Science & Technology Q&A for Kids (and others) [Part 111]

Hello, everyone. Welcome to another episode of Science and Technology. Q and A for kids and others. And I see we have a number of questions here that are saved up. Let's see, let's pick one here. Paul asks, are we close to making face recognition a ubiquitous replacement for passwords and electronic systems which require login, negating the need to remember and constantly change passwords? Well, so the. Okay, few different things. One question is sort of, how does face recognition work? And another question would be sort of, how good is it? Another question will be sort of what other methods are there of sort of identifying yourself for the sake of your computer and so on. So in terms of face recognition, there's sort of a question of how, what, what is there that is the same about a face? You know, you have your ordinary face, you have your face making some weird expression, you have your face wearing sunglasses, you have your face wearing a hat, you have your face when you, you know, cut yourself somewhere, et cetera, et cetera, et cetera. You have a lot of different small variations of your face. And the question is, can your computer still tell it's you? Or actually the big thing in recent times, the masked face, so to speak. Now, most face recognition methods failed on masked faces, so to speak, and I think they failed also for humans. Often on masked faces, we couldn't tell who somebody was in that case unless we were picking from a fairly small set of possibilities. So the question is, what can you tell about a face that is independent of sort of the detailed conditions of the face? And maybe also if you don't look at your computer or your phone for five years and you're five years older, particularly if you're young, that means your face will look quite different. And can one tell it's the same person, so to speak, or, you know, if you're, if you're set up that way, you know, you grow a beard, you do do this, all kinds of different things you could do. So what's tended to happen is one common technique is to try and use kind of the three dimensional structure of the face. And the way that typically works is your phone or something tries to sort of visualize your face in 3D. And a common way to do that is, is using this approach called structured light, where essentially you project a grid of infrared light that we can't see with our eyes, but which the computer's camera or the phone's camera can see. You project this grid onto your face. What's the point of doing that? Well, if you have something which would ordinarily be a grid. If on a flat surface, it projects as a grid, when the surface isn't flat, that grid will be deformed and the deformation will show you kind of the shape of the surface. And so that's a technique for figuring out kind of what is the three dimensional shape of the thing that you're projecting the grid onto, which in this particular case might be a face. So that's a way to get a slightly more robust sense of kind of what, what is that face that's in front of my phone, so to speak? Well, it has this three dimensional shape and that's. So that's a kind of an approach. Now the next question is, well, you know, between the person who just put on sunglasses and the person who, you know, decorated their face in this way or that way, how much change can you make to one person's face and how far away does that leave you from another person's face? Or are you going to get confused about whose face you're actually looking at? This is, I don't know all the details about sort of how far you can go and so on. I think what is often the case is that for things like passwords, you are really just trying to say, is it plausible that this person is the person who they say they are, so to speak. You know, the username and you're trying to verify is this in fact, plausibly the person whose username is that? And even if it could be the case that out of a thousand people, one of them would turn out to be one who would successfully trigger that face recognition, that's still okay because, you know, 999 times out of a thousand, at least the, the thing will correctly verify that, yes, it is the person who it said it was. And if that number is a million or something, it gets good enough that it's probably considered okay. It's a different problem from the problem of the. The thing that I think is, is slightly problematic in the world of, you know, you just have some random camera on a street somewhere and it says, who's that person who just walked by? That's a different problem than the question of verifying the person who just typed in their username. Is it plausibly that person who is actually using this phone? The problem of being able to say, okay, out of a billion people who might walk along that street, which particular one is this? That's a more difficult problem than verifying that, oh, we think it's this person because that's the username they Typed in. Now let's see if we can verify that. I think people, different times have told me different things about what aspects of faces are the most important for modern face recognition systems. Sometimes people say it's things about, you know, the ears and the height of the ears and so on, because that something doesn't change when you put on sunglasses or whatever else. I'm not sure what the current thing that's used is, although I think the three dimensional structure is an important piece in modern times. I think this. So in terms of, well, what are the different ways you can tell it's you, so to speak? There are a whole variety. I mean, you know, there's fingerprints, there's things to do with the iris of one's eye. There's lots of, lots of detailed things and you know, even between. Oh, and there's things like if you're prepared to talk to something, there's details of the, of your voice. Most of these things can be changed by sort of stuff that happens every day to people, some less than others. I mean, you know, if it's a, if it's the iris of your eye, you might put a contact lens on that has some different pattern on it. If it's your fingerprint, you might have, you know, smooshed your finger somehow and that will change your fingerprint. There are all sorts of different things that, that can change kind of the, the, these aspects of the sort of the, the, the you that the computer recognizes, so to speak. I mean, ultimately probably your, your DNA is a pretty good indicator of the who you are. And it isn't yet the case. It's still the case that to sequence a piece of DNA or even to get what's called a DNA fingerprint, to pick out certain features of a piece of DNA that might be characteristic of a person or a species or whatever else, all of those things are not kind of the kind of thing that could conveniently happen when you put your finger on your phone or something like this. But yeah, I think that that's. No, I mean this question about passwords and sort of writing down passwords and so on. I mean it, it used to be the case that people would say, oh, you should change your password for everything once a month. The problem with that is unless you have some system for writing data, for