Business, Innovation, and Managing Life (February 15, 2023)

Hello, everyone. Welcome to another episode of Q and A about business, innovation and managing life. And I see a bunch of questions here. Ollie is asking, how did you begin your journey into the live streaming world? It's not something I see most CEOs doing, and they say it's enjoyable to see someone like me being available in an open capacity like this. You know, I realized it's been about five and a half years since I started doing live streaming. And originally, well, we started with a bunch of internal meetings that we have that I always thought were really interesting that are about software design, process and so on. And somehow the. The process of designing a computational language is the most concentrated sort of intellectual activity that I know in terms of you've got some domain that you're working with, and designing computational language around that domain is kind of the way that forces you to drill down with as much clarity as anything on that particular domain. It's kind of more so. I thought that was an interesting process, and I thought, what the heck, maybe other people will find it interesting too. It turned out it's had many benefits. It's very useful for us. Just recording these meetings, I think is very useful to us to see sort of instant feedback on things from users and experts and so on who watch these live streams. It's been great for recruiting. It's of people who ask all kinds of interesting things on live streams, and it's like, hey, who are you? You know, maybe you should come work on this stuff. Sort of on the other side of the. Of the live stream, so to speak. I think that that was sort of the. The first thing that I did in terms of live streaming was live streaming, these internal software design. We're in a interesting position in the world in terms of our technology situation because, you know, I started building this kind of direction of technology, of kind of making the world computable, so to speak, well, more than 40 years ago now. And we built a very unique tower of technology. So I don't think we would view ourselves as having kind of competitors breathing down our neck, so to speak. And I think it's perhaps a different situation for other people. But, you know, we built something where the real challenge is that we built a bunch of things which maybe sort of artifacts from the future. Kind of 50, 100 years from now, everybody will say, well, of course one will build that. But as of right now, we're mostly sort of the challenge is mostly to have people understand what it is that we have today, as opposed to wait 50 years and then and then have them understand rather than, oh, somebody's going to take that little piece and make use of that and so on. So that's, I would say also. Yeah, so that's, that's one piece of live streaming. That was the first thing we started with. Then I, well, let's see. Then the, I guess the pandemic happened and in the first sort of week or two of the pandemic it was like all these kids are out of school and what are they doing? And it's like, what can I do to help with this? And I was like, well, you know, I, I, I enjoy interacting with kids and so on and I'm, I know a certain amount about science and technology and things like that. Let me just put on a kind of live streamed Q and A about science and technology for, for kids and others because I knew perhaps other people might find it interesting. And so I started doing that three years ago and I found it really, really interesting to do because people ask all kinds of interesting questions and I find it the exercise of trying to just sit in front of a camera and explain myself, a very helpful exercise in terms of clarifying my own thoughts. And there are an awful lot of things where over the last three years I first realized how to explain something while I was sitting doing one of, one of these live streams. And that's very helpful to me. And I must say that in these live streams that I added later about business, innovation and managing life, I found the same thing that there are things where, well, I do something some particular way and one of you will ask, well, why do you do that? Or what do you think about this? And I realized I've never really actually with clarity just sat and thought about that. One doesn't usually, one just usually does what one does. I mean, I like to believe that I spend, you know, I, I do a lot of things, but I also like to spend a little bit of time kind of planning what it is I should be spending my time doing. And I've always sort of wondered, you know, what percentage. It's a, it's a single digit percentage of time that I spend planning. You know, I, I know people where it's sort of multi, you know, double digit percentage of time that they're spending planning. And that's often a bit silly because it's like I'm planning, I'm planning and planning, hey, when are you actually going to do something? Or that's kind of off on the, on the side, so to speak. But I, but if you are Like I never plan, I just do the next thing, next thing, next thing. I think that's a mistake. I think one should spend a small part of one's time planning things. And these live streams help me kind of think about a bit more abstractly things that, that are sort of come up in my, in my life now. You know, I found there are some other sort of live streamed series that I've done. So when