful people, need more of them in the world. But the problem is that all free thinking comes from the individual, and the moment you make a group, then the group has to have agreements to stick together, and so group cohesiveness becomes the overriding phenomenon, rather than looking for truth.

David: I couldn’t agree more.

Naval: Well, thank you so much, David. You’ve been incredibly generous with your time.