recording those passwords, which itself has certain vulnerabilities, people, people have the terrible habit of saying, oh, I have to write it and change my password every month. Let me write my password on a piece of paper so that I will remember it. Otherwise, you know, you change it every month and you keep on not remembering what it is. Or you come up with a scheme where you say, well, if it's October, I'll put a 10 in my password. And by the time you've got schemes like that, it's much easier to guess what's going on. So a few thoughts on that. Let's see. Well, the questions here about quantum mechanics. Maybe I can tackle one of those. That's hard work, guys. All right, let's see if we can do some of these. I was asking about correlations, as Aaron asking about correlations among qubits and how that differs from ordinary bits and so on. All right, well, let's see. So let's talk a bit about quantum mechanics. Quantum mechanics is this theory that's supposed to describe ultimately, small things in the world, like atoms and so on and electrons and such like. And sort of the big difference from big things in the world is that when it comes to big things in the world, we generally have the view that definite things happen to big things in the world. Like when you throw a ball, it moves in a definite trajectory, moves in a definite path. But when it comes to small things like electrons, what quantum mechanics says is that it isn't the case that definite things happen. Instead, there's a whole set of possibilities for what can happen, and. And somehow what we perceive is some sort of probability that of what happened, that's based on all of those different possibilities for what could have happened. Now, in our physics project, this becomes kind of more explicit and clearer. What's going on in our physics project, the way that everything gets represented in the world is as this giant network of kind of atoms of space. So we imagine that space, physical space, is ultimately made of things. It's made of atoms of space in the same kind of way that water might be made of molecules of water. And all those atoms of space have this feature that they're related to each other. And we can represent that by having this kind of network of relations that represents the structure of space, and not only the structure of space, but also the structure of kind of everything in space. And then we imagine that time is the process of sort of the continual rewriting of this network that represents space and everything in space. And so that's kind of the nature of time very different from the nature of space. In the model that we have, which seems really to be a very convincing model, the time is this kind of computational process of continually rewriting these relations, these connections between the atoms of space. Okay, so here's where things get tricky. When we say time consists of the continual rewriting of these relations between atoms of space. It turns out there isn't just a unique way to do that. At any given moment, There are many possible ways that you could rewrite this set of relations between atoms of space. Each one of those possible rewritings essentially gives you a different step in history. It gives you a different possible thing that could have happened. And when you put all those together, you end up with these many different paths of history that can occur. And so, as I said, sort of the core idea of quantum mechanics is that there are many possible paths of history. One of the things that comes out in our model that was not, I think, very clear before is that these different paths of history, which in our model are just the different possible sequences of rewritings that can exist of this network that represents the structure of space and everything in it. The different possible rewritings can both branch. That is, one state of the universe can end up having two possible outcomes, but they can also merge in the sense that two different states of the universe can end up evolving to something which is identical. Now, I should say one thing that makes all of this possible to talk about and realistic to think about is in our model, space is ultimately discrete. And everything about what I'm saying is ultimately consists of. I can just count. I can say there's this number of atoms of space. There are this set of possible paths. I can enumerate those things. You know, in the history of people's studies of kind of matter and things in the universe and so on, people didn't know. 200 years ago, nobody had any idea that things like water were made of discrete molecules. Some people thought that might be the case. Other people violently disagreed and said, no, no, no, it's a continuous thing that is cannot be broken down into discrete atomic pieces. Well, it nevertheless turned out that in terms of making a theory of what was going on, like things like the theory of heat that was constructed. In the end, the correct theory of heat was constructed on the basis of the assumption that things like water were made of discrete molecules that could move around. And then people started constructing the idea of, let's look at enumerating all possible configurations of this sort of glass of water or something, all the possible ways, places the atoms could be, all the possible configurations the atoms could be in. And people, this is in the late 1800s. People kind of very routinely kind of assumed that things would be set up in this kind of discrete way. And then when quantum mechanics was discovered, that was originally discovered in the context of photons, particles of light, that light was actually made of sort of discrete lumps of stuff. Then it became clear that there were atoms, that matter was made of discrete stuff. It was still the case that people imagined that space, for example, and time were not discrete in that way. That's something that's new in our model, is that space and time, like atoms, like photons, like other things, are in fact discrete. Well, as soon as you say that everything, space, time and so on, it's all discrete, it becomes a lot easier to kind of talk about things like all these different possible branches of history and so on. If you're trying to do that using continuous mathematics, using mathematics that's based on calculus and so on, it's really very hard. It's very hard to define what you're talking about because it's kind of like you've got these many paths, but they're not distinct paths that you can count. There's some kind of complicated mathematical structure, structure that's very hard to define. So one has a huge handicap if one's saying, let's think about these things in terms of traditional kind of calculus based mathematics. What's made possible our physics project is kind of this idea of discreteness, which is an idea that's very fundamental to computation. We talk about having discrete bits of data in our computer and things like that. These are all like, it's a 1 or it's a 0. It's not a 0.725 value, so to speak. It's a definite, it's a yes or it's a no, so to speak. So that idea of discreteness is important in setting things up. Okay, so now what? Well, now there's the question. Okay, so