Business, Innovation, and Managing Life (June 14, 2023)

Hello everyone. Welcome to another episode of Q and A about business, innovation and managing life. I see a bunch of questions saved up here. Jamie is asking here, how do you balance work and life? Do you have advice for men maintaining a healthy lifestyle while working? How do you find time for socializing and exercising? Well, I like my work, so I do it seven days a week, many hours per day. I have been for, well, decades, I've been sort of taking time off at dinner time, so to speak, when it's been sort of a constant. If you look at personal analytics data that I have that shows outgoing emails sent, you find this big band around 6 to 6, 6:30 to about 8 local time. Where, where I am, that's a band where the, where the email volume goes down because I'm, I'm off having my dinner or whatever. But other than that, I, I tend to, I like to work and, or what people would consider, many people would consider work. I don't know whether, for example, doing these live streams, does it count as work or does it count as entertainment? I'm not sure. Hopefully, you know, I, I do these because I like to do them. I find them interesting. And it's the same with most of the things I do. I do the things I do because I find them interesting. Now, as a real reality, when one's doing a big project, running a company, whatever else, there's always a certain overhang of you, like doing the basic thing that you're doing, but in order to get the thing you want to get done done, there's a certain fraction of the time that you have to spend doing things that you may not be as thrilled about doing as you might be. And I try to find, you know, most kinds of things like, oh, this is going to be really boring and oh, I don't really care about, you know, this particular piece of sort of software engineering build out. But then I get into it and I've found almost uniformly that once you get into something, there's interesting things to be done there, particularly when you try and do the things well. It's usually the case that to do things well is not as easy as you might at first imagine and requires a lot more kind of intellectual engagement than you might have thought. And for me that makes them interesting. So that's always a, a nice feature. Now, I think in terms of healthy lifestyle and exercising and so on, I've, I've taken a very, I, the, at the beginning of my day, I will go and walk for basically the first two meetings of my Day, which is about a little bit less than two hours usually. And I have to say I think I'm on a five year streak right now of walking more than 10,000 steps every day. And I like to, when I can, I walk outside. And for a while I was trying to have a kind of elaborate system of first computer with a kind of a popcorn popping type, a popcorn selling type type thing that I put the computer on. And then I went to tablets, iPads and so on. And then I pretty much gone down to just a phone. And I'm often in meetings with people where it's necessary to see the screen. But I found it's like, okay, somebody else share the screen. I'm not going to be able to type when I'm walking, at least with that setup. But I find that so long as I've arranged the meetings that I'm doing earlier in the day to be ones that are as we describe them in doing scheduling walking meetings, that works just fine. I like to walk outside. I found that, I don't know, it's generally. Well, I even noticed years past when I did sort of an unintentional controlled experiment. I found that my resting heart rate was lower when I had been walking outside on my walks for a few weeks as compared to walking inside on a treadmill. When I'm walking on a treadmill, I've figured out a long, long time ago now sort of how to set things up so you can have a keyboard and you have displays and so on, and you can walk up to about two and a half miles an hour and still be able to type just fine. It's kind of the trick is you tip the treadmill up a bit. So you usually have it up like 5 degrees or something. And that means you're doing more work to walk and still walking at a lower speed so that you still can successfully type. And I think so long as you kind of set things up so that the, the kind of mechanics, the biomechanics are right. I find I can type up to about 2.5 miles an hour. After 2.5 miles an hour, I find it difficult to kind of maintain sort of my, my fingers in the right place on the keys while I, while I walk. But I, I have certainly for me kind of taking that time, you know, walking a bit every day really, really helps. I think every so often I sort of every so often is every quite a few years I kind of reset a bit the things that I'm eating and try and sort of the take things which, which don't really, don't really help but just provide more, more calories or something and, and try and take those out. But that's a, that's a gradual process. I think the other thing I tend to do, I'm a very consistent, I have a very consistent schedule and I sort of try to go to sleep at more or less the same time every day. I actually just moved my sleep schedule back a bit because have a number of things coming up for which it's useful to be up earlier in the morning and also as it gets hot out on the east coast of the US at this time of year, being able to walk outside before it gets really hot in the day is a good thing. So. But I tend to be very consistent in that and I tend to try to sleep a decent number of hours every day. And I have to say it's perhaps a sign of an unduly unconstrained life or something that I basically don't use an alarm clock more often than once every few months for some very special reason and usually because I'm pretty consistent in, in, in my kind of schedule during the day, I find I, I wake up actually embarrassingly close to when I think I should wake up within five minutes or so often. And so those are, those are good things. And I, I suppose for me the, the things I do are things I like to do and that always sort of helps me lead a happy life, so to speak. There are things which, you know, for each person, there are different kinds of things that different people find stressful. You know, for me, I tend to find sort of being in charge. I don't find stressful being in a situation where I'm trying to sort of consensus work with a lot of people to, to sort of all collectively agree on something. I, I find that fairly stressful and difficult and, and not what I like to do. And so I tend to avoid those kinds of situations. And that sort of has a effect on the things one does. But I think, you know, one, one, it's a, it's a common thing to just try and find the things that you like to do or you're good at doing and then figure out sort of where in the world is there a niche where that fits in. And I've been lucky enough to have a niche that involves running tech companies and doing science and so on, where the things that I find I like to do and maybe even a reasonably good at doing are things that fit in with that niche. I think in terms of Jamie Asks here about socializing. I suppose I'm one of these people. I work with people much of the day virtually because I've been a remote CEO now for, what is it, 32 years. And so I've always been doing kind of remote things, usually not with video, very rarely with video, usually just audio. But that still feels like kind of a social interaction. There are people other than me involved, so to speak. But I also like to sort of do things on my own. And I spend, most days, I'll spend many hours just working on typically creative and scientific type projects where it's just me, so to speak. And I like that too. I think the, I find sort of the formalized, you know, I'll go to events from time to time where one does a certain amount of intense socializing. And I find that. One of these days I'll have to get my wife to join in. I don't know whether I'll be able to do that, get her to join in one of these pieces of commentary about life and times. I'm not sure what she would say about my habits of life. She's always quite critical of my kind of, I would say sort of assisted life where I delegate many kinds of things. She's like, you don't any, you don't know how to do this or that thing anymore. It's like I really, I do well and then I try it and I find out actually I don't. But in any case, I think that the, the thing that I, I find I have been lucky enough to accumulate a very interesting network of sort of friends and acquaintances over the years. One of the things that I do make some small effort to do is to maintain that. And I suppose I've maintained it a long time as evidenced by the fact that, for example, I recently did a kind of 50 year virtual reunion of kids from my elementary school class. And I hadn't kept up with all of them, but I kept up with quite a few of them. And that's obviously over a long span of time. And I've just found that insofar as I've met people that I find interesting over the years, that the difficulty of keeping up with them is not very high and the value is high. And it's always very strange when you talk to people you haven't talked to in decades and you realize you're