we're working on our physics project, we're first doing our physics project. Well, in the preamble to announcing it, we didn't live stream that. We recorded a lot of meetings which we then put on the web subsequently. But we didn't live stream that because I thought it would be too incredibly confusing. Plus we also didn't know where it was going to go. And it could have been like, well, we're trying to make this fundamental theory of physics but whoops, doesn't really work. So sorry, come back in 30 years. So that wouldn't have been that exciting. But we did record all of those meetings and as I say now put them on the web. But after we launched the physics project, then we did our kind of internal working meetings for that in a live streamed way and did a bunch of those. I think we kind of stopped doing them a little bit because some of the topics we started talking about were a little bit less, what can I say, they required more background. Also had the very practical issue that they were sort of happening at times of day that aren't terribly convenient for most people, so not as good for real time presentation, so to speak. Then I started doing another series which is kind of interviews and discussions with people. I really should do more of those. I just did one yesterday with my friend Terry Sinofsky, who's a person who's been involved with neural nets for, oh, close to 50 years now. And we were talking about kind of the history of neural nets. I thought it was a really interesting discussion. I encourage people to check it out. It's out there on the web, on YouTube and things. But I find these discussions, particularly I would say about history with people very, very worthwhile. And unfortunately there are people where I'm like, I really want to do this history discussion. And then the, this just happened. Recently one person I really want to do this with died. And I was like, I was, I was sad that he died, but I was frustrated that, that I hadn't managed to get my act together, so to speak, to do that discussion which will sort of now never be so. But in any Case I, I've that this is a, the, the point with doing these kind of historical discussions is to do them well, one has to do a certain amount of preparation and it's just there's only a certain number of hours in the day, so to speak, but they have a certain time urgency, particularly when one's dealing with. Much older people. And as a matter of fact, just next week I may do a few because I'm going to an event where it's a conference about complexity. It's a subject that I was sort of involved in initiating in the 1980s. And it's a little bit shocking to me that I think I might be one of the youngest speakers at this conference. And I consider myself at this point ancient. So in any case, that's another thing. I may do some live streaming of that if I have a chance. But yeah, so that's what I've done there in terms of, of kind of the openness of answering questions, you know, that is, I have kind of led my life in a way where I like to believe that I can kind of explain everything I'm doing and I'm not embarrassed about the things I'm doing. I might make mistakes, but, you know, I, I, and so for me it's kind of a, a, It's both helpful to me to understand things and, and it's something where I feel like perhaps some of the things that I figured out might be useful to other people. And, and that's a good feeling to be able to pass these on. So. Okay. Qwerty is asking, do you ever get overwhelmed with the decision fatigue and dealing with so many topics and the feeling that you will never get everything done? It's a risk, you know, I just. Oh, when was it? Two weeks ago now. Posted what's about 500 pages of material that were an attempt to finish a project that I started when I was 12 years old about 50 years ago now, and which I've been sort of steadily making progress on. It's a science topic about the fundamental origins of the second order of thermodynamics. It's, it's a topic which many people think was solved 150 or 100 years ago. It wasn't. I think we now have something much closer to a solution and has many interesting consequences. But the fact that I was sort of finishing a project that I started 50 years ago is a little bit shocking. I have been, you know, for a while, I was a little bit pessimistic that a bunch of projects that I thought I wanted to do at some point would ever. I would ever get around to. Actually, in 2019, I kind of decided. I kind of. I. That there were a number of projects where if I was ever going to do them, I really had to get on with it and do them. And. And the first of those was my effort to find a fundamental theory of physics. And I was. I was fortunate. There are a couple of young physicists who kind of came on board to help work on that and helped add a lot of ideas and energy to the project. And the fact that. That in about six months was really very dramatically successful is a great encouragement to actually finish things that one thought one should do. That project on fundamental theory of physics, I had sort of first thought about that direction for working on this in the early 1990s, and had done a bunch of work on it in the 1990s, then a little bit of work in the early 2000s. But I was sort of put off because in that particular case, the target market for my work there was really professional physicists, many of whom are friends of mine and so on. But