we've got all these possible histories for the universe. There's sort of a big experiential problem with that, which is that we, in our normal existence imagine that definite things happen in the universe. We don't perceive the universe to be something where there are all these different possible paths of history going on. We perceive there to be definite things happening in the universe. So the question is, how does that happen? Well, the, the answer is in, in large part, the answer is that we are kind of averaging over large chunks of sort of the space of histories of the universe. When we are looking at something like, I don't know, a gas or liquid or something like that, we put our finger in water and we're running it along and we're seeing what happens. There are lots of individual molecules, trillions and trillions and trillions of individual molecules that are hitting our finger and making the water move around with certain and certain speeds and things like this. But we don't notice that. We just notice the aggregate effect of all of those molecules that are hitting our finger and making the water behave in the way that the water behaves. Well, that's what happens in the case of sort of averaging over different molecules in water. We're also averaging over different atoms of space. Whenever we perceive space to be kind of continuous, we can just move from here to there. We don't go in jumps from here to there. That's because we're pretty big, very big compared to the atoms of space. So for us, what would be kind of discrete jumps really we perceive as continuous. Kind of like we know that a computer screen is made of pixels, yet we can watch a movie that seems to have kind of continuous things happening in the movie, even though in the end we know it's made from a bunch of discrete pixels. And so it's the same thing with our perception of space and motion in space and so on. Now it's a little bit more abstract to talk about the fact that we are sort of averaging over a bunch of different histories in the universe. But the reason that we end up sort of inevitably doing that is that just as the universe is branching and merging and so on with all these different possible threads of history, so too are our brains. And so in a sense, what's happening is our brain, which is doing all this branching and merging and so on, that branching and merging is causing. If we thought our brain was sort of at a definite place in what we call branchial space, the space of possible branches of history. If we thought our brain was kind of at a definite place in branchial space, well, really it's not, because progressively over time it's kind of moving around in branch hill space. And what we perceive as being sort of our. Our perception of the universe, our perception of our internal kind of consciousness, so to speak, is something which is quite extended in branch hill space. There are many different possible branches of history that we aggregate together to make what we consider to be our sort of thread of experience. And so that means that what where we then can say, oh, there are all these different threads that do slightly different things in detail, but we're aggregating them all together. Just like all those individual molecules in the water, we're doing different things in detail, but we aggregate them all together and come up with some definite conclusion. And that's kind of our typical perception of the universe. Now, if we look at exactly the right kind of experiment, we can start seeing the fact that there isn't. It isn't the case that there really is just one branch of history, one thread of history. We can start seeing the fact that there are these separate threads of history. And it's just like when it comes to perceiving, I don't know, the fact that water is made of definite. Of discrete molecules. That became possible when people had microscopes, when people realized you could look at pollen grains being pushed around with Brownian motion or other kinds of effects like that. It's something where with the right instrumentation, we could kind of get a sense that, in fact, for example, water wasn't a continuous kind of thing. So similarly with Quantum mechanics, around 1900, between different studies of kind of light from atoms and things that involved essentially electronic amplification, we were able to get to the point where we could actually sort of see those individual threads of history, at least in some very special cases, and notice that things weren't. There wasn't just one kind of thread of experience. And this is kind of the. The basis for quantum mechanics is not the way it's usually explained, but that's the way that we can now understand it from our physics project. The. What happened is the mathematics of describing these different, essentially branches of history, these different quantum states that one can get that mathematics developed in the 1920s, mostly. And essentially what happened is that one started to be able to say, well, we've got a thing. Let's say it's an electron, and the electron can either be spinning this way or that way. And there are two possibilities. And quantum mechanics says there's only two possibilities in this setup. And. But we can end up with something where we don't. Where our kind of perception of the universe is such that both those possibilities are there. Both those possibilities can happen in some sense. Now, at some point, if we kind of want to get to the point where we have a definite thread of experience, we're going to have to make a decision was it actually the up direction or was it the down direction? We're going to have to make a quantum measurement that decides that. And it depends on. But at the level of individual electrons, individual electron spins, and so on, we can maintain kind of this kind of. We can maintain those both possibilities for quite a while, and we don't have to. Our kind of common experience doesn't Sort of force us to make a decision between these. In other cases where it's some great big effect in the world. The rather gruesome one that's commonly discussed as a cat, is it alive or dead type thing. And you don't sort of imagine that. You imagine that you can definitely tell it's alive or it's dead. And you don't have something which is a superposition of states of the cat being alive and dead. But that's less obvious for something like an individual electrode on. Okay, so one of the questions is, can you make use of the fact that there are all these different threads of history, that there isn't just one thing that happens, that there are many things that sort of internally happen in quantum mechanics. Can you make use of those multiple threads of history to, for example, make a more efficient computer, a quantum computer? Can you? For example, there are many problems that might be faced by a classical computer. I don't know, typical ones might be things like, oh, you're trying to find a way to go through a graph. You're trying to find a path, the shortest path through a graph. Okay, you can try all these different possible paths and you know, can you? And then you might compare them and say, what's the shortest one? Or