picking up the conversation from just where you left off. Now, maybe it helps that I have a reasonably good memory and I tend to. Tend to kind of remember those interactions. And so it's easier to sort of pick up from where one left off. But I think people have a certain. There's a certain similarity of the way that one reacts to people that sort of persists in that way. But certainly for me, that's been something that has enhanced my life is to sort of maintain a. A long term, very large network of people I know in all sorts of different kind of activities and walks of life and so on. And over time, there are people where their fortunes will rise, their fortunes will fall, and I kind of stay in touch through all of those kinds of things. And that because I'm mostly interested in the people rather than the fortune, so to speak. And that's been something that for me personally, I found valuable. Let's see, There's a question from Wilf here. When will you retire and devote all your time to your physics project? Or is retiring out of the question? You know, it's terrible. At my kind of ancient age, a lot of people I know who are my contemporaries, you know, are retiring or have retired in fact, because I happen to start my kind of career in science and technology and so on rather young. A lot of people who I know who were my sort of sort of work contemporaries, maybe as much as 20 years older than me. And so they're very much in the kind of retirement mode at this point. And so I get an awful lot of questions from people I know saying, hey, when are you going to retire? And I'm like. I look at them and like, I'm not going to retire. Why would I retire? I like what I'm doing. And, you know, I have to say that, you know, people always make the claim and I think it's true that the single most medically dangerous thing a person can do is to retire. And somehow it's kind of when one is, you know, in action with. With sort of things one is looking towards doing that keeps one kind of vibrant and, and alive, so to speak. And when one's just like, well, I'm just hanging and I don't really have a thing that I'm trying to do, that's a much more difficult situation. Different story. If. If people have kind of the quotes hobby, that is only a hobby because it isn't something you seem to be able to get paid to do, but it's something that you really want to do, that's a slightly. I suppose, in a sense you could say that might be my situation in that I like doing science kinds of things which are not something that. Well, I suppose I Used to get paid to do that when I was a professor and so on. And I suppose in principle one can get paid to do that, but that hasn't been what I've done for the last few decades. But I think the thing that I found is that there is a certain energy that one gets from being sort of exposed to real issues in the world and solving kind of practical problems. At least for me, that's something that I find quite energizing. And a lot of things where it's kind of like, like, okay, there's a practical problem, but then that practical problem leads to some intellectual questions. And many of the things that I've done in science were the results of essentially something that started at some level as a practical problem that then sort of I realized, oh, well, given this practical problem, then there's sort of a question that is slightly more basic. And then eventually you're, you're going down all the way to the foundations. And you realize I have to figure out these foundations to be able to, to go back up, even if I want to just solve my practical problem. But also somewhere in the middle, there's all sorts of really interesting intellectual work to be done and interesting science to be done and so on. I mean, to give an example, right now one of the things of interest is one's looking at LLMs and generative AI systems and so on. And there's kind of the question of, okay, do you really need hundreds of billions of weights in a neural net to be able to do a good job of reproducing kind of the production of human language? And it's something, we didn't know that it was possible to reproduce kind of high quality human language with any kind of automated system before about six months ago. Now we know it's possible, but does it really need hundreds of billions of neural net weights to be able to do it? Almost certainly the answer is no. Almost certainly. We now know it's possible with a neural net and so on. Now we ask what is the essence of what's going on there that makes it possible for that to happen? And what I strongly suspect is that really we are seeing certain aspects, certain sort of core aspects of the structure of human language and the structure of kind of human reasoning and common sense and so on, and that there is a sort of almost a theory of those things that is really a quite small theory that doesn't need hundreds of billions of neural net weights to be able to implement it. But there's some essence of that that's Much smaller. And I think I've talked about this on some of these live streams before. But the kind of. The big example of this is something like logic that sort of deduced from looking at the typical forms of argument in pieces of text that was deduced back a couple of thousand years ago as this kind of formalized version of a structure of kind of human language, of structural elements of human language that corresponded to this sort of formalized thing that is logic. Well, my guess is that there's a lot more of that that can be lifted from human language and from what we might consider common sense reasoning, but it hasn't been done. It's kind of like one of those missing pieces of science that didn't get done for a couple of thousand years. Now I think we get to do it. And I've been interested in. I've just been getting started on really working on that. The result of that will both be something that I think is of significant sort of foundational intellectual interest, but also something that relates directly to these kinds of practical questions. And in part, the way that I kind of fell into thinking about this now, I thought about these questions in the past for slightly different reasons, was digging into really understanding ChatGPT and making technology around it and writing kind of an explanation of it and these kinds of things. And so that sort of very practical activity then led to this, what is ultimately a quite foundational set of questions. And this has happened to me many, many, many times. So for me, having this sort of driver of practicality and sort of connection with the world I find useful. I also will say that the effort that I put into kind of exposition, explaining what I'm trying to do and explaining things in general, I find that that's extremely valuable to me, just purely if all I wanted to do was to optimize kind of the intellectual things I can figure out and the intellectual work I could do, it greatly contributes to that, the effort that I put into exposition of various kinds. It really helps me clarify my thinking about things so that I can move forward, even just for myself, so to speak. So it's. To me, I mean, I've been. I've been lucky enough that most of the time the actual sort of practicalities of running a company are things that I somehow, you know, I've been doing it long enough now, basically been running things for what, 45 years now, my current company, for 36 years, that in the end, one gets fairly experienced. And yes, there are always crazy things that happen that have never happened before. But One has a certain, with a certain degree of experience, it's possible to make decisions reasonably quickly, to not get too agonized about different kinds of things. And so for me, that's not a kind of high cost of the value of having a company that has sort of the machinery to be able to execute projects, amplify ideas one has and so on is far outweighs for me the kind of personal kind of drag in a sense of doing some of those kinds of things that are just the practicalities of running a company, which I suppose again, after one's been doing it enough years. And also I suppose the other constraint is if one's goal is to make the biggest, strongest, most commercially successful company ever, then that's a different set of things that one's pushing for. My goal has always been to make a company where I'm pleased with the things we build, where we have a good set of customers and, and employees and so on, and where, you know, we can do something which I feel good about doing in the world, so to speak. More so than let's maximize our, our financial returns or whatever else. Fortunately, in my company we don't have like, outside investors and so on who are, who've been sort of, you know, who've had pitchforks for all these decades saying, you know, why aren't you making more money for us? That's something where it's kind of, it's basically on me to, if, you know, to that I'm the main beneficiary or one who loses with those kinds of decisions. So the answer is I'm not planning on retiring. And it is very nice to see that, for example, in our company there's lots of younger people who are very energetic