Business, Innovation, and Managing Life (May 15, 2024)

Hello, everyone. Welcome to another episode of Q and A about business innovation and managing life. And let me see, what interesting questions do we have here? This question from Mikhail. Can you patent an algorithm or a machine learning model? Well, I'm not a lawyer, but, you know, this has been a long discussed issue of what does it take to reach the bar for what can be patented? And I have the impression that, that the bars have been sort of moving around a bit. I mean, there was a time when it was like, no, you can't patent a raw algorithm, but if you describe how it can be implemented in a piece of hardware, then you can patent the thing with the sort of hardware wrapped around it. And then there were a variety of different versions of that. I mean, it's worth having a little bit of a think about. There's questions of what does the law and its precedent say right now? And I know that, for example, in some copyright areas with machine learning and AI and so on, there are even complexities within the US that different parts of the federal court system have different precedents that are inconsistent with each other. So there's a fair amount of turbulence in what's going on there and what should go on there. What should happen to the world's kind of collective knowledge, and how should it be ingestible or not by AIs? How do you t the threads of copyright law that were intended for very different purposes from the ones that they're being applied to now. And is it important that somewhere in the workflow inside the AI training process that there's necessarily copying of data? Is that really a trigger for things? Does it matter whether the machine learning model is going to be freely given away or whether it's going to be proprietary? Does it matter? There are all kinds of questions there. And in terms of algorithms, it's worth having a distinction between, well, things like laws of nature, which have generally been viewed to be not patentable. But on the other hand, for example, specific genes were viewed years ago as being patentable. And then there's sort of questions about, particularly when you get to the level of kind of a structure of things like our physics project, where kind of everything in the universe is ultimately represented as rewriting rules on hypergraphs. It's kind of like the distinction between what's physical and what's merely abstract begins to break down. And then you can ask questions like, well, what would society like to have be the case about what's protectable and what's not? I think in the US Constitution, there is a little Bit of a couple of sentences. I think they were added by Jefferson. I'm not sure about how. Which. Which became the basis for patent law, which were something along the lines of to give sort of people who create things in. I think the term is science and the useful arts the exclusive right to use their inventions for a limited period of time, something like that. And notice that it says, you know, science and the useful arts. But nobody ever figured out how to do any kind of intellectual property protection for, for example, scientific results. That's something where, you know, you. You write about the scientific results in your paper and then it's on your book or whatever, and then it's kind of free for all for people to use that idea. And, you know, so I think it's all pretty complicated. And it's a question of what will be, what's good for society. That's, again, something people absolutely wouldn't agree on. I think that in the different forms of protection, from trade secret to copyright to patent and then all sorts of other ones about trademarks and trade dress and mask works and all kinds of different exotic forms of intellectual property protection, they've all sort of had their place. Whether there will be new categories that come in in the world of AIs, in the world of being able to take kind of things that were sort of part of the kind of the world as it is and then sort of modify them with AI and so on, and where that lands up. I mean, and the question of whether, for example, in the copyright case, there are now some precedents which I don't think are terribly convincing about how things that were sort of had enough AI in them didn't pass the test of being copyrightable by the person who sort of pressed the button on the AI. You know, this is an issue that we thought about quite a bit in connection with our Wolfram Tones project back in 2007. I thought about it a certain amount in connection with things like the pictures and the programs in my book A New Kind of Science. In most of these cases, it's like we just wanted people to use the stuff, but we didn't want crazy things to happen with it. But it was sort of a question of what was protectable by various kinds of intellectual property law. I think one sort of exercise is given a number. What numbers are protectable by copyright law? The number 30 is presumably not protectable by copyright law, but a number that is 25 digits long probably is protectable by copyright law because nobody just comes up with that number as a. Oh, I'll randomly generate that number. A UUID is presumably protectable by copyright law because it is intended to be, not reproduced in some way. Now, that's not a test for copyright law. I mean, the test tends to be. Well, I think there's things about structure and sequence and organization. There's things about originality, there's things about creativity. I think there was a famous Supreme Court case, if I remember correctly. It's the feist case, where somebody had taken a phone book, which is clearly not a creative work in the sense the phone book is the phone book, so to speak. It is merely a representation of some form of external reality. And the question was, is the phone book. To what extent is there copyright that can be asserted on the phone book? Even though it is, in a sense, a purely mechanical thing to take all those subscribers, phone numbers and stick them in a phone book? And I think the way that came down was, yes, that was a copyrightable thing, even though it had no kind of spark of human creativity involved in the creation of the thing. And I think that that is sort of a thing that certainly will relate to whatever one might want to think