basically, at that time in that field, everybody was like, we have it solved. We've got a path we can. We can find. You know, we've got. We've got things from string theory and whatever else that are going to, you know, solve this problem of finding fundamental theory of physics. We. We don't need anything different. And people really. It seemed like a huge uphill battle to kind of have anybody sort of care about the things we were doing. And. And so that kind of put me off, and I thought, look, if the target market doesn't want this stuff, I've got plenty of other things I want to do. Let me work on those other things. So I was glad that I think both the things we figured out were much stronger than what I had figured out in the 1990s. And really, I think we've nailed it. It's really very exciting. Plus, the world of physics had changed, and there was much more interest and appetite and enthusiasm for kind of new ideas than there had been at the beginning of the 2000s. So, you know, now, for example, there's a project that I've been meaning to do for many, many years that has to do with finding what I call. It's not a very good name, and I really need to rename it a symbolic discourse language. It's the problem of taking kind of everyday language and making kind of a precise computational representation of it. We have, with Wolfram language, we have A precise computational representation of many kinds of things about the world. But we don't have a sort of precise computational representation of things which we say in everyday language, you know, I want to get a. I want to eat a piece of chocolate or something. We have a good representation of the piece of chocolate. We don't have a good representation of the I want to eat part. But I become. I've been interested for a long time in whether there's an extension to things like the ideas of logic. You know, logic is a kind of a way to take what would otherwise be a kind of human statement about something and make sort of a precise framework around it. Come on, do that more generally. And actually the most recent impetus, I've been meaning to get to this, but I sort of got derailed again by the physics project, but now with kind of the success of ChatGPT and its friends, it's kind of a very clear indication that there is more structure to human language than we knew. And Aristotle sort of noticed this when he originally invented logic by sort of taking things that were making statements of rhetoric and noticing that you could kind of abstract from all these statements of rhetorical statements, you could abstract this kind of structure that he then made into what was originally syllogistic logic. And I've sort of realized that what ChatGPT has discovered is that there are other regularities in language. There are other kind. There's a sort of a construction kit of things that you can put together to make a sentence that could plausibly be meaningful. And it's a little bit shocking that in 2000 years, I think not a lot has been done that sort of follows up on sort of Aristotle's. He did it in rather primitive way at the time, but it was 2,000 years ago. So different thing. But in any case. So that's a project that I'm just now embarking more seriously on. And it's a project I've been meaning to do for 40 years or so. For me, it's kind of a lot of fun because I work on a bunch of projects where the kind of time span, I should say I made comments like the things we're doing is sort of making artifacts from the future that will sort of seem obvious in 50 or 100 years. I feel pretty confident in making those statements because I've kind of watched over 50 years kind of the arc of what gets figured out and what people think is obvious and so on. And I've also studied a lot in the history of science and can Kind of see how those sort of arcs of progression work. But a lot of things I've worked on have been things where sort of there's been a big piece of progress, for example, in physics, big piece of progress 100 years ago. And I think we're able to sort of take a new step now based on sort of new ideas that particularly come from computation and so on. I also worked a bunch in the last year on mathematics and the kind of foundations of mathematics. Again, it's a field where a lot of progress was made right at the beginning of the 20th century and then it kind of sort of decayed. And I think now we've got new things to say that, you know, I'm, I'm, I'm pleased with being able to say them, but I would say that the, the sort of ambient development of kind of the computational ideas that I've been certainly much involved in, but they, they come also from the world at large. Those are the things that make that possible. So I've, I've worked on things which have 100 year, the second law of thermodynamics thing is like 150 year type timescale. Let's see if we can nail the things that, that people started thinking about at that time. But I haven't come close to working on something where it's a 2000 year time span. So that's one thing that at least I'm kind of interested in thinking about this symbolic discourse, language idea. But this whole question about what does one work on next and how does one deal with all these different projects? One of the things that is critical to me is that, and perhaps it's irrational, is that I'm quite an optimist. So, you know, when I said, okay, I'm going to finish off writing about the second law of