and I, I hope that as time goes forward, they'll, they'll take on more and more of what has to be done at the company. But I don't expect that I will sort of zero myself out anytime soon. Let's see. Nightmare asks, have you ever thought about constructing a business model based on the principles of our physics project? You know, when one goes, one sets about studying sort of fundamental theory of physics. One basically thinks of oneself at least I think of myself as kind of in that project, checking out of anything that is about business models and so on. I kind of my, my statement when I was starting to do that was people said, well, you know what applications are going to have? And I said, I don't think it's going to have any applications for 200 years. It turns out that Wasn't true. Turns out that the formalism of the physics project, these ideas of multi computation and so on, do seem to have applications. And they have applications in the quite short term to things like distributed computing, possibly to things like molecular biology and sort of the practicalities of biomedicine, also potentially to machine learning actually. And I've just been exploring that very recently of kind of the analogies between quantum mechanics and the different paths that you can take in sort of both inference and training of machine learning and so on. And where the formalism of our physics project may be relevant, I don't know yet. But it's also the case that a bunch of the ideas, particularly ideas about the RULIAD and so on, seem to give one a bunch of inspiration about ways to think about things that are much more sort of directly applicable in the world today. And I think, for example, the whole multi computation idea and the way to sort of organize computations in parallel using the kind of physics type notions seems to be a very valuable direction. We had a project which didn't come to fruition yet, the so called quantum money project, which was about taking the formalism of physics project and using it to sort of understand the different version of distributed blockchain kinds of ideas. I mean, in ordinary blockchain you've still ultimately got this one ledger that's recording the transactions that happen. There may be many copies of that ledger, but there's still one consensus ledger that is telling you this transaction happened, then this one, then this one, and so on. And the problem with that is that to sort of settle that ledger takes a while, it takes 10 minutes, it takes 24 hours, takes a while. But often, particularly if the transactions are quite small, you don't really care whether you're off by 2% in some settling. You know, that's something that you can gradually sort of settle out and get it right. But you know, you'd really like in a second to be able to know, oh, is this roughly right type thing. And it's kind of a money uncertainty principle. You can know your bank balance exactly or you can know it quickly, but you can't sort of know both of those things together. Well, anyway, using the formalism of our physics project, I think it's possible to create kind of a distributed blockchain system that has that kind of property. We started doing this, didn't really finish it. It's one of these kinds of things where it's, it's a complicated kind of strategy question because one can figure out sort of the theory of how to do it. But until you really engage on some of the practicalities, you don't really know quite how valuable it's going to be. But to engage on the practicalities, you actually have to start building large software systems and worrying about registering all these crypto kinds of concepts and so on and doing all the, all the sort of now nowadays increasing amount of regulatory work and so on around crypto kinds of things. And that's a, that's a lot of work. And it's something that is, you know, when my interest there has been more in the kind of theory of economics that are associated, that one can think of being associated with these kind of multicomputational processes. And. But I wanted to kind of ground that theory because, you know, it's easy to have a theory of economics in a sense. It's, it's. You kind of know, you know, what you're talking about if you can actually build, you know, a cryptocurrency, if you can actually see transactions happening, if the things you imagine in your theory are actually what you can observe happening. But that's the thing where sort of bringing all those things together haven't really done. It's always the case when one thinks about sort of intellectual things and their relationship to business. It's always for me a question of there's an intellectual idea and there's some business opportunity, but the business opportunity has pieces to it that involve practical engineering, build out, practical development of distribution channels, practical kinds of marketing things and so on. And there's sort of this question of that's an irreducible amount of work typically to do all of those things to really make practical in the business world some idea that one has. And for me, one of the things I've tried to do with our company and so on and with the products we've built is to make it be the case that for a certain spectrum of ideas we have a kind of a matrix into which to put those ideas. So that given some idea about how one can do some new kind of symbolic programming construct or something, or some new thing that makes the world computable in a different way. That's the kind of thing where given one of those ideas I have a matrix into which I can put that idea, where I have the distribution channel, have the whole engineering build out and so on. So it's a sort of pure idea that flows through to something that's practical in the world. But when it comes to something like, okay, let's deploy a new you know, a new type of, a new form of cryptocurrency. That's something where one has to build that whole business structure to be able to do that. It's the same thing when I've been interested in molecular computing for a long time and sort of thinking about what it would really take to build out something in that area, I think getting closer to being able to do that for a variety of reasons. But you imagine, okay, I'm going to make this molecular scale computational thing and what am I going to do with it? And I can kind of make it in some lab or something, but then there's a whole sort of question of how does that actually happen? Do something in the world. And yes, you can sort of do the academic distribution channel where you say, I'm going to write a paper about it. But for me, I always find that, I mean, I like writing about things and I write a lot. I don't write academic papers. I like to think I write in a more lively style than one finds in typical academic papers. And I write in a way that I like to write and that I think people like to read, which is a different optimization from writing in a way that is kind of the formalism, the formulaic writing of modern academic papers. The kinds of ways I write tend to be a lot more like the ways that academic papers were written in, like the 1600s when things like academic journals first came into existence, when it was much more like kind of blog type writing than the sort of more formulaic writing that's emerged today. But in any case, let's see the thing that if one's doing something like let's make some piece of molecular computing, well, yes, one can write about it, but I feel like to really know what one's got, it's much better to actually build something practical. And then one has this whole question about how do you build the sort of distribution channel in effect for what one has? I mean, I always noticed that I suppose for me, you know, I have ideas about lots of kinds of things. I tend to try and control myself to only have ideas about kinds of things for which I have some plausibility of a distribution channel. You know, I was, I'm reminded years ago, I think I mentioned actually earlier in this live stream, you know, walking on a treadmill and typing on a computer and so on. And I, you know, have figured out how to do that and figured out the biomechanics of it and so on. And it was like people were saying to me, oh, you should Patent that, that thing about, you know, how you do that. And it's like, well, yeah, you know, but what on earth would I do with it? You know, you should, you should market this, this sort of thing that is the, the way of setting up, you know, sort of retrofitting stuff to a treadmill to be able to type on a computer with it. And it's like, yes, I mean, in principle I could do that, but that is something for which I absolutely have no distribution channel. I have no kind of, I have no structure for doing something like that. It's a whole bunch of sort of irreducible work to go and sort of connect with that part of the world, so to speak, and just wasn't, wasn't at all worth it as far as I was concerned. So it's, it's, I mean, I, I tend to, I tend to sort of like to say, okay, I'm going to build a good matrix for the kinds of things I like to do that lets them be practical in the world and then feed, you know, restrict the things