about in terms of AI and its kind of. At what point is it the creator of the AI? At what point is it the person who presses the button? At what point is the person whose little fragments of their intellectual property were ground up by the AI in creating the AI as it was? Now, there's a somewhat cleaner version of that in a lot of what we do with Wolfram language, for example, where I think there's an argument, which we've never particularly pursued, that when we generate from an algorithm a piece of output, let's say you're doing an integral, there isn't just one answer for that integral. There are many different forms that the answer could be put in. And the choice of what form it is is made by algorithms that we've created. Now, some of those algorithms might be. They're not typically tables that we just say, oh, for this one we want this, for this one we want that. And those tables could have been made by a human just saying I creatively choose to use a sine function instead of a tangent function in this case. I think in that particular example, it's much more a human created the algorithm, which then flows down to whether it's a sine function or a tangent function. But then the question is whether the thing that is output is something which one could consider to be subject to copyright by the creator of the algorithm, by the person who pushed the button to make the result come out. Where is it? Where does it come down? And I think these are questions, as I say, that first of all involve taking laws that were created mostly for other purposes, precedents that are not really quite aligned with this, and sort of teasing them out to this case. Or you can sort of take another point of view of kind of what should be the answer based on kind of where we want things to land in society. And I think those are all interesting questions. And some of them, particularly in the AI case, are very much things that will be sort of, I'm sure, heavily litigated in the years to come. And I have to say that the, you know, I'm sort of the things we create are very much created from scratch. So we don't really have these issues, but they're certainly issues that are of great relevance to lots of companies and the ways that those companies are operating. I think the thing that it's clear that certain kinds of things that we generate, like the complete output from WolframAlpha, is clearly a copyrightable thing. And I think that's very uncontroversial. The question of the formula generated by the integrate function in Wolfram language, I suspect that is a copyrightable thing too, but that has had a somewhat different character in terms of the way that it's used. Let's see. Yeah, it's gosh, hula. Comments. We'll need a new division of law for AI cases. Yeah, for sure. I mean, there are a lot of new issues there, a lot of new issues that are in terms of kind of how one takes the collective knowledge of the species and sort of runs it through the AI, so to speak. In a sense, what we've been doing for years of curating data, you know, we've been doing that in a way where we know where it came from. There is a specific piece. We license it from this place. We use it because we can, based on the copyright status or whatever else. And it's been a very controlled process, so to speak, for the AIs that just say, let's grind up the Web and do this or that. It's been a much less controlled process. I think people are realizing it needs to be a more controlled process, a bit more similar to what we've done over the years and where we can say, yes, we've curated all this data, we've got this stuff, and we know what our rights are in terms of the ability to use it, not use it, provide it to partners, et cetera, et cetera. Et cetera. It's a very well defined thing there, which is something that we've been careful about in setting up the way we've set things up for our business. But I think that there's the increasing situation of the AI did it, so to speak, who's responsible, and all these questions about whether you can foresee from, for example, training data you gave the AI, how you constructed the AI and so on, how you could foresee kind of what would happen. And what do you do when you've got my favorite kind of computational irreducibility phenomenon in the middle, where you say, look, I set up these rules and I couldn't have known what would happen in this circumstance or that circumstance, who's responsible in that case. What do you do? To what extent can you say, well, you should have known it would do this when we can sort of know that it's unknowable whether it will do it or not. And I think that's an interesting issue. I mean, there are places in law where it, it's clear it was an act of God or it was a thing that was where you just couldn't have known. And the law has a definite notion of you couldn't know, but it hasn't been really applied to things like computational irreducibility. I think that one of the things that sort of was a fortunate piece for the software industry back in the day was there was the possibility that sort of every time there was, you know, to what extent were software makers, how were they responsible for bugs and so on. And there were a bunch of early precedents that were precedents that kind of sort of missed the key points because they were precedents. There was a famous one I recall from the, from the 1980s, early 1980s, involving a spreadsheet where somebody had underbid a job based on some results they got from the spreadsheet. And, you know, if that had been a bug in the spreadsheet program, that would have been one thing, but it wasn't. It was a misreading of the documentation by the user, and which is a very different kind of problem than, oh, it was an actual bug in the. It didn't add the numbers up correctly. But anyway, it's, it's a, it's a complicated area. I think this, this whole question of if and when there are autonomous AIs that don't have owners. I mean, generally most things in the world have owners. Although, I don't know. I think, you know, the birds in the, you know, who are Making international journeys all over the place. Those birds don't have an owner. I mean, in some countries, I think in England, the queen owns, owns, or king now owns all swans. But that was probably put in place in the 1200s or something like