thermodynamics, that was in August of last year. It's like this is going to be very straightforward project. I've thought about it for 50 years. I think I've more or less figured it out. Let me just write it down. Okay. Turns out writing it down was about 500 pages. And turns out it took several months to do that. But at the beginning my assumption was, look, I can do this fairly easily. Let me just, you know, charge forward and do it. If I had thought to myself, oh my gosh, this is a six month project, I think I might have had a much harder time kind of jumping into doing it. So, you know, sometimes projects are quicker than I expect. So it's not the case that I'm always irrationally predicting that, that things will, will take less time than they do. But somehow for me, when I do projects, you know, I'm usually doing projects that have never been done before, but that's what I'm interested in. And so it's very hard to make a clear prediction of how long it's going to take because, you know, it's like, okay, let's make a time estimate for finding the fundamental theory of physics. Let's budget it out. Well, it's hopeless because, you know, in the history of our civilization that hasn't been done before. So, you know, you can't estimate how difficult it's going to be. It's like there are things where, for example, in working on fundamental theory of physics, there are things where my first estimate is it's going to be 50 years before we know how that works. And then we figured it out. It was really exciting. There were a bunch of things which I really thought were much on a much longer timescale than they turned out to be. But in terms of projects I do, it's, you know, for example, the Symbolic Discourse Language project. I kind of think it's reasonably easy to make a certain kind of progress given all the things that I've developed before. But it might not be, but I need that optimism to really get sort of the, the, the enthusiasm to do it. I would say in leading projects with other people, one of the things that is one of the most important things to do in my experience of leading sort of projects that have never been done before is to sort of create that kind of optimism. And it's like, yes, we can really do this. This project is really doable. And you know, it's what, what tends to happen for me is I have a, a vision of this project should be doable. And then sometimes I realize the original goal isn't quite right. So you make kind of a mid course correction. You don't. And that mid course correction, you know, sometimes is taking you to a much better goal. But I, I certainly particularly remember in the developments of Wolfram Alpha now, I don't know, 14, 15 years ago now, at the beginning of that, everybody was saying, look, this project is impossible. There's all kinds of evidence. You know, people in the AI community have tried to do things for years. People, et cetera, et cetera, et cetera. It's just not doable. I was pretty sure it was doable. And an important part of my contribution to it, in addition to a bunch of technical figuring out of things, well, was just, look, it really is doable. And what that really meant was whenever people come to me and say, look, we've run into this roadblock, it's impossible, then that was my cue to really try and figure out, okay, let me kind of roll up my sleeves and try and figure out the solution to this engineering problem. And sometimes the solution was to change the question. It's, sometimes it's like, look, that's not the thing we want to actually do because that really isn't doable. But let's do something that will be very useful to people that is kind of, you know, avoids we, we've gone around that barrier rather than just sort of bashing our heads against that same thing. But yes, it is a, it's a, it's a big issue for, for somebody like me kind of picking what projects to do. I think that I've tended to maintain a certain collection of projects which are kind of always rattling around for me. And I kind of wonder when do I really get to jump in and do this project? And it's partly a question of what sort of ambient understanding there is in the world, what tools I have at my disposal, what people I'm sort of interacting with, and so on. I mean, for example, one area that I'm very interested in is molecular scale computing, nanotechnology, these kinds of things. I've been interested in that for a long time. I had decided ages ago that it would take a decade or more to build kind of the technology stack to do, to make a particular kind of bet on how to do nanotechnology. And I'm like, I'm not up for doing that. It's not what I think is the most valuable thing for me to do conveniently. Now there's been a bunch of sort of ambient technology development, particularly automated lab technology and things like this, and ambient understanding that's particularly come from our physics project. That means that I think there really is, is an opportunity now to think about sort of molecular computing. And I, I hope to do that. And that's, that's sort of something where at some moment I think I might decide I've really got enough sort of tools set up that we can really do something and make progress in a fairly short time. But yes, there are, there are always projects where I, you know, I keep this big list of projects that I'm kind of considering doing. And usually what I do is as I see material or interact with people or whatever