that I think about to things which can be fed through that matrix. And I, it's something I recommend for anybody because I think it's a thing where, you know, if, if you're a person who likes to have a lot of ideas, it's. Well, I mean that people have different, different kinds of set points in this regard. I mean, there are people who are like an idea every five minutes and sort of little interest in the kind of follow through from the ideas. And there are people where it's not so much kind of the spark of the idea as it is the kind of long actual building of the project and the sort of operational construction of things. And people have different set points about what they like to do. I think it's kind of, for me, it's not so much fun to just have raw ideas that never go anywhere. And it's also not so much fun to do things which are sort of purely operational and don't have kind of an intellectual spark to them. So I've tried to sort of set things up so that I can operate somewhere in the middle of those kinds of things. Let's see, People ask all kinds of questions here. Aaron asks random task delegation question have I ever changed a flat tire? The answer is yes. A long, very, very, very long time ago. And I can't even remember the circumstances, but I do remember thinking, you know, I know something about physics, I should be able to figure out how this all works. And it more or less I could figure it out. I think it was in days before, certainly long, long before cell phones and things like that. It was a. Gosh, I'm trying to remember. I remember sort of jacking up the car and sort of figuring out how to, you know, remove bolts and things like this. But I, I don't remember whether I think it wasn't my car, I think was a rental car and I think it was, was a long, long time ago. So it's. So the answer is yes, although only once. And I think in modern times it's not so much really a thing because they're usually these run flat tires that let you go on for a while. Let's see. Lala asks, have you ever researched health benefits of certain foods? Does this influence your diet? I pay attention to what people say about sort of the health features of different foods. I have to say it's a, it's a complicated thing to pay attention to because there, you know, when one tries to dig into the basic science, which I have done a few times, it's kind of complicated and messy and it's not really clear what effect is the most important effect. I mean, there are certain things that are pretty clear about sort of, you know, foods prepared in certain modern ways are kind of, you know, are sort of, you know, they may taste good at the moment but aren't really very good for you. I mean, there are sort of basic things like that and things that were kind of the, the fruits that people could have picked back in the day off a bush or a tree or something, and those are probably sort of natural and okay type thing. And I guess my tendency has been to kind of pay some attention to what kind of the big waves that go through of, oh, it's, this kind of diet is great and that kind of diet is terrible and so on and to try to just sort of be sensible. And I think also for myself, I do notice that if I eat some terribly greasy, goopy food, I feel terrible. So I don't eat that stuff very much. And I think one's body tells one a certain amount about what's good and what's bad if one pays any attention to it. At least that's been my increasing theory as I get older. I think that in terms of kind of the details of, you know, the, the magic food, I've not tended to go in very much for that. I have for many, many, many years. I've, I've, I have a, a good friend who's a medical researcher who told me probably what is it 40 years ago now, said, you know, pointed out that, you know, things like coenzyme Q10 was probably going to be a good thing and, you know, fish oil was probably going to be a good thing and you know, et cetera, et cetera, et cetera. So I, I always, I just have a little pill sorter and every day, you know, every morning I just, you know, take the, the pills from that day's collection and stick them in my mouth and swallow them. And, you know, that includes various kinds of, I suppose, supplement type things, but nothing particularly wild. I've sometimes wondered about the wilder ones and I sort of do do a little bit of research and it's kind of like, well, I don't really know. And I seem to be doing okay as it is, so I'll just keep doing what I've been doing. But, so I haven't really, I haven't been somebody who's deeply researched that I would say that, that it's very easy to not be sensible and to eat things that are clearly, you know, not, you know, not very good for you. And I've, you know, I, I've tended increasingly to just avoid doing that. Although I have to say for me, the, the challenge is always for me, if I get really hungry, my brain doesn't work so well and I, and that's bad. And I, and then I tend to, to, you know, it's kind of like I, I feel good with the effort of kind of staying without eating for a while just because I, I feel virtuous, so to speak. But it's actually a mistake because then I, you know, my brain doesn't work very well and, and I end up overeating when I, when I sort of come back. And, and I mean, I, when I, it's not like I, I go for long periods fasting, but if I haven't eaten for, I don't know, four or five hours after the beginning of a day, I will get to that. So I'm afraid I don't probably have anything terribly insightful to say about the foods and diets question. Let's see, Navalji is asking, do you now write with the help of an LLM? No, I have to say I've used LLMs occasionally. You know, in the past I've sometimes used a thesaurus. Now I tend to use an LLM for that purpose of, you know, I just can't think of the way to say this. There's got to be a simple way to say this. There's a word I'm. I'm not thinking of Ask the LLM. And it does pretty well at those kinds of. Those kinds of tasks. I would say that the. And occasionally I've asked it things which are kind of sort of average questions about the world, so to speak, and kind of things about sort of overall trends and overall things. Things that you can't. That aren't really a keyword search kind of question. They're things where it's more like, you know, what are the statistics of the world about this? Although you don't ask it that way, it's been quite useful for that. And I think in. Well, another thing I found recently really useful is summarizing papers and so on and even just taking abstracts and summarizing them. I have to say it's made me not appreciate a lot of modern academic writing because you take an abstract of a paper and you think, oh, well, then I can figure out what the paper is about. And you're reading this thing and it's like it has a whole formulaic structure and it's like, what the heck are these people actually trying to say? You feed it to an LLM and say, just summarize this. And it comes out with a nice clean summary. And it's like, okay, great, now I know what these people were talking about. Now sometimes the LLM will probably inject its own nonsense there, but honestly, when you're looking at a large collection of papers and you're wondering which ones of them might be interesting, if you miss one, well, you miss one, you might have missed one. Anyway, having the LLM in the loop doesn't seem like it's a bad thing. And to me, that's been a pretty valuable. That just started doing that now. It's a pretty valuable thing. It seems to be able to do that. I suppose I haven't done that for certain long emails I get and things like this. Now that I think about it, I should start doing that. Because one thing that I do do when I get email from people outside of my company and circle and so on, you know, I have a little system that automatically searches for the sender of that email in my archives. I have good archives of about 3 or 4 million emails that I've sent and received over the last 30 years, more than 30 years now, and also all of my scanned documents from earlier than that time. So I have a pretty good record of every sort of body and everything that I've interacted with over that period of time. You know, assuming it ever hit anything in written form, which it pretty much will have done whenever I knew the name of somebody or something like this. So I have a good archive. So that means when I, you know, when I get a piece of email I have a little bot that automatically searches those, those archives and does various other searches and so on, gives me a summary of those things and it sends it back to me within a few seconds in the thread that the original email came in at. So I can sort of immediately see, okay, yes, oh yeah, I interacted with this person in 2007 and that reminds me of something and so on. And then it's a very useful way to contextualize the interactions that one has. And now that I think about it as I'm talking to you guys, I realize for anything that's at all long I should get an LLM summary of the thing that got sent to me because that way I'll actually be able to read it. If I say make a two sentence summary, at least the LLM will say something