this, for who knows what reason. I think in the us, is it maybe the case that the federal government owns all the eagles? I'm not sure. But in any case. But there clearly are unowned things in the world, and there's the question of will there be AIs that are unowned things? And under those circumstances, what will be the kind of legal status of the AI and what it does? And will we say, well, the person who created it? Let's say that you have a bird, and let's assume that I'm right. I think I am. That the birds are not owned. Mostly, then you have a bird and you chose to hatch this bird, and it was a very vicious bird of prey, and you put it out there in the world and it's flying around, and suddenly the vicious bird of prey attacks somebody's pet. And then the question is, you know, you say, well, but I don't own that bird. And then there's sort of a question, but you, you know, you created the bird, you hatched the bird, you know, maybe you bred the bird, but then the bird was off free as a bird, so to speak, flying around, who's responsible for what? So I don't know the answers to these things. These are kind of. But that's a sort of analogy, perhaps to the case of the free AI, so to speak. And then, of course, there comes the question of at what point do we feel that we should emancipate the AIs? And at what point do we feel that? I mean, one very basic thing people sometimes bring up is, oh, you've got an AI. Just if the AI does the wrong thing, just kill the AI, erase its bits, and you quickly realize that it isn't quite as simple as that, because the AI could be friends with 10,000 people, and erasing its bits would be of great harm to the people who had befriended it. I think there's also a certain sense in which there is maybe some societal issue of if you've got an AI that stores a bunch of data about the world and stores a bunch of things that are part of the heritage of the world, so to speak. And then you say, well, we don't really want this AI. It's just a bag of bits. We'll just erase them. You are destroying Some of the heritage of the world there, even though that heritage of the world might have been not specifically owned material, so to speak, somehow there's something a bit wrong with that. And it makes one. It reminds one of sort of environmental law, kinds of issues of what are the rights of inanimate objects, rivers, things like this, and how should that be thought about in the law and so on with respect to sort of what effect that has on society. Kyle asks the question, can blockchain mitigate AI facts or does it have its own threat to communications? I think one of the issues is, did you know that such and such a thing really happened? Did it get made up? Is it a deep fake? What is it? And there's been, we've known for quite a few years that yes, you can sort of centrally ledger things to say, yes, it really happened that way. Or at least this thing, this hash of this photograph got put into this ledger at this time. And if you say, well, if you post modified it, then you know that that original timestamp won't be, you know, that you post modified it, so to speak, in terms of. There are now, for example, starting to be cameras that will put digital signatures onto photographs locally to say, based on the hardware of this camera, this thing was taken at this time at this geo position based on GPS or whatever else. And one can imagine a version of that that has sort of a backing with a blockchain where you're putting those. That kind of information, you're putting a hash of the image on the blockchain. So you don't just rely on the fact that somebody says, yes, my camera hardware did this or that thing. You're saying, well, actually there's a central ledger that tells me that this is the way it was, not just I'm relying on not having messed with the hardware of my camera or whatever else. So I think there are, there are definitely things and I wouldn't be surprised if in the this sort of year of lots of political activity that there aren't kind of some pushes to validate certain kinds of things that, you know, I don't think it's going to work to have. I'm not a television watcher, so I don't get to see all of these remarkable political ads and all this kind of thing. But, you know, I don't think it's going to work to have the voiceover that says, you know, this thing was paid for or whatever by so and so because, you know, the faking voices is a thing of this year. And that's not really going to be sort of a secure indication that it really came from that person. I mean, it's kind of similar to the issue of, for example, signatures where, yes, you know, the signature you write comes from your cerebellum or whatever else, and you know, it's. But, you know, that is something where there hasn't been a lot of pressure on that particular thing because there have been, you know, a big move towards digital documents and digital ways of signing things and so on. But there are all these things where it's like what was unique, what was not unique. Then we go to biometrics and things like this and there's a complicated set of issues about when do you want to really commit that it was you, and to what extent is it, you know, what should depend on that? And so on of complicated issues there. Let's see. Oh my gosh. There's a question here from Samuel. Can you rationalize a world where free to use AI services use licenses that allow free use, but then the operating company can capitalize on the content that they generate? Oh, I think I understand this question. I mean, so in other words, imagine that you are a provider of image generation services and you provide those image generation services for free, but people tweak the prompts to make the perfect image. And then you as the provider of those free AI services, say, and now because we are the provider of these AI services, you, even though you, as the user went to lots of trouble to tweak these prompts and chose the picture you thought was great, we as the, as the kind of generators, as the folks operating this free to use service, say we own the results of that. I don't think personally, I mean, my initial reaction is, look, if you say that's what you're doing, do it. I mean, I think there is always a, you know, the free to use service model is, has been a bit of a, you know, often