you could take some thing like factoring a number where you say, if we could guess this set of integers that are the factors of our number, when we multiply them together, we can easily check it's our number. But to determine from our number how to break it down into those factors might be really hard. There are a whole bunch of problems where one can, for example, well, where sort of trying out multiple paths is important in some problems. Even the case that once you get a path that is a winning path, it's easy to check that it's a winning path. Those are the so called NP non deterministic polynomial time problems, typically of which, well, factoring is an example of an NP problem. It's not known whether it's so called NP complete, but it's an NP problem for sure. So anyway, what one might think is with this quantum computer, you've got all these different paths of history. Surely one can use one path of history to try each of the different possible factorings of the number. And that way we can in parallel run all these different computations. And that's going to be a better deal than just having to use an ordinary computer where we have to run one thing, then the next thing, then the next thing. Obviously modern computers have, for example, multiple Cores where one can do a certain amount of parallelization just by virtue of sending one instance of the computation to one computer core, another to another computer core. But kind of the idea for quantum mechanics would be use the intrinsic parallelism of the universe to be able to get all these computations done in a faster way. Okay, well, that's well and good. The real problem with this is how do you know what the answer is at the end and how do you figure out in the end? You've got all these different threads of history, and they may all have different things going on on them, but we as humans want to have a definite, this is the answer type result. And so that kind of knitting together of all these different threads of history to come out with the one definite answer, that's not a trivial process. And that's a process which has been pretty hard to describe in kind of the traditional formalism of quantum mechanics. It's a little bit easier describe in our models of physics. And it's not obvious that the effort to kind of knit together the threads of history is any less great than the effort to just follow all the different threads to begin with. In other words, do you actually get something fundamentally different, fundamentally more efficient by saying, oh, we're going to break this quantum computation up into all these different possibilities in sort of the insides of quantum mechanics. And then, but then there's a question of, do we, how do we knit them together to get a definite conclusion of the kind that we humans want? Now, if we humans were operating ourselves right down at the quantum scale, if our brains had features where we could in our brains hold separately those different possibilities in the superposition, it will be a different story. But what we typically imagine when we talk about quantum computing is quantum computing for the human user, so to speak, not quantum computing for the, for the quantum alien, so to speak, who's operating purely in a quantum way, and who can hold in their brain these two different superpositions of what's going on? Okay, so I think the question had to do with, with correlations between different bits and so on. And I think the thing to understand is that in, okay, when the different threads of history in the universe are running and doing their thing, these threads of history are operating independently or they might be merging, they might be branching. It's a thing going on in there that has sort of its own self consistent operation. As soon as you say, oh, I'm going to force those threads of history to get knitted together so that I can get a definite answer, you've broken that internal mechanism. It's kind of like the computer that is you'll void the warranty if you open this computer. As soon as you kind of go in and humanize all those sort of individual threads of history that are happening at the purely quantum level, you've broken that quantum mechanism. And so the question of sort of what happens at the level of the quantum mechanism is a separate question from what happens and what we perceive from that quantum mechanism. So, for example, okay, one of the issues is if you have two bits, two qubits, which are these kind of things that you would represent. So in a standard bit in a computer, when you store one bit of information in a computer, how is it actually stored at different times? There have been different technologies for doing it. These days, it's actually stored by having a certain amount of charge, a certain number of electrons, maybe 10,000, 100,000 electrons per bit these days that are either present in this kind of configuration of transistors or not. And that's what determines is it a one or is it a zero? Back in the day, when I first used computers, there was a thing called core memory where you would have a little tiny piece of a little circular magnet. And the one bit was recorded in having the magnetization of the circular magnet going one orientation. And the zero bit was the orientation of the magnetization being the opposite way. And that's pretty analogous. Those magnets were big things. You could see them with your eyes. I mean, they were. You know, they looked. They were like. They were sort of somewhat, you know, took effort to see, but you could definitely see them with your eyes in a sense, in a. In a sort of quantum computer. You're imagining those magnets have shrunk down to the point where they're just single electrons each. And the question of whether the magnetic field is one way or the other is the question of whether the electron is spinning in one orientation or in the other orientation. And so that's kind of the idea of a qubit is something like that microscopic magnetic core memory just made from single electrons each. Now, one of the questions then is what are the. When you. When you have these different things happening and you have different configurations of electron spins, for example, that represents different things you're storing. One of the issues will be if you've got two electrons and they start off having some correlation, like one of them is up, one of them's down, and you know that they're always paired together, one up as one and one down. And then you say, well, Somehow we're going to separate those two electrons in space. We're going to move them apart. Well, normally when you move things apart, they're no longer particularly correlated with. They don't have any reason to be correlated with each other. You know, one electron can have one thing happen to it, the other electron can have something different happen to it. Whatever correlation might have existed originally between those two electron spins will be destroyed, one would think, by separating those two electrons apart from. And in general, if you wanted those electrons to cooperate again, the way they'd have to do that is by like sending a signal from one electron to the other. And so