which I will look at as opposed to somebody sent me a long thing that may be really interesting, but when I'm processing 500 emails I'm just not going to read it at that time. So that's a way I should be using LLMs. Let's see, Several questions here. The Paula is asking what does a typical summer look like for me? Well, I have a few definite things because we have our summer school and our summer camp and et cetera. For the last 20 years we've done a three week, well, three to four week long summer school for people from around the world, usually undergraduate to graduate student to postdoc ish level people. A summer school about science and technology and kind of our way of thinking about and doing things. And it's usually about 70, 75 people I think. And that's a three week long, it's really now pretty much extended to four week long extravaganza that's coming up actually next week. It starts this year. And that for me is a, is a. It's. It's always very interesting because it's. I meet all sorts of interesting people, many of whom stay involved with us in one way or another over the course of many years and many of whom kind of build it very interesting careers often based on things that they were doing at the, at the summer school and such like. And for me my role in the summer school tends to be the person who tries to figure out what project each person should do. And I'm actually terribly remiss this year because I usually make up a list of a few hundred sort of potential projects that seem like they're worth doing for people to try and pick between. And unfortunately it's always a little bit bad conditioning for me because, because I come up with all these projects which I think are great projects, and then people, it ends up that only a small fraction of those projects actually get picked by people and done. And usually the projects are some amalgam of things that we've sort of, I've thought about and things that somebody has, that the student has, has sort of separately been interested in. But you know, one of my big roles there is to try and make this match of inventing projects, connecting projects with students and so on. I said, I'm remiss this year because I haven't made my project list this year. I always accumulate over the course of the year a large number of kind of ideas for potential projects, but then I have to write them all out. And were it not for the fact that I'm in the middle of doing a science project right this week, I would have probably done this by now. But I haven't done it. I probably need to do it in the next day or two. But in any case, so for me that's sort of an intense three or four week period around that. We, we also, not only do we have our sort of summer school for grown ups, we also have a kind of junior version of that for high school kids, another 60 people or so. That's always lots of fun for me. I also get involved in figuring out the projects for those folk. And there are sort of many pieces of things with the summer school. I'm happy to say this year the grown up and the high school summer programs are both in person. I learned a number of years ago that because we have to arrange these programs long in advance, I better arrange them fairly near where I live. Because one time we arranged the, the summer school actually in Italy and we'd arranged it like, oh, nine months in advance or something. And that was right around the time when we released WolframAlpha in 2009. And it was like, oh, of course we'll have WolframAlpha released long before the summer school. Well, it didn't happen that way. In fact, WolframAlpha got released like two weeks before the summer school. And it was a huge mess with me being in sort of places in Italy with bad cell phone coverage and things like this and trying to deal with, you know, WolframAlpha release plus two weeks type thing. And so after that experience, we're always like, let's just do the summer school in a place where I know I'm going to be and I know I'm going to have sort of infrastructure and so on. So anyway, so we get to do those in person. We also have a middle school summer camp that we added a few years ago, that is, I guess. I guess was added during the pandemic. So it started virtual, and it's still virtual, but that's another thing that's interspersed in those weeks. And then I've tended to. I mean, I don't know, my family has sometimes taken me on various trips over the summer. I think sometimes it's been a where can we go? Where you can't sort of interact with the locals and so on, where there's no conceivable work activity to do in that place. And I think the most extreme version of that picked by my younger daughter was quite a number of years ago. Well, not so many years ago now was Svalbard, which is a sort of northernmost inhabited place, which is sort of a very obscure place. It's a very interesting place. It's kind of like. I imagine it being a bit like what the Wild west must have been like of sort of a frontier town type thing. But in that particular case, a frontier town that's sort of somehow vaguely attached to Norway and so on. But any case, it was like, there won't be any. The theory was it's still remote, there won't be any sort of place you can drop in and visit there. But it turned out there was a group that does ice physics that actually uses our technology and so on that I happen to find is there. So that turned out not to be true, that there was. Even in that remote location, there were interesting people, sort of locals to visit and so on. But usually that's a little while during the summer doing those kinds of things. I tend to that. But I don't think, as far as most of my work is concerned, the summer is very similar to the rest of the year. I have to say. Sometimes there are things where there's a. Well, you know, it's always a funny thing because I get a lot of invitations to a lot of different events, sometimes virtual these days, often in person these days and in the past. And I'm. I very much. I'm. It's been very difficult for me to figure out, should I go to these events, should I not go to these events, when should I go, when should I not go? And so on and I sort of made a decision back in 2019 that I was going to stop going to business conferences because I just found that I, you know, I. It's nice to meet the people, but I meet people in all sorts of places and wasn't really having a great sort of payback for our business and so on. And it wasn't something that I was finding intellectually that interesting. And so I was sort of going to cut back on that. But then the whole landscape of conferences changed as a result of the pandemic and so now there's a whole new generation of these things and it's unclear what to do with them. But you know, for me, I find it's, you know, what I tend to do when I'm traveling is I tend to try and stuff as many things as possible into the sort of travel that I do. So if I go to place X, I always maintain a list of sort of who do I know or what do I want to visit in place X. And I've been maintaining that for years and years and years. And when I go to some place where I think I'm probably only going to be here once in my life, you know, it's always good to have accumulated these are things I want to see people I want to see things I want to see there and then do them. And I tend to take the point of view that if I'm going on one of these trips, I might as well stuff it full of, you know, morning to night meetings, events, et cetera, et cetera, et cetera, and just not even think about doing my kind of normal everyday work during that time. You know, I try to keep up with sort of email traffic, but I don't do kind of real time interactions during, during those times. I'm not sure if that's the best optimization, but it's the one that I've tended to adopt because I find that, you know, I'm traveling somewhere and then there's some scrap of time, can't really get much done, then I might as well, you know, given that I'm going to be traveling there, this is my one opportunity to be in Place X and go visit this or that, you know, organization that's in that place. So why not do it then? But that tends to mean that for me it's kind of a, A, you know, any trip I do is kind of a blackout period for projects that I'm doing. And, and that's, it's a, it's a complicated trade off and I haven't haven't quite figured out how best to deal with that. And again, it's been interesting to me to see that sort of what's come back after the pandemic is quite somewhat different from the kinds of events that existed before that time. And so, for example, one of the things that's crazy thing is that I've often thought, let me analytically figure out when in the year are the most invitations coming in and from what parts of the world and things like this. And let me be sort of scientific about saying, well, I'm going to have this gap in the summer where nothing's happening and I can go off and do this and whatever else. And it's very frustrating because, you know, I have plotted some of this data and it is extremely variable from year to year, but it clusters in different years. Like for example, I know in this year, like in