deceptive model. I mean, I think we are sort of some of the only fools in this respect. Because of Wolfram Alpha, for example, it really is free to use. I mean, it has, there are paid things you can, you can add on to it. But WolframAlpha is free to use. And we're not doing anything nefarious or otherwise with the data other than, you know, using it on a very aggregated basis to help improve WolframAlpha. But we're not using it. You know, it's not, we're not, you know, selling data about users or any of those kinds of things. Or running ads or any of those things. The vast majority of free services, it's, you know, it is, it is free to use, but there are all kinds of entanglements behind, associated with what's happening to the data and things like this. I would say that the concept that, I mean my own view is in the end somebody has to pay for the free stuff. And there are different ways one can imagine doing that. And the idea of essentially crowdsourcing the selection of nice pictures by making use of the users who are using the free to use stuff doesn't seem like a ridiculous thing to do so long as you tell the users that you're doing that. My initial reaction at least having thought about it only for the time I've been yakking at you. Let's see, another question here from M. Rudo. What is a good process for knowledge transfer from employee to employee? Boy, I'm not sure I've solved that one. In modern times, we increasingly record little videos and screen sharing kinds of things. That's a mechanism. We certainly have internal talks and things like this. I think in the end, sort of apprenticing the new employee to the older employees, so to speak, or not the fresher employees to the more seasoned employees is probably one of the better mechanisms. I don't think I'm an expert on how that's worked. I mean, in the history of our company it's worked fairly well. But I don't completely know the on the ground mechanism by which it works. I mean, I know that for individual roles there is a certain amount of let's train the next person how to do this. I think largely by apprenticing to the existing person. I think that in our company we're pretty good in terms of sort of recording archives, having internal documentation and so on. I think that the. Oh, you should read this internal documentation. I don't know how successful that is. I don't know. I mean, certainly at some level when people are into a job and they're like, how was this done before? Let me go read the internal documentation, let me go look at the archives. That's definitely a thing that happens, but I don't know how systematically this is, this is done. And I have to say that I think in a sense this is a general, kind of the general problem of education. How do you do that? I mean, in modern times maybe it's like ask the bot that's read all the correspondence of this or that person. I don't think that's yet a thing. Maybe it will Be. So I don't have a great answer to that. Um, Well, let's see. There's a question here from Heidi. If you had the chance to redo your life, would you choose differently in terms of what choices you made academically? Well, I don't know. That's pretty specific to me in the years in which I've lived, I think, because in. I'm not sure, I mean, I, I think, you know, I did. I was a traditional academic back in the days, I think, you know, back in the, at the end of the 1970s, early 1980s, I was kind of a, a traditional professor type character. And I think the things that I did at that time, I tended to do things that were sort of more innovative than the typical. And I sort of had license to do that because I had done things, particularly in things like particle physics, which were done at a time when those were the popular thing to do. And so they were pretty mainstream at that time. And so I had built up some degree of credibility and people would hire me to just to do that, so to speak. And so when I was doing all kinds of other more innovative things where people didn't know what to think about them, so to speak, it was like, well, but at least we did these credible things in particle physics and so on. I think in my own personal trajectory, I decided, oh, in my mid-20s, I guess, that I had developed this kind of area of science in terms of content and that I should sort of try and develop organizational structures around getting that science done. That was a mistake in the sense that it really was. To build those organizational structures is a different life from doing science, creating things and so on. It's. For me personally, it wasn't a particularly good fit. And also the timescales for doing many of those things are quite long. And although I was able to sort of make deals with universities and say, set up this thing, the, oh, government, why don't you set up a funding program for this? That's more like a decade type proposition, not something that happens in a year or two. And I think I didn't know that with enough clarity. I think in terms of the organizational structures that I set up and the kind of directions that I pushed these fields, I think there were some mistakes there in terms of not really distinguishing the sort of pure basic science from the, oh, look, there's this cool application over here kind of thing. I think in my own trajectory of. I mean, I think academia has become probably much more rigid in many ways in the 30 something years since I've not been involved in it than it was at that time. It's a general rule that smaller areas that have fewer people are less institutionalized, have much more flexibility than areas that have become very giant and industrial. And everybody has to follow the same rigid path. And I think academia has become more rigidified. I think in many areas, probably in areas where there is more sort of, it's a new area, nobody knows how it works yet that may or may not even be in academia at this point. There's plenty of work that in these years is done in companies and other kinds of places that aren't traditional academia. I mean, that was always the case. It's just that's shifted around. I mean, there was a lot of stuff that was done by essentially the government, whether it was NASA, whether it was government labs, whatever else. There was a lot that was not at universities, not in academia as such, although