relevant to those projects, I'm sticking those, those papers or those notes or whatever in the folders for those projects. And I fully know that in a few years, when I'm finally ready to do that project, I open up that folder, all the stuff that I've accumulated is there. Then I start going through it. I don't put that much effort into figuring out how I put those notes and so on into those folders. I just throw them in there. And because I know that the way I'm going to be thinking about that project when I finally come to do it, it's probably going to be different than the way that I thought about it when I put those notes in there. And I shouldn't spend the time sort of cleaning those notes. I should just throw them in there. And then when I really get to that project, I'll probably have some conceptual framework which will allow me to think about those notes in a much clearer way. Suggestion for memes to call symbolic discourse language. Modern mimetic theory. Okay, I will think about that. Thank you. And let's see. I appreciate people saying the neural net discussion was. Was good. I thought it was good. I thought it was really interesting. The. But it's. Let's see. Okay. Zayden is asking, have you ever made a bad decision on a project? Do you own up to it? What do you do about it? Of course I've made bad decisions. I mean, my calculation is if I can hit, you know, 90% correct decisions, I'm doing well. And for me, one of the things that's important is make the decision quickly and move on and don't sort of agonize about, oh, look back, did I make the right decision? Did I not make the right decision? It's just in any given day, I'll make a bunch of decisions, do this, do that, do that. Some of them will be wrong. And I think the. You know, it's funny because when it's wrong, I just say I got it wrong. And sometimes I find it very useful. You know, I even talk about decisions we got wrong and I got wrong in the past as we're trying to make new decisions. It's like, remember that time five years ago when we did this? I made this mistake. And in fact, I was. I just was sending a piece of email to someone just before this meeting talking about a mistake that I made. Not really a project mistake, but I'll describe the mistake because perhaps it's interesting to people. It's a mistake I made in the 1980s, and it was the following mistake. So I worked on a bunch of different things. And when I moved from one thing I was working on from science to technology, business, et cetera, I just sort of dropped the previous thing I'd been doing and I just, you know, I've been working on science. I made certain progress, said, okay, I'm done with that. I'm going to move on, I'm going to do technology development. You know, let me, you know, I'm, I'm entering, turning over a new page and I'm just going to work on new things. I realized after that decade, I realized that was a mistake. I realized that I should have, I had developed a lot of relationships with people, a lot of kind of momentum in these areas and so on. I realized I should have kept, I should have maintained more connections with those sort of areas that I moved on from than I did. I mean, I've subsequently recovered those things, but it took a long time to recover them. And it was one of these things where the sort of cost of maintaining the relations, so to speak, is not high. But it's helpful both in terms of the, you know, one's own sort of return to those different areas as well as just sort of knowing what's going on. I suppose one of the main reasons I didn't do that is that it's like, well, I'm not going to actually work in that area in the foreseeable future. So finding out all the things that are going on there is not really that interesting to me because there's nothing for me to do with them. You know, what I tend to do now, for example, when I've written something about a particular area, I know I'm not going to write something about it again soon. But I also know that I might do something again on it in the future. And so when I have my folders, I usually have these pretty big folders actually of, well, it's subfolders and subfolders and so on. But for example, the project I worked on several last few months about thermodynamics, that was, oh, I don't know how many files, I'm going to guess, a thousand files of different kinds of documents and notes and notebooks that I was doing computations in and drafts of things I was writing and so on. And then I have an additional folder which I haven't standardized the name of as well as I should have done. But it's basically follow up. I mean, that's the usual name I use for it. It's just follow up, but that's not always quite the best name. And what that means is things that come in after I wrote this where I'm not really going to put a lot of energy into it right now because I moved on to the next thing, but I want to store them somewhere because I may come back to that area. And I also want to, as I say, I think one of the, so I'm describing my mistake from the 80s was just sort of move on completely. Just ignore, just, you know, stop reading anything, ignore everything, stop interacting with those people. That was a mistake. And, you know, spending a little bit of effort to kind of keep those, keep one's awareness of what's going on, keep the relationships with the people, organizations, whatever is worthwhile. And I've tried to do that in subsequent decades, so