October, for example, in mid October, there are I believe, four events in different parts of the world literally landing on the same days and the, and similarly, you know, and then there are times when it's just like, oh, there are these things that show up in August when there weren't usually events at all and now suddenly there are. So it's been very hard for me to have a kind of a good scientific approach to when in the year am I going to do this or that thing. I think for me, I suppose it'd be an interesting question. I haven't really done the personal analytics to know this. When in the year do I end up initiating the most projects? I know many summers end up being times when I initiate projects. I'm not quite sure why. Perhaps, perhaps there is. Well, I don't know, perhaps some things are less intense then. And so I end up being able to finish things I've been doing previously and get started on new things. I'm not really sure. The question here. Let's see. RBS comments that they were just looking through some of their old emails, found that they ordered a new kind of Science on 13th of July 2011. Okay, all right, that's, that's all right. I was, I was wondering whether you would quote a date that was before when it came out, which was May 14, 2002, but no. Well, thanks for getting it then. Cody asks, has being a remote CEO become easier over the years? Do you ever miss being physically in the office? Being a remote CEO has become less unusual, obviously. I have used screen sharing, web conferencing type stuff for close to 30 years now. And that's the most important tool for me in terms of sort of interaction and collaboration with people. I think that. I mean, I have to say back when I first was being a remote CEO in 1991, it was. I kind of. It's kind of funny to think about this these days, but, you know, my main connection, I guess that was before I started using. Doing web conferencing. So it's been. Yeah, I've started doing web conferencing probably around 93 or so. But at the very beginning it was kind of like. Well, I had a computer. It was actually a next computer, a couple of them, I think. And you know, I would be getting email, but the email will be delivered through a phone line and the computer would dial up every 15 minutes and go fetch its email. It seems incredibly primitive these days, but that worked okay. And then when things had to be sent otherwise, I had a fax machine connected to my computer and things would show up on the fax machine and so on. And then I would, you know, download files and things like that. It's. I think in those days I was probably much less structured in the way I used my time and I would. I was working on. In 1991, I. I was. When I kind of was. The reason I became a remote CEO was in large part because I wanted to work on a science project that I thought wasn't going to take that long, but ended up taking 10 years my project to write my book, A New Kind of Science. I sort of decided I wanted to kind of sort of separate myself enough that I could really concentrate on doing that project. I also. The company was quite young. I'd started the company in 1986 and we released our main product in 1988. The company was growing pretty rapidly, but I had many more ideas than a company of the size that we had at that time could possibly absorb. And so it was kind of like I was talking about sort of matrices into which to put ideas earlier. It was sort of a frustrating thing because I could feed in ideas much more rapidly than the company could process them. And so that was something that for me just was. Would get frustrating because it's like we could do all these things, but like, well, we can't actually do them because we just don't have the machinery to be able to do that much stuff. So for me it was better to sort of throttle back on sort of effort spent on the company and sort of feed it ideas that the rate it could actually absorb them and put more of my energy into doing the science project which Is, which is what I did. But at that time I would say that my sort of daily calendar was considerably less structured than it is today. But I think during the time I was working on the book for 10 years, all that time I was a remote CEO. I think it was actually quite healthy for the company as well as for me in the sense that the company built structures that could deal with the fact that the CEO is not in the office every day now. There was a time actually in the middle of that 10 year period when we were trying to get out actually version 3 of Mathematica where it's like, this is a total mess. I really have to spend. I spent pretty much six months more or less in the office, give or take at that time, and really pushing to get things to happen. But mostly that wasn't either necessary or desirable in terms of what's it like being in the office versus sort of being a remote CEO. If you're a remote CEO, you have to be self motivated. But I am self motivated. That's sort of why I do the whole CEOing thing in the first place. I think that there are. Being in the office, there tends to be an awful lot of, oh, you know, if the person's right there, I might as well just ask them something and it probably isn't necessary. It's probably not a good idea. It's probably doesn't, you know, it's probably kind of a waste of time now. You know, do you get sort of the occasional kind of, oh, I happen to bump into this person and so this idea came up? Yeah, you know, that can happen. But I think by the time you really develop sort of this geodistributed way of working, you can surface those kinds of things in lots of meetings, talking about different kinds of topics and in meetings that are really tagged as brainstorming meetings, you can surface those ideas pretty efficiently rather than just, oh, I happen to bump into this person. And for me, I suppose it's very easy for me to sort of get distracted if I'm in the office. And people are worst cases. It hasn't really been this way for many, many years. But you know, when it's people kind of like knock on the door or just, you know, you leave the door open and they just sort of poke their head in and say, you know, oh, can I ask you this? Or that thing? It's like, it's much more, much more efficient for me at least. Just like, let's get the email and you know, I'll read it sometime. In the next day or something, and then I think about it and I'll respond rather than interrupting something I'm doing. I mean, you know, for example, right now I, I basically never ever get incoming phone calls or anything like that. I suppose, I suppose modern people don't get phone calls either. They get, they get text messages and things like that instead. But I have, you know, I don't get those things from sort of the general outside world at all because I find that for me, it's important to sort of concentrate on the thing I'm actually thinking about doing and get myself really into the thing I'm thinking about doing. And as soon as there's sort of an interrupt, it kind of ruins that. And so I, I tend to avoid that and just sort of have my time separated off. You know, I'm always getting email coming in, and if I really sort of want to take that break, I'll look over at my email and, and try and process some of that. But I don't like being interrupted. It's much, much better to sort of interrupt myself in that regard. But in terms of, I mean, our company, as a result of my sort of example of being remote, our company has had more and more and more remote folk all over the world. And companies adapt to that. We adapted to it decades ago. And, you know, I think the thing that doesn't work very well is when there's a team and a few of them can be sort of in a conference room together, and then the rest of the team is remote. That tends to not be a very good dynamic. It's a kind of us versus them type. Dynamic. But by the time everybody is remote, or at least everybody is, you know, web conferencing, so to speak, rather than going into that conference room, that works just fine. And I do know that at our company, it's kind of shocking the extent to which people who are just down the hall from each other will actually be talking through, you know, through web conferencing rather than by walking down the hall. But it's probably from a business productivity point of view, I think that's probably a good thing. So, you know, in terms of, of things like doing events and so on, you know, there was a period of time, oh, 20 years ago or so now, when I decided I really didn't want to travel very much. My kids were young and it was, it was kind of a, a just, it's like, why do I need to travel? And so I really actively didn't do that. And I said okay, instead of traveling, I'll do a video conference with some event or another. And people, people like that, because we're kind of a novelty item to do video conferencing. Occasionally I would send a telepresence robot just for added sort of interaction, so to speak. That worked sort of okay. Not great, but okay. Kind of led to some, some amusing and fun things. But I would say after a while, so the novelty of video conferencing wore off and it was like, kind of like, oh, you