in some cases it was kind of academic like environments, so to speak. And then companies at various times have tried to have academic like environments with more and less success at different times. I think so. I don't know it that that's a. In terms of things, you know, I made various decisions. I haven't published a sort of an academic paper in an academic journal since 1986. A shocking fact is that many people who are academics, for example, who sort of interact with me and so on, wouldn't know that. They've assumed. Often people will just assume, oh, you must have been publishing papers in academic journals for the last, you know, 30 something years, 35 years, whatever it is, the. And well, actually I haven't. And the fact that people don't even know that makes me feel like, oh, it was the right decision not to do it because nobody would have noticed either way. As it is, the things that I tend to write, and I write a fair amount, as people may know, you know, I just post them and sometimes they get republished in different places, sometimes academic journals, sometimes other kinds of places. But I'm happy to say that the readership I get by just posting things freely, openly, et cetera, is rather good and much, much, much better than it would be in a typical academic journal setting. And for me it's much more flexible to be able to, to put things out there where I can have, you know, every picture, you can click it and copy out the Waltham language code and run it and get that picture that you don't get to do that if you publish in an academic journal. I also have the opportunity to be stylistically More varied in the things that I can write by the fact that I'm just putting them out there. And I don't have to worry about whether I'm sounding kind of as highfalutin as one might think one had to sound for a modern academic journal. I mean, it's worth realizing the things I write probably are stylistically very similar to the kinds of things that were written in early academic journals from the 1600s, when people were just writing as they thought about it and writing as they would say it to people, rather than trying to formalize the structure of it, which I don't think really does anybody any good. I don't think that it, you know, perhaps it makes some people think, oh, it must be more real and serious and deep because it's written in a highfalutin way. But in terms of the pure communication of content, I don't think the highfalutin does anything for one. I think it just can often sort of paper over things which were actually quite muddled. But it's like you say it in a fancy enough way and you can sort of get away with saying it. If you said it more directly, it would be very obvious that it was very muddled. And for me, for example, when I write things, part of my reason for writing things is to understand them myself. So it's kind of like the fact that I'm trying to write it in a plain way is the way that I'm forcing myself to really understand things. And I hope I'm managing to communicate those things to other. Let's see. There's a question from Yulin here about. I've written books. Do I have thoughts on child education? They say I have three kids at different stages. I find it challenging that school curriculums are trying to make conforming students while lacking the efforts in individual talent discovery. What are your experiences and how have I approached it? Well, I have four kids. They're all now grown up. And I think they mostly don't like me to talk in too much specificity about them in these public settings. But. I would say that I think everybody's pretty happy with the way they've quotes turned out in terms of kind of the. Well, what can I say about educating kids? I mean, my older three kids did various forms of homeschooling. Actually, my youngest one did for a while, but then. And that was mostly a kind of kid decides what they want to learn about, you know, writes ad to try and find teacher who can teach that reads a bunch of resumes. You Know, interview people, then sort of, then they learn stuff from people that worked really well. You know, a thing that was notable there was the types of people who did really well at that were not the same types of people as the people who could do well at classroom teaching. Typically, you know, classroom teaching is a different problem where you've got 20 something kids in a class. And you know, the number one issue is I think, can you keep the whole thing sort of driving in a definite direction rather than five of the kids are going crazy and things like this. If you're doing more one on one, you know, two on one, or, you know, one on two, whatever it is, you know, two kids, one teacher, or one kid, one teacher, it's a very different problem. And really what that's about is are you finding people who are sort of knowledgeable and passionate and capable of explaining themselves about the topic that they're talking about? Much more so than, you know, is the kid going to stay paying attention because the kid has chosen that subject, they've got a person they're talking to, that's not the issue, so to speak. And you know, plenty of people who did great with our kids were, you know, I'm sure, would have been kind of, you know, would not have been successful as classroom teachers. And occasionally we would have people who had been successful as classroom teachers and they were much too regimented for kind of a one on one kind of setting. But then I think, you know, my kids had a habit of, of doing sort of all kinds of projects. And projects are great because you get to, I suppose, sort of pace yourself. You learn how to do that, you learn how to produce something that you're really interested in, all those kinds of good things. I will say that a sort of exercise I did a number of times is kind of, you look at the encyclopedia and you say, what is there to learn about? And the kids will say, well, this thing's kind of interesting, that thing's kind of interesting. And if you ask what fraction of the encyclopedia is covered in typical kind of school education, it's not such a large fraction. And so there are all kinds of things that one might learn about. Now another question is, if you just learn what you want to learn, do you eventually sort of fill in everything? And the answer in