to speak. But in terms of a project, for me it's very important to make the decision and move on. And for me, as I've gotten older and more experienced, it's gotten faster to make decisions. And sometimes there's a question, okay, so when you're interacting with other people and you are sort of making decisions with them, I'm probably not up there in the top of the consensus decision making crowd, but what I like to be able to do is I have a lot of very talented people that I work with and my goal is to be able to get the best sort of decision support and advice from those people as possible. It might not be saying to them, go make this decision, go tell me what you think the decision should be. Because that may not be the optimal thing that they can do. The optimal thing may be they will think about some part of it. And if you ask them about that part, they'll give you a very good response. But if it's left only to them, they're only going to think about that part and they'll make the wrong decision because they're not paying attention to another piece of what's going on. So, you know, the challenge is, can you, can you extract from the people you're working with the parts that they have the greatest ability? And you know, sometimes I will just say, why don't you guys decide? You know, I think, you know, there comes a point where I realize they're going to make as good a decision or a better decision than I am, and why am I in this picture, so to speak. I think that the other thing that can happen is people are like, oh, oh, I think you're making the wrong decision, I think you're making the wrong decision, et cetera, et cetera, et cetera. And it's an interesting problem when you listen to that and when you don't. And sometimes there'll be people who are quite junior people, quite new people, people one has barely interacted with, and they'll say, you're making the wrong decision, you're making the wrong decision. And sometimes they're just completely talking nonsense and sometimes they're not. And you know, distinguishing those two is very interesting problem. I mean, I think it's always like, well, can you explain what you're saying? And you know, for me, I always like to believe that I actually understand things. And if somebody is like, you're making the wrong decision, let me just tell you, you're making the wrong decision. That doesn't really do it for me. If they can say, well, this is why and this is why, and so on. That's, that's, that's very useful, I have to say. It's a, I seem to have developed a bit of a habit over the decades of times when people are interact with me in various kinds of settings and a sort of, kind of almost heckling me about, about things and so on. And, and those end up being people that I end up hiring and they do really, really well. I don't know that that's a, it's always one of these. And of course there are other cases where it's like, oh gosh, I don't want to hear this, this is not interesting. But, but it is, I think it is an important thing for somebody like me at least to feel that, you know, I'm continually expect to be challenged by people. And the only thing is that, you know, you can have your time wasted by that by people who are like, oh, they keep on saying the same thing, they keep on saying the same thing. And, and you know, it really isn't that interesting. And, but you know, it's, it's great if one can surround oneself as I think I have, with people who can provide challenges where the challenges they provide usually make sense. And, and sometimes, you know, sometimes I'll agree with them, sometimes, you know, I'll convince them of, of that I'm right. But it's a, it's a useful dynamic, so to speak. Let's see. Okay. Time Lord. Good, good. Monica, for the, for the question being asked, estimating the time to build a feature is hard. Have you found any task estimation practice that works well? Yeah, you know, the thing for me is that I guess I've been doing things with technology, software, whatever, long enough that I usually have a sense of how hard things are going to be. And my usual procedure is, I have some idea, this is going to take a month, this is going to take two months, whatever. And I'll usually say that's what I think it's going to take. If it takes longer than that, tell me about it. And then what I usually find, or often find is if my estimate was way off, maybe there was just something I missed, maybe it just turned out to be harder. And person's going to tell me why it's harder. And I'm like, okay, you're right, I missed that. Well, sometimes it was. And this is a quite common thing that people are trying to solve a problem that they thought might have even been a problem that I presented. They're trying to solve it and it turns out that problem is really hard. It turns out they're trying to make some feature and to make that feature as advertised is really hard. But what went wrong was they didn't understand that the reason I, for example, asked for that feature is for some upstream purpose that we're trying to achieve in terms of users or whatever else. And they were just stuck on, you know, he said, implement this feature. Let's go implement this feature. Oh, that's really hard. And so when you come back and you check in and it turned out the time estimate that you had for that was, was, was off, you often find they were trying to implement this thing that was really hard. You say, wait a