know, well, let's do this event in person. And then it's like, well, let's not. Because I don't really want to travel there. And so my. But I think in, you know, in the pandemic, I was noticing that one of the things that happened was that a number of very outlying places around the world that, where it'd be very, very unlikely I would travel there. And you know, maybe once in my life I would, but not more than that, were like, okay, let's organize, you know, an event. And I was sort of impressed with that initiative. And indeed that caused me to have interactions with people sort of in parts of the world where I really would be very unlikely more than probably once in my life to, to go there. And so that was kind of a nice thing. And that sort of made it easier to be remote CEO, so to speak. I think that, yeah, I wouldn't have said that in terms of. I have to say that the sort of web conferencing, screen sharing technology has really not at some level changed that much in a really long time. And so that's not been something, you know, it's been something one can, you know, it works very nicely and smoothly and has a nice user experience these days, but wasn't terrible even decades ago. I would say that what I found in response to what's it like physically being in the office? I found more and more that being in the office is largely, you know, when I visit, you know, main headquarters or other offices we have, it's really mostly social and not mostly sort of business purpose type type thing. And it's nice to see people in person every so often. But it isn't really achieving direct business purposes in my experience. It's sort of. I. I feel a little bit like, my gosh, why am I here? This kind of waste of time. I'm just, you know, and I might as well just organize. Oh, let me just go meet a bunch of people, just sort of quote, socially, so to speak, rather than even pretending I'm Trying to do real productive work. When I'm visiting the office. Aaron asks how helpful was it to have already built SMP before starting Mathematica? So I just to give a bit of history back when I was, well, 1979, so I was 20 years old or whatever, I was sort of on a track of being a physics type person. But I had become sort of the world's largest user, I think of sort of mathematical computation systems and it sort of outgrown the systems that existed at the time and so made this decision, well, I better build my own. And I started that in 1979. First kind of released version was in 1981 and that caused me to start my first company and things like this. Then by 1986, for all kinds of reasons, SMP the thing I built was, you know, the company had gone off in sort of a not so useful direction. And the product, there were all kinds of issues. And so I'm like, okay, now I'm going to sort of try and make the version of this that I really always wanted to make. And that's what led me to start building Mathematica in 1986. Completely fresh start. I mean, no, no code in common, only, only ideas in common, so to speak. How useful was it to a built smp? Unbelievably useful. I mean it's, you know, in smp I did many, kind of had many sort of radical ideas about symbolic programming and things, some of which were quite good, some of which were, were too extreme for people to be able to use them. And also the whole process of doing the large scale software engineering for smp, the whole way of explaining it, the way people reacted to the way that, you know, you wait five years, you see what the training seminars about the software product are like, then you realize some of the things people say in those training seminars, you're like, oh my gosh, that was a design mistake or that was a strategic mistake in the system. Also in terms of the setup of the company, very different first time around where I kind of brought in a CEO to run the business side of the company and so on, versus the current time when I've been CEOing it myself for better or worse for the last 36 years. So it was hugely useful. I mean, I think it's often people say when they're talking about software systems and so on, you know, plan to throw one away, plan to make one and then, you know, see and then throw it away and then make the one you really wanted to make. And then usually the Other line that people say is, but you never will. Well, in my particular case, for reasons of complicated circumstances, I did that. And it was hugely valuable because a lot of the things that, as I say, were sort of too extreme in their design in SMP I didn't do again in Mathematica, but I took the things which were good ideas and built from them. And that was probably the first several years of things that we did in Mathematica were things that were sort of. I already kind of knew how to do that from experience in smp. I mean, there were a lot of things about kind of the software engineering practices, I would say, in the industry in general, but also from my learning about things, they definitely were better. They were not actually that terrible in smp, but they were better in. In Mathematica and what's now from language. But yes, the answer is it was hugely useful. And psychologically, it's a strange thing. You build something, then you throw it away and you build it again. Usually that doesn't work out very well. In this particular case, perhaps because the thing I was building is pretty difficult to build. It worked out really well because sometimes when you like, I'm going to do this the second time, it's kind of like, oh, I know everything about what I'm doing. You know, it's all just plain sailing and easy to do. And that's usually the time when you don't do it very well. In this particular case, it was kind of like, you know, I could see the first time around there were all kinds of issues and all kinds of things could have been done better. And it's like I had an opportunity for essentially business reasons, to just make a fresh start. And so that was. That was a good thing to do. So it worked out really well. Guillermo asks, in terms of building software, was the biggest challenge you ever faced? And how did you solve it? Was it building Mathematica or was it earlier on? Look, the hardest thing, the thing that is most intellectually demanding is overall design, in my view, is kind of the functional design of the system. How do you make the right decisions that are going to be sort of the bedrock on which you construct the system? How do you. How do you build things that you're not going to regret later? And it's taken me a long time. I think I've gotten very experienced at this point at doing software design. You know, you can kind of watch it in action because we live stream a bunch of our internal software design meetings. But it's kind of like how do you. How do you. Every time one does that sort of software functional design of, you know, what's going to be the language design for the system and so on, you really have to drill all the way down. And every time I haven't really drilled all the way down and really understood the essence of what's going on correctly and perfectly every time I've not done that, which is very rare in these years, I've regretted it because you end up building on mud, so to speak. And it doesn't work out well when you really can drill down and really understand the essence of. And then you can successfully design things and sort of successfully build on solid bedrock, so to speak. But I think the most difficult thing is sort of continual vigilance of doing hard software design. And the fact that it's about to be the 35th anniversary of the launch of Mathematica on June 23, it will be and it's pretty neat that one can take kind of a Mathematica Wolfram language program from 1988 and it will still run today. And that's because, well, didn't make too many mistakes in that Design back in 1988. Actually we just found somebody just found a version 1.1, an unopened version 1.1 Mathematica distribution box for the Mac. And so I'm hoping on June 23rd to open the box and try and I've got a Mac from that period of time and try and stuff the floppy disks in and see whether I can get version 1.1 to boot up and run and then see whether we can take a version 1.1. I think version 1.1 was the first one that actually ran on the Mac without. Well, I'm not sure. I have to look back at the. The details of that. I think version 1.0 wasn't. Was really a version only for workstations class computers in any case. So I think this is a t equals 0 Mac version. And I think the sort of maintaining sort of vigilance for 36 years to make sure the design doesn't go astray, to make sure there aren't things which in another 10 years it's like, oh my gosh, that was a bad idea and that it's now we have to make an incompatible change and so on, that's a difficult thing. It really requires a sort of strong commitment both for me personally and now for our company and for many people who were sort of aligned with this mission of, you know, keep the design clean and good. That's Been that's the single most difficult thing in my experience now in terms of the actual mechanics of building the software. You know, when we first built first version of