my observation is by the time you get excited enough about learning things, you're going to just keep learning more and more and more things for your whole life and you gradually fill things in. I know for myself where I went to all sorts of fancy Schools. But much of what I learned, I learned by myself. And it's always been sort of, in a sense, amusing to me that there are areas where I just never learned it. I never had a reason to be interested. I never learned it well until 20 years later when suddenly that particular topic came up as a central thing that was in the line of some project I wanted to do. And then I learned it. But it's almost been something. I will say that the set of kind of holes that. And there definitely are holes and things that I know about. But I would say that over the years, as one's kept on learning more and more stuff when there's a hole. I almost find it fun when there's a hole because it's kind of like, like, okay, there's a thing I don't know about. Let me learn about that. I can probably learn about that reasonably easily if I really want to learn about it. So. But I think in, in terms of, well, it's, you know, different kids have different reasons to be sort of interact well with schools. Not so well with schools. I mean, my, my. I think, you know, it's always a question of is there a match, is there a kid where there's a school of particular kind with particular characteristics or whatever. And you know, if you're dealing with public schools, American public schools in Britain, obscurely, the word public school means. Means a private school. But that's for gosh, that's because back in the day, you know, kids would either learn at home or they would go out to a public, out in the world school. And so it got called that. But when I say public schools, I mean government set up schools as opposed to privately paid for and so on schools. But, you know, so you may not sort of have choice about what's on offer there. I mean, I will say that one feature of modern kids is their tendency to sort of stock themselves up with an infinite number of activities to the point where I will often claim that high school students are the busiest people in society today. It's probably more true of elite high school students sort of than the absolutely general one. But it's something that always sort of amuses me that they made themselves and the world has made them busier than anybody else. I think that's a mistake. I mean, it is the case in certain situations that you keep the teenagers as busy as possible and that way they don't get to do crazy things. But I'm not sure that the other side of that branch, which is then they have no Time to think about anything. And they're always like, you've got the next step that you should take and the next thing on some track that you should go down. And I don't think that's great. I think that one of the things that sort of some freedom of self determination, so to speak, does seem to be a thing that is very, very valuable to people. And if everything is on a track and everything is like, you do this because that's the way this is done. And you know, every hour of the day is scripted, it doesn't seem like the right thing. I mean, I will say that, you know, and in a sense, the end of that, I think, is often. And then you go into a very scripted profession, which may be a profession that's great and it's a good fit and that, you know, and you live a happy life and it's all good. It seems like a pity to me, although that's maybe from my point of view, which tends to be more the make it up for yourself rather than follow the script kind of kind of approach. I mean, I think there are, there are lots of other issues. I mean, you know, one of the ones I. Okay, this is a pet peeve of mine is kids who have no idea what their parents do for a living, which I always think is a pity because it's kind of like, well, you can learn something from what the parents do. They may like what they do, they may hate what they do. There may be things about what they do where you can get a sense of how it works and et cetera, et cetera, et cetera. And in my own case, my kids have been exposed to lots and lots of things that I've done and projects I've started. And they'll have told me that's a totally stupid project and then it will turn out to be successful. And maybe they learn something from that or they'll be, you should do this and it won't work out or whatever else. But I think it's sort of useful to get some kind of sense. I mean, you know, I know for myself as I try and learn about different fields, you kind of have to get into the spirit, the culture of that field to kind of know how things work in that area. And for kids, if they're sort of around their parents, their parents are in some area, at least the kids learn about the. Learn the ropes in that area, learn kind of the intuition. And the culture of that area may not be the area the kids want to pursue when they're Grown up, I suppose a funny thing for myself, my mother was a philosophy professor in Oxford and when I was a kid I probably said this a bunch of times. I would always say the one thing I'll never do when I'm grown up, it's be a philosopher. How could one be interested in a field where people are still arguing about the same things they were arguing about 2000 years ago? So said the 8 year old me or whatever. I'm not sure I would agree with the 8 year old me anymore. But it's sort of an interesting thing that I realized decades later. Yes, I knew a bunch of probably very fancy philosophers and they would find it entertaining to talk to, you know, the kid in the room because, well, I probably, I don't know, maybe it was this particular kid or maybe it was kids in general, I don't really know. But you know, the end result of that was that I learned a certain amount about how, you know, how philosophers think and which I didn't use for years. I don't, at least I don't know that I used it for years. Maybe I did use it, but only implicitly because one of the meta things about how philosophers, at least the, the then rather I would say not quite arrogant, but there was a certain degree of confidence in particularly Oxford philosophy at that time and there was a certain sense of the first question, but how can you know that's true? When you say