minute, we don't really want to implement that. Let's think about what we're really trying to achieve. There's a different path we can take. And oh gosh, that path is really easy. Great. I think that's an important reason to do that. Check in when your immediate estimate is wrong. So that's the thing I've tended to find is that people will say to me, you know, oh, that's a unbelievably over optimistic estimate. And I'll say, well, why? What are the steps that you have to go through to get to this point? And sometimes they'll point out a step and I'll say, okay, you're right, I'm wrong in that estimate. Now it's also important when working with people. There are people I work with who are always optimists as I am, and there are people who are kind of pessimists and there are people who, who I sometimes enjoy, who I think, you know, in the Winnie the Pooh classification of, of different people, they're the Eeyore crowd. And they're always saying this is wrong. This is wrong. This is never going to work. It's never going to work. And I'm thinking one person in particular who's a senior person at our company now who is just always in the. It's not going to work, you know, it's going to be impossible. And it's like, well, I say just, you know, what I learned is I say just give it a try, just see whether you can make progress. Comes back in a few days. Well, actually, I think we've been able to solve it. Blah, blah, blah, blah, blah. That was the kind of the Eeyore response which I've learned in the case of that particular individual means just tell them to go try it. Don't tell me all the reasons it isn't going to work. Just think about it yourself and try it. And they'll have success and they'll have, and everybody will feel good about it and we will have achieved something. Then there are other people who say kind of, yes, yes, I'll do it, I'll do it. You know, we'll, we're going to march in that direction. It's all going to work. And I'm like, wait a minute, do you really know what you're doing? Are you really going to be able to do it? And, and sometimes the answer is no. And sometimes it's like, you know, they're very diligently going to, going to do, but they don't really even understand what they're trying to do. It's. The assessment of people with respect to what's going to happen is very tricky. I remember one time years ago, a person who was doing computer system administration for me was, was, was going to leave because some friends of his were going to start this company in Silicon Valley. And I was a good guy and like, few days, but he was building this big, There was an early Linux system for me and he was like, you know, don't worry, I'll have it done before I go. And it's like, you know, and then it's, it's, you know, it comes to the actual time and it's like, here's the system. It's done before he leaves. And I'm like, this is not going to end well. This is going to be, you know, it's, it's going to be, there's no way that it's not going to have a tale of issues. It didn't have a tale of issues. Actually, I think it was up for a year or two or something. That particular individual went on to Have a. The company he went to was a company you all will have heard of that was very successful, and I think he was retired, but then he got unretired again, but has had a very successful career. But that was a case where somebody just says, don't worry, I'll take care of it, but I'm off. I'm out the door. And, you know, it's like, you usually think this is going to be a disaster, but I. I kind of. I kind of gave it even odds in that case because it was a good guy. And I thought maybe he's actually going to pull it off. And he did. But there are many other cases of. Of situations like that. I can think of another. Oh, boy. Another situation where as a long time ago, a woman who was very talented and. But who was working on this difficult project, and I kept on saying, are you sure it's going okay? You know, tell me about what, you know, what's happening? And it's like, don't, you know, just trust me, I'm going to do this. I'm going to do this. And then it came to the sort of moment when it was all supposed to come together and total, you know, meltdown, disaster. It didn't work in that particular case. Just a different person, different personality. And, you know, it's. It's a. I was. In that particular case, I was. I was deeply concerned that that was going to happen. I mean, I was. I was sort of 90% that's going to happen, 10%, it's going to turn out to be great. It was not. I mean, you know, it's. It's happened with. But this question about assessing, you know, should you have confidence when you're told this or that thing, should you have confidence? My general response to that is if somebody can explain themselves, then have confidence. If they can't, then be concerned, so to speak. That's a decent heuristic. It's not a perfect heuristic. There are people that I've worked with for a long time who are very effective in what they do, but just really terrible at explaining themselves. And it's kind of like, I think they communicate sometimes more with computers than they do with. With humans. And it's not even that they're. It's not even that if you met them, they would be socially awkward. It's just they're not very good at kind of explaining themselves in a way that other people understand. And so that heuristic of mine, of if they can't explain themselves, don't