how do you send a signal? Well, you might send a radio signal, you might send a light signal, you might send a neutrino signal. Whatever kind of signal you send. The maximum speed at which that signal can go in physical space is the speed of light. So if these electrons are far separated, that the, the, the, the, the to make sure that they're correlated will take a time. That is depends on how far apart they are and the speed of light. Okay, but when you're dealing with quantum mechanics, you can end up saying, well, in a particular thread of history, we know the electron has this configuration, even though we know that we've got this up, down configuration. Even though that's what happens in one thread of history and another thread of history, something different will happen. And we can set it up so that even though the electrons are far separated in space, in that particular thread of history, we know that there wasn't this flipping around one of the electrons, because that thread of history is a definite threat of history, a definite thing happens, another thread of history, something different will happen. But so what one can set up here is something where if you could separate out one thread of history, one would know that there was this correlation between these two electrons, even though the electrons may be in space very far apart. But because it's a single pure thread of history, we know that the electrons have this form. And because nothing have kind of, because it's a single thread of history, nothing kind of, kind of got in there to change it. Because that will be a different thread of history if it had done that. So that's why in quantum mechanics it's called a pure state where one's got something, where it's a sort of single thread of history that one's dealing with. And that single thread of history can make it look like there's these correlations between electrons in one place in a completely different place. And that's Kind of a weird effect. Because if one could really home in. On a single thread of history. One would know this correlation existed. Even though those electrons might be very far apart. And so there are actual experiments that have been done. Which show that, yes, you can home in on essentially a single thread of history. At least in terms of the electrons. Other things in the universe might have correspond to different threads of history. But at least as far as the electrons are concerned. You can have this. The single thread of history. And so that's how you can get this kind of. This surprising correlation. Between these electrons. Because that correlation exists because you're dealing with a single thread of history. As soon as you knit together the threads of history to make an actual measurement. You kind of lose this effect. Of being able to have this sort of separated thing. Go on. So that's kind of why you can get correlations between qubits. That so long as you are maintaining the sort of pure, pure single thread of history. You can maintain those correlations in a way that is very different. From the way that you can make things be correlated. When you're dealing with sort of, we know what happened. Definite kind of behavior that you get. When you perceive the universe, as we humans do. Where we're averaging across a large number of threads of history. This is complicated stuff. The mathematical formalism that describes how this works. Apart from the measurement part, which is a mess. The other stuff is clean formalism that we've known for 100 years. Although it is clearer what's going on. When you think about things in terms of our physics project. In terms of these very discrete set of possibilities. And all that kind of thing. But that's. That's a little bit of an idea of how that works. Let's see. Okay, so Spare is saying perhaps a conscious observer. Is in fact the result of the underlying physical system. Building a model that averages out all the parallel threads into a coherent story. Yes, there's a lot to that. I mean, I think one of the critical things about consciousness. Is this idea that we have a definite thread of experience. That is persistent through time. And without that, we really have something very different. From our human experience of the world. Now, as far as the world itself is concerned. There are many different threads of history. There's not only are these threads of history. That are associated with quantum mechanics. There's also this whole even greater collection of threads of history. Associated with a rulial space. Associated with the set of all the possible rules. That can be used to update the universe. Which we think are Sort of all operating at the same time to make this whole kind of ruliad construct of which sampling just a tiny piece. And so the fact that we have a definite view about what's happening in the universe is a consequence of the fact that we are sort of aggregating. Okay, so, so there are several different effects. We're taking a small slice of all the possibilities. Just because we're, we're a certain size in the universe. We don't span the whole universe. We don't know what's happening everywhere in the universe. We don't span all of rural space. We don't span all the branchial space. We are localized in those different kinds of space. So we only have a kind of a parochial local impression of what's going on. That's one thing. The other thing is that's what happens because we're small compared to those kinds of the whole universe in those kinds of spaces. But we're also quite big compared to the individual atoms of existence, so to speak, or the atoms of space. And so we're averaging many of those. And that's one of the reasons that we can get a robust notion of, yes, this is the definite experience we had happen. At the level of all of this microscopic stuff that's going on. There's a lot of complicated, random looking stuff that's happening, but we get to aggregate lots of that because we're pretty big. And that's why we imagine there's a single thread of experience. What's not obvious is that that aggregation will be consistent. It could be the case that we could take these two threads of history and we say we're just going to assume the same thing happens on these two threads. And it could turn out that that assumption is just vastly wrong and that we get completely incompatible. We can't make up laws of physics because there are all these weird things that keep on happening and that are incompatible with any consistent laws we make up. The fact that doesn't happen is a consequence in our model of this phenomenon called causal invariance. Which basically is a thing that says even though there may be two different branches of history, those branches will always in the end merge. And that's a necessary feature. By the time you have branches that correspond to all possible rules by which things can be updated, it's inevitably the case that any branch will eventually be able to merge. It's less obvious that that would be true when you specialize to a particular set of rules. When you do something more like the pure quantum mechanics and so on. But it seems to be true there. We think it's true there too. So that's