but look, the moon is not made of cream cheese or something, but how can you know that's true? A philosopher would ask generically and in a sense that's a useful question to ask. And I suppose the fact that I was exposed a lot to that when I was, I don't know, 8 years old or something might have had some useful effect even though I didn't know it at the time. But that's a case where even though I thought nothing about that field is something I want to do, well, maybe I changed my mind decades later, but you know, it was useful to be exposed in that way. My father ran a company and you know, I was exposed a bit to that and that's kind of useful. When I came to start companies myself and although I didn't know the details and I didn't know, you know, I did have some sense of what, you know, what companies were like and you know, you have employees and you do this and that thing and etc. Etc. Etc. And that was again a useful kind of thing to have had something rub off on me. And I always find it Sort of a little disappointing when I talk to kids and I say, oh, what do your parents do? And they say, I have no idea. Or, oh, I think they're lawyers, but I don't really know what they do. It seems like a shame. And that's just one of my sort of pet peeves, I suppose. I think that what else to say about, you know, I don't know to what extent. And I mean, you know, it's like kids want to do X, Y and Z. Should they just go do those X, Y and Z things or should parents sort of say, you should do other things? I have to say I, you know, maybe it's not great parenting on my part. I pretty much let my kids do whatever they wanted, and they didn't really want to do things. They want to do things that were sometimes unusual. They didn't want to do things that were, you know, damaging or whatever else. And maybe that's the particulars of my kids. I, I do tend to think that the, you know, perhaps again, there's a certain amount of, you know, the, the, the things that I do myself are, you know, I, I, I don't do things that I sort of would regret doing or would not think that other people should do, so to speak. And that's, you know, probably other people have a different point of view about that and the, or different, different way of acting. And I think that's something that kids, you know, perhaps, you know, the, the, there probably are situations where parents are like, trying to get away with all kinds of things and the kids kind of see that and they model that, that type of behavior and, and that probably causes trouble. I mean, I, I don't know what the, But I, I kind of think, you know, whenever you get into the sort of, the whole contention. I mean, I suppose one thing that perhaps I was lucky about is that I didn't have kids until I'd been, I'd been managing people for, oh, gosh, how long? Probably. Oh, what is it, more than 20 years, 25 years? That right? Yeah, about. Or 20 years. At least 20 years before, know, before I had kids. And I, I do think sometimes that having seen, you know, having done a certain amount of management of adults, you are perhaps a little bit more relaxed perhaps about dealing with kids and so on. And you know, I, I sometimes by the time you're kind of, you know, when you're managing adults, by the time you're ordering people to do things, usually sort of all bets are off. You've lost the game by that Point. And I think, you know, often in the case of people dealing with kids, it's like you're ordering the kids to do something. You know, I, for whatever reason, didn't, didn't really ever have to do that and, or didn't do that for better or worse. And it's more like, you know, manage the kids, like, like you deal with the kids, like you, like you work with adults. My wife sometimes thought that was crazy, but it seemed to work out okay in the end. Just a few comments about that at least. Let's see. Ulan asks, do you still check your mail regularly? I find it a distraction for productivity, et cetera, et cetera, et cetera. Yeah, I get about five or six hundred emails every day and I grind them down. And that's my main way of interacting with the world. I don't DO texts or WhatsApp messages or other kinds of, or social media things. Those kinds of things, all of those that I interact with, they all flow into email. And I have all sorts of systems for dealing with email. That is my main mechanism. And it is nice and asynchronous in the sense that email doesn't have the expectation that I feel like texting has. Although I think for the young it no longer has this expectation that if you get a text you expect to reply immediately, so to speak. Email. It's just not the expectation. And for me, I'll let emails accumulate. I have know, a whole chart of, of that I will bring up of, you know, number of pending emails, number of unread emails and how it goes over time. Like for example, about a week ago I was at a horrible. Yeah, yeah, just about a, a little bit more than a week ago I was at a horrible maximum of 4,000 unprocessed emails. And it's always good when some, many of those are unread, but unfortunately those were mostly read but unprocessed like they needed things done with them. And I had, had been working on some projects that were fairly intense and I was like, I'm putting this off. And then I had to spend a day or two pretty much grinding down that email. And what usually happens with me is that there are emails where I will kind of, I know it's difficult, I will, you know, it's just complicated. It requires real thought to respond to it. And then I eventually get around to doing it, I'll do it. And then some that sort of require more thought. They age for a while and sometimes they, and I have people who help me, although I sort of gave up on the idea of sort of the primary place that the email goes being somebody who kind of reads it first. Because the fact is most email that I get the way that people, you know, my, my email address is publicly findable, that say, that's a choice I made years ago, which has worked out well. I mean, some people are like, let me hide my email address. But it's like, look, if somebody wants to contact me, I don't want to hear from them. Worst case, I delete their email. But if somebody wants to contact me or needs to contact me for some reason or, or whatever, why not let them contact me? Why make