Mathematica, I wrote a large chunk of the code about eight people, seven, eight people who worked on the original version one and the, in the, in the kernel I wrote a bunch of the sort of core functionality. I'm always a little bit traumatized when people tell me that code I wrote and things like the pattern matcher is still running today. Because one of the things about a lot of that code is that you end up, it's complicated and it ends up being a lot of custom algorithms that you have to invent. And those algorithms, one of the things that I think I learned to do was sort of judgment based algorithms where okay, you've got this thing, it's got, you know, it's mostly going to hit some cache. It's going to be, you know, you can have, you have something that, you know, you're kind of guessing, oh, it's enough to have a, this size of cache that's probably going to be okay. You don't measure it, you don't have a way to measure it. You don't have, you know, using usage yet. So you just have to kind of guess. And there's a lot of kind of judgment based software engineering that I did back in those days, which actually worked out. You know, when it has been measured, it's worked out surprisingly well. And there are a lot of detailed things like, I don't know. One of the things, I don't know whether this is a thing these days is collisionless hash tables. So what I did was once, you know, where you're putting a bunch of things in memory, you hash those addresses and the question is you change the hash code, you change the hash coding algorithm until the particular addresses that you're trying to hash don't collide. And turns out you usually can do that very quickly. I certainly didn't prove a theorem that my collisionless hash strategy would eventually converge and that at startup you wouldn't get some crazy infinite loop as it tried to build collisionless hashes of things. But that's an example of some algorithm that I just invented in a few minutes at some point and that's in the code. And turns out it's worked really well actually for the last 36 years, so why worry? But you know, in, so there are a lot of sort of small scale software engineering kinds of things. I remember probably the hardest piece of code in version One of mathematical morpheme language is the pattern matcher. And I do remember that that has a piece where it's sort of doing its thing and matching patterns and partially matching things and so on. And I think it has three stacks running all at once as it tries to recurse in different ways on different things. I mean, another feature of that code, sort of interesting again is that I was, you know, very concerned about optimization of it and I broke it into, I don't remember how many, 30 different cases. And again, it's very judgment based programming. It's kind of like I'm going to pick this case, this case, this case, I'm going to optimize those code paths. I'm going to take the cost of figuring out at the beginning. I think the way it works is it records what kind of a pattern match thing this is going to be and then it goes down that code path. But even knowing, sometimes you have to test, you know, is it going down this path or that path. This is in a sense standard software engineering. But what's interesting about it is that it's, you know, in that particular case, in developing that system rather quickly, I had to make a lot of judgment calls about how to break up all these cases. And mostly they worked out well. In fact, I don't really know of any examples where they went horribly wrong. But, you know, so that was, that might have been really, really difficult. But I suppose I had enough experience with the software engineering and so on that I was able to make these judgment calls and get them. It seems more or less right. Had I not had that experience, those probably would have been super difficult things to write because I would have not known what to do and I would have tried this and tried that and had to test a lot of things and it's too big a space to really test everything and a big complicated mess in terms of more modern times. I mean the, the thing that always tends to happen is these big distributed systems are hard to debug because, you know, they're running in real time. There are lots of different things, there are lots of different kind of real time conditions that develop. And I have to say I, I've probably at the rate of, about, what is it, once a year or two, you know, I've kind of jumped in to try and debug complicated things that our teams haven't been able to debug well. And it's always a little bit horrifying to see just how complicated these software systems end up being. But it's Also something where I have consistently found that just using Wolfram language as an analysis tool to go through and like, okay, give me that log file, it's got a hundred thousand entries. I'm going to go analyze, you know, and bin these different entries. Maybe I even use a little bit of machine learning here or there. Certainly I use visualizations of all kinds and you know, go in and do these analytical kinds of things. I found that that's a really powerful tool in terms of debugging, even these distributed system kinds of things. I'm trying to think in terms of earlier software engineering adventures. I suppose. My very first, very first computer that I used was a machine with ferrite core memory, 8kW of 18 bit memory and a simple instruction set with 16 instructions. It was a pretty fast for its time. I think it had a two microsecond cycle time. And on that computer, well, debugging was certainly a, a funny activity. And there was some. One of my early efforts was to build a physics simulation which I never managed to get to work. And years later I realized that the fact that I didn't get it to work was scientifically very interesting. But that's a quite different story. But the process of debugging on a machine written where one was writing in assembly language and where it had a teletype and it had a tape reader, paper tape reader and so on, and that was, it's, that was, that was not trivial. I think that the, my kind of, probably proudest achievement on that computer was writing a paper tape loader, which is something where you would, you know, the tape was running through this optical tape reader, so running pretty fast. And the paper tape had little poles punched in it and sometimes a little piece of confetti would get stuck in one of those holes and then, well then the computer would read the wrong thing and the program wouldn't work. And so you had to kind of detect was it wrong or not. And so I wrote something which did. It had some very basic error correcting code where you'd run through this whole bunch of tape and it would do a checksum and at the end it would say, oh, something went wrong. It's like drat. You know, I just had all this tape that went into this, you know, bin and I had to roll the tape up again and it was a big mess to, to go back and redo it. So I wanted a paper tape loader where it would tell you sort of more or less immediately when it had gone off track. And so I Wrote what I guess was an early error correcting. Well for me, early error correcting code. I didn't know about official error correcting codes at the time, which would sort of in the. As. As the tape was, was running through the tape reader, it would check blocks and if the block failed it would stop in the tape reader. And then my sort of proudest achievement was you could physically pull the tape back in the optical tape reader and tell it to start rereading again and it would correctly align itself and be able to read that block again. And that was a bit non trivial because you had to have something where I don't actually remember how I did it, but something where the thing could know sort of what, what where it was that the block started and so on and align itself correctly. But you know, that was a different style of software engineering from the one that one gets to use today. I suppose from, from those constraints I kind of took the point of view that that one should in the end, you know, in building Wolfram language my goal has been to define sort of a formally correct in the sense of being sort of a clean formalism for making things computational and not to kind of pander to the details of the particular computers and hardware and memory allocation and you know, color maps and whatever else that exist at any given time in the development of computers. But more think about the raw concepts, the sort of idealized concepts of computation. And that's worked out really well for, for us. Well, let's see. I probably should wrap up here. I think I need to go back to my day job. There are all sorts of interesting questions here which I think I should leave to another time. I think the next one of these probably falls within our summer school and I might try and do one of these with a live audience. I've never done that before, but maybe I will try that this year at our summer school. That will be fun. Talking of what's virtual and what's not. So. Well, I look forward to talking to you all again then and should, should wrap up for now. So bye for now.