kind of a thought about that. But I think that the important thing about consciousness is it's big, but it's small. It's big in the sense that it is able to average over. And sort of take this consistent slice. Of all of these underlying sort of random details. Of what's happening underneath. It's big at that level, but it's small compared to the universe of all possibilities. If it wasn't small compared to the universal possibilities. It would be hard to have a coherent notion of us. In other words, if we, for example, were very extended. And even in physical space, it's like, what's our experience? Well, you know, we can have an experience based on where we are. Or we're seeing these things, we're seeing those things. But if we experience the whole universe or all at once. It's much harder to say. We have a definite story about what happened to us. It's like all these different things are happening. It's even worse when we look in rural space. At all the possible different description frameworks for the universe. Because that's a place where we end up with sort of this very incoherent view. Of all the possible things that are happening. We don't have this coherent sense that this is how the world works. But these are all these possibilities for how the world can work. And we are kind of spanning all of those. And so in some sense, if we're. If we're too big. Then we don't really have a coherent sense of self. We don't really have a coherent sense of what we are. And how we kind of fit into the universe. If we're too small, we're too subject to all of this sort of random stuff that we can't really track. So we're kind of this intermediate size in the universe. That makes it possible for us to have kind of the kind of consciousness experience that we typically have. If we were much bigger. We could no doubt have a. A perfectly meaningful experience of the universe. But it will be different. It will be something where, you know, if different parts of our brains. If it mattered what the speed of light was, but. And, you know, communicating thoughts between different parts of our brains. That would be. We would have a very different experience. And very different kind of way of thinking about things than we do now. And it's pretty hard to imagine, at least for me. Kind of that level of difference. In our kind of way of perceiving the universe. I mean, we're kind of stuck in the way of perceiving the universe that we've sort of grown up with and that our senses make obvious to use. To try and imagine a different way of perceiving the universe is very interesting, but it's also very hard. Even within your average whale or dolphin or your average whatever. What way of perceiving the universe does one have there? It's hard to know, but it's interesting to try to imagine. What other kinds of ways of perceiving the universe might there be? And what kinds of very different physics might you construct if you had different ways of perceiving the universe? See, William is asking about destructive interference between threads of history. Okay. All right, let's talk a little bit about that. So what is destructive interference? What happens is when you have waves, ripples on surface of water, for example, you'll have a wave, and it'll be going and rippling. And it'll be going up and down and so on. Well, if you have two sources of those waves, you know, you have your. Somebody's like sort of bouncing something on the water in one place and another place, you'll get these two kind of circular patterns of ripples going out. They'll each correspond to waves. So what happens is if waves. The two waves from those two sources arrive at a particular place and they're in phase, in the sense that both waves are up at the same time, then they'll add together. If they're both down at the same time, they'll also add together. But if they're out of phase, if one wave is up whenever the other wave is down, they will cancel each other out. It's like you're saying one sort of thing is telling the water, go up by a millimeter. The other one's saying, go down by millimeter. Those two things in the case of water just sort of add together. And the thing destructively interferes. You've got waves from those two collections of waves. And you might say, well, they'll always add up and you'll get twice the height. But actually, if they're out of phase, they can interfere and you'll get zero height instead. So one of the things that was observed ultimately, well, in the. Was observed for light, oh, in the 1600s or so. And it was observed for atoms in the 1920s or 1930s. That it's possible to have destructive interference between things that you thought were kind of material objects where you. Where it wasn't just the water is going up and down. It was like, there's a photon there or there's an electron there, and it's. It's like one plus one equals zero, so to speak. So a typical experiment, you have these two slits and you have something where you have a source of light. And the light can either go through one slit or through the other slit. And what happens is that you don't actually get to see the sort of the waves that correspond to the possibility of the thing going through one slit or the other. But the end result is that you get this interference pattern which you can actually see on the. You know, the photons can land at this place on a screen at the back, behind the slits or in that place. But in between those places, there will be destructive interference and no photons will show up there. How does that work in our kind of models? Well, what you might think is if there are different threads of history, these different threads of history must always just add up that if there's one way that the photons can go and there's another way the photons can go, then they must just. Those must just be combined. Well, here's the problem. The problem is those different threads of history. For those to be experienced by us, we have to have knitted together those threads of history. We have to have somehow taken the. It went through the left slit, it went through the right slit and combined those and say, well, it went through, you know, the combination of those as it. Well, it went through one of the slits. And we can sort of say there's a certain rate of 1, certain rate of the other. We could combine those together. That's all well and good. If we somehow can successfully knit together those different threads of history. In other words, let's say we've got a certain size in branchial space. We've got a certain set of different threads of history that we can say, oh, yes, we're aggregating across these threads of history. These are all things that are consistent with being what we experience. Just like when we're looking at, I don't know, the finger in water, we're aggregating over all the molecules that are spatially near us in the water. So similarly, in branchial space, we're aggregating over all paths of history that are somehow near a particular place, maybe some particular path of history or whatever that are near somewhere in branch hill space. Okay, so that's well and good. But now the question is, what happens if. If this. If the paths in