it difficult? And that, and that's worked out really well over the last, I don't know, 30 years or something that that's been the case. And I think. But, you know, I've tended to peel off things which really shouldn't be going to me. You know, like, I don't know, something that winds up being some technical or commercial thing about our products or something like this. It's just like, don't send it to me. I'm not, you know, I'm not the one to respond to that. We have a whole company that deals with those things and mostly those don't come to me. And I've got mechanisms to kind of peel those off. But then there's an awful lot of none of the above email that comes in. That's just stuff where, you know, if you tried to imagine what are the possible emails people might send a person like me, you just won't guess this particular one. And sometimes they're, you know, they're things of no interest, but sometimes they're like, oh, that's kind of interesting. You know, somebody wrote a biographical piece about my grandmother or something. You know, it's like, that's not a bucket of email you expect to get. Or somebody made a, you know, a dance performance based on some piece of science I did. Again, not a piece of email you expect to get. So there are all these or things where people are asking, you know, I don't know, I could, I don't know whether I can, you know, somebody was asking, oh, they're analyzing gestures made by people that psychologists or something or cognitive scientists. And they want to know, you know, am I left or right handed? Because they have a guess based on watching some video I did. You know, those kinds of things, you just can't, you know, you can't. And obviously that's easy to respond to and why not respond? And by the way, Ask some questions about, you know, that you might learn something from. So, you know, I've generally taken the point of view that there's enough none of the above stuff. Oh, and there's a large. The majority of my email is internal email associated with our companies and that's pretty well organized and it's stuff where I need to make some decision or whatever else. And it's presented usually in a pretty good way so that I can extract what I need from that email quite quickly and say, yeah, let me use my judgment and say the answer should be this or whatever. So I find that that works pretty well. I would say that I have thought about how to use kind of a bot of myself to respond to email and I think one thing that we will do is when people, and this happens quite a bit will ask questions about science or other things or questions of the kind that I try to answer in these live streams that we will have a bot that basically says, look, the question you've asked might be answered in the following transcript of some live stream or might be answered here. Let me give you that answer back. If you're happy with that answer, we're done. So to speak. Everybody came out ahead. If you're not happy with that answer, say, oh, you are a stupid bot. I want to actually communicate this to a human. It might take a couple of months before you get a response in that case. But we're not sort of cutting it off, but we're making it easier to get the more surface level response that, well, might be exactly what somebody wanted as a response, but it's something that we already had on the shelf and we didn't need to custom create and I didn't need to spend time on, Let's see, Gosh, Thank you for a comment from Harry here saying that they enjoy my writings and that they have a certain unique humor to them. And yes, that's a thank you for appreciating that. It's, you know, the starchiness of. I mean, again, I would like to say that I think in cases where people are writing very formalized prosecution and very kind of highfalutin prose, it really is too often a cover up for not really having a clear way to say things and which makes it more difficult all around for people to read. And by the way, one thing that for the last year or so I've been reading LLM generated summaries of papers about various kinds of things and it is frustrating that the LLM gives me these, these boring flat summaries but they're much easier to read than a lot of these abstracts. All right, one last question and then I think I need to go because I am late for something for my day job. Let's see. Wow, this is, this is an unusual one. I'm not sure I can do much with it. From Frictionless. An interesting handle, given the question. I want to do something that should be an Internet based business and thermodynamics and thermodynamics and its application should be a central part of the business which friction is kind of part of the picture of the second law of thermodynamics that you have kind of systematic motion and it gets turned into heat. Thermodynamics and an Internet based business. That's an interesting combination. I think that in. There's, well, probably a lot of things there. I mean from the. Oh, I've got a source of random numbers that I'm feeding out to the world based on thermal noise, which is a business that people have had, in fact still have kind of. Can you, you know, let me, let me sell you some randomness, so to speak. But more, I would say more. I think the relationship between thermodynamics and economics is probably a quite close one. And the sort of equilibrium in thermodynamics and kind of the equilibrium of prices in economics probably has. There are probably a bunch of analogies there. I would say that it's sort of. Okay, so here's a thought for you. If you look at quant finance, quantitative finance, back from the earliest days of kind of quantitative finance and things like the black Scholes model for options pricing, things like this, those are stories of the heat equation and of diffusion models and so on. And in a sense, and this is a rather bizarre answer, the. That's kind of a thermodynamics story that is sort of played out in quantitative finance. And I'm being told I really need to go to my day job thing here. So I'm sorry, but the end result here is that the CEO can't spend all their time having fun answering questions on live streams. We actually have a new product that we're trying to get out and I need to go meet with the team that's working on that new product. So thanks for joining me here and see you another time. Sat.