branchial space wind up in very different Places in branch hill space. If the it went through the left slit and it went through the right slit, if those two different threads of history end up far, far away in branchial space, then we as observers can't knit those together. They're too far apart. We are only of a certain size and we don't aggregate over that region. And so that's, I think, the qualitative picture of what's happening in destructive interference. Basically what's happening is the different possibilities, the different threads of history wind up being far away in branch hill space. And our effort to aggregate threads of history isn't able to span those two things, isn't able to bring them together. So we just say, where'd the electron go? Where did the photon go? We don't know where it went because neither of the threads of history where it went through either. Either of these slits is something within sort of the domain that we're aggregating to extract our thread of experience. I think that's the qualitative picture of how sort of one plus one can equal zero in a case like that. That by having these two different things that wind up sort of in different, different corners of branchial space, they're far enough apart, we never get to aggregate them. And so we can't say that either of them ever happened. At least that's a qualitative picture. We're trying to fill in more of the mathematics of how that works. And it's kind of tricky. We sort of thought we had an idea how to do that a couple of years ago, but the mathematics of filling in the mathematics is a complicated story there. And it really relates to kind of the correspondence between the way things work in branchial space, the way things work in physical space, the kind of very weird sort of way in which this sort of space of possible branchings can be kind of put in correspondence with physical space. It's kind of like if you have a tree which has sort of an infinitely branching collection of possibilities, how do you take that tree with its infinitely branching possibilities and arrange them in ordinary space? In that particular example, when it's a pure tree, you can use hyperbolic space and so on to make various kinds of arrangements. But the situation we have here is more complicated than that. And that's kind of the type of thing that's involved. Well, there's a question here. Could our brains be a quantum computer? Well, not if you believe in. Not in an interesting way, in the following sense. If you believe that you have a single thread of experience, you're not using quantum mechanics in a serious way. Because if our brains were sort of truly in the innards of being a quantum computer, they would be able to kind of. They would. We would not have our description of the world that says, this happens, then this happens, then this happens. Definite things happen in the world. We would have a very different sort of description framework for the world. Now, can we imagine that one day there will be a device that operates purely at that level that never sort of externalizes to be relevant to the human observer and so on? Yeah, sure. You can imagine such a device happens all the time in kind of the innards of quantum mechanics. But the externalization of that depends on, well, what experience do we as humans have based on what's going on? And if we had brains that were fundamentally down in sort of the quantum domain, then we would be able to fundamentally understand these quantum processes. But our brains would be very different from the way our brains actually are today. And in particular, this thing that has been crucial to our experience of the world, this is this idea of single thread of experience wouldn't be there anymore. So, you know, you imagine the future evolved humans that have true sort of quantum brains. And you are communicating. You're talking to one of those future quantum brain humans. It will be very confusing because to that future human, so to speak, if you, you know, classify them as a human, they would be kind of imagining all these different threads of history that were going on. And, and you'd be sort of saying, but what actually happens? What's actually going on? What actually happened? And they're like, I don't know what you're talking about. You know, it's. It's kind of like all things are happening. All these different threads of history are all happening at the same time. I don't know what. And, you know, it's a different level of experience. I mean, it's sort of. I'm sure there's an analogy. If you look at modern sort of modern technology and you view it from some time in the past, and you say the modern technology has sort of expanded our experience of the world. And you might say, well, how would you describe that? How would you feed that back to a. A, you know, a consciousness from 2000 years ago or something? How would you feed back the things that we now are able to experience? How would you feed it to somebody who. To a mind, a brain that doesn't have those, that doesn't have the ex. The experience of having those kinds of experiences, so to speak. And I think That's a challenging thing. So I think in the, you know, could we enable sort of the quantum bit in our brains? Well, if we do that, if that was possible, and I think the scales are somewhat wrong for that to really be possible, but if that was possible, and I think probably biology goes to some trouble to make our brains actually believe in a definite sort of thread of experience, because that's what's necessary to operate in the ways that biological organisms tend to operate in the world. And so, you know, there's, there's quite a bit of effort the brain goes to, to kind of aggregate our sensory experiences to this one kind of layer of this one kind of attention string that is, that is going through time, so to speak. And that's. So I think whatever the hardware is doing, it's still the case that being human is a lot about sort of aggregating things together into that single thread of experience. And if we say, well, we don't really care about being human, well, well and good, fine, but you may not have something which to us today, in, you know, 21st century humans could, can actually relate to, so to speak. Right. There's a comment here, I need to go in a moment here, but, but there's a comment here about. Neurons firing in the brain and, and they're kind of, what kind of computation is going on there. I mean, one of the, the sort of, I would say it's almost the fundamental problem of neuroscience is to know what matters in the brain. That is to have a description of the brain that is below the level of just saying, oh, thinking is happening and we can do psychology about it. And above the level of, and this is how an individual neuron works and this is how the electrical signals go from here to there and so on. What is the intermediate level of description that is between. So for example, for a computer, we have an intermediate level of description. It's the typical kind of machine code and then the layers of software that we build up above that machine code. But in talking about a computer, we're not down at the level of saying where'd that individual electron go? Which there's a level of description that we sort of understand is there, that is above the level of the individual transistors switching in the computer and all those kinds of things. And so. That's kind of the, but the question is what in the case of the brain, what is the kind of low level language of the brain that is above individual neuron firings and below just this whole big thought happened? We don't know. And we get some better clues maybe from artificial neural nets that seem to have some characteristics in common with the biological ones we have. But even with artificial neural nets, we're kind of at a bit of a loss when we build up some complicated machine learning system to give a kind of science level narrative description about what's happening in there. So, Yeah, so anyway, a few thoughts about that. I see a bunch of questions here which are all very interesting, but I think that I have to go and do some day job kinds of things. And you guys are tempting me with a few other questions here that I just. Aaron is asking what Wolfram language functions would be most improved if they could use 20 million logical qubits on a quantum computer? It's kind of an interesting question if, if there was even, even just let's say we had. I mean, okay, the computer that I used when I first started using computer 50 years ago now had a cycle time of around 2 microseconds. Computers that I use today have cycle times that are about 5,000 times faster than that. So what is different now that the computer is 5,000 times faster than it was back then? What would be different again if the computer I used was 100,000 times faster than the computers today? A million times faster than computers today. And it's not inconceivable that quantum mechanics as such, forget these parallel threads, which I think are going to be hard to really make use of in a computer, just reduce the size of the components, just be dealing with individual electrons rather than 100,000 electrons, doing things with individual molecules rather than whole aggregations of molecules. It's possible to imagine much faster computation, forgetting about whether you're doing this multi threaded kind of multi quantum type thing. And you could ask the question, well, what would be enabled by having a computer that's 1,000 or 100,000 times faster than the computers we have today. I think it's interesting because a lot of that question revolves around. So in the end, it's a human user and the human user has, you know, our eyes are sensitive only every 30th of a second or something to what the image looks like. If we started switching out images a million times a second, we wouldn't notice. We'd only notice what comes up every 30th of a second or so. Similarly, we know our ears are only sensitive to maybe 40,000, if you're lucky, kind of changes of sound pressure per second, things like this. So there's a question of, well, the stuff that the computer is actually showing to us. There will come a point not too far from now when our sensory inputs are actually the bottleneck in terms of getting stuff from our computer to our brains. Right now we can make sounds. We can't completely synthesize sounds fast enough on computers. We have today to make the synthesized version of some spoken voice that is transformed in some way or some way. We can't make something which is a completely realistic every pixel is right image that we could see through our some display that's displayed to our eyes. We can't quite do that yet we will be able to do those things. Those things are very much on track to get done, you know. Exactly how many years it will be is hard to predict, but a modest number. I think that we will get to the point where we can routinely synthesize much of what is sort of our at least static, everyday kinds of phenomena in the world. And to go to make the computer faster at doing that, it's like irrelevant because the bottleneck is us and our ability to absorb that data. Now, on the back end, there are all kinds of computations where to work out that thing that you can feed to the human that could take a lot of effort, and it's an interesting question that I have not thought that much about is what could we really achieve if we had computers that were 100,000 times faster? What kinds of things that. I mean, I will tell you, in the development of Wolfram language, I've always taken the point of view that we should develop the language as we think we should develop the language. We should have a clean, elegant specification of making some particular kind of thing computable. And then the computer hardware and systems will always catch up. So, for example. But there are things that we can have a clean and elegant specification, but it's kind of annoying for a human to use. There are things like, let's say a user interface where it's a clean and elegant specification, but things move around at the rate of one little movement per second. Way too slow for a human to be happy with it. But then computers get faster and there's more and more that you can kind of do in the loop in a clean, elegant way within the time frame that for a human constitutes kind of sort of a continuous flow of time. So that's one of the things we've noticed in the building of Wolfm language that there's more and more, for example, user interface that we can just specify in a nice, clean, elegant way and it runs fast enough that for a human User. It seems to be something that is a smooth kind of experience. And so there are things like that where. There are things where the sort of the clean, elegant specification, it's like, can we go from a clean, elegant specification of something to how it actually works? Can we do that in real time? Are our computers fast enough that we can, I don't know, do something where we have defined some axiomatic framework and. We then simply say the computer is fast enough that in real time, every time we ask for something, it's computing all these different paths and so on and telling us the result. I think those are the kinds of things we can imagine that it's possible to go from the high level specification to what we experience. It's possible to go from a higher and higher level specification to the things we experience. So, for example, we might be able to describe something in some visual scene where we just give it in terms of something where we've specified kind of the almost in the end, the rules of the universe, so to speak. And then we just sort of say to the computer, okay, go, go work out their consequences and show me the result. To do that for the whole universe is clearly not possible. But to do that for larger and larger chunks of the universe, as we have faster and faster kind of pieces of programmable matter, so to speak, represent our computers, gets to be something we can do. All right, I think I have to wrap up for today, so I look forward to continuing another time and addressing some of the questions that I didn't get to today. And thanks for joining me. I was a little tired today, so I might not have been as coherent as sometimes, but look forward to seeing you another time. Thanks for joining me and bye for now.