Believe it sometimes fails. But for me, that's been a pretty good heuristic to know whether you should have confidence in something. Zach asks, do you have unconditional confidence? I don't know in, in what? I mean, I think, I think it would be. I certainly don't have unconditional confidence in myself. I mean, I, I feel like I've, I've, you know, I'm always doing difficult projects and I always think that, you know, the projects I do, I suppose I, I arrange them maybe to be so difficult that I'm not sure I'm going to be able to do it. I think I'm going to be able to do it. I have enough confidence that I'm going to push forward. But is it going to be trivial? Is it going to be like, I have this one, I can do it. I know I don't really have to think about it. No. And by the way, whenever I get to that situation of this project is so easy, I can just, you know, stream through it. There's no problem at all. Those are the projects I mess up. I think in. I think that the. Somehow, at least for me, the fact that there's a project don't really know what to do. I really have to think. I have to really engage in thinking. That's what causes the project to be successful. And when it's a, you know, a cruise type thing, I'm not quite thinking enough. And, and it, and that. And when I don't quite think enough, I don't see all those issues that are coming in that are going to sink the project. Let's see. Junior is asking what allows us to be successful despite disbelief, discouragement, et cetera, attempting to solve problems in tech and science and other domains. You know, this question about confidence is an important one because an awful lot of things I do and have done, there's an awful lot of negative feedback from the world. Even people you work with and people you. I think what happens is I was lucky that I did some things in science and other places that go in pretty well when I was pretty young. And so that developed, I developed a certain amount of confidence. Yes, I can do things. I think at the very beginning when I was, I started, you know, doing physics research and writing physics papers when I was kind of a mid teenager. And at the beginning I was like, well, I think I can do this. But I would say that my early papers were not nearly ambitious enough. They were really things where I was just sort of incrementally adding to knowledge that already existed. It took me a while to get the confidence to really say I can do things where I'm sort of asking fundamental questions and I'm sort of building from the foundations and so on, rather than just I'm adding a little piece on the top. It was only after I'd had some success at the kind of traditional, adding some piece at the top that I kind of got the confidence to say, well, let me try something bigger. And I think that's been for me an incremental thing that you see things that you're successful at and then you can go further. I have noticed that when I work with people and we work on a project and there's a project where people at the beginning are like, this is impossible, it's never going to work. We're starting from nothing, we're never going to be able to figure it out, et cetera, et cetera, et cetera, and then by golly, the project works. It really is a very eye opening experience for people. And I think there's sort of a. I kind of have the theory that there's a sort of a 10 year glow for anybody who's been involved deeply in a project like that that for at least 10 years afterwards, they, when they come to a thing that's really difficult, they say, wait a minute, I saw this other case that was really difficult. You know, we can do this. And so I think it's useful for people to be successful at doing projects, to build confidence from being successful at doing projects and then build for the next project and so on. And maybe if that means you have to do some easier projects at the beginning to build that confidence, probably depends a lot on personality, how important that is, that might not be a bad thing. But you know, if you're doing difficult projects and you're doing projects that are going to be sort of not in the mainstream and have the potential to define a new direction in things. Yes, People will say we'll give you negative feedback and they'll give you negative feedback. Sometimes with the best of intentions, they'll say, you don't want to do that crazy thing that's off in that direction, just do the mainstream thing. Because you know that's going to work out decently well and you know it's not. And sometimes the intentions won't be so good, but that is. And sometimes it'll be like a variety of things. Whether, whether it's like, well, if you go in that direction, you're going into my turf. And you know, we don't want you in our turf type thing or whatever else it is, or just people, you know, you describe an idea and people just don't get it. They don't understand it. They don't imagine what it's going to be like. And they say, that's a stupid idea. And again, that might be said with the best of intentions, as in they really, you know, I'm sure I've made that mistake. You know, I try not to, but I'm sure I've made that mistake. When people describe ideas to me, I've said, you know, I don't really believe in that. I don't really understand it. As a matter of fact, I'll give you a good example of that. Let's talk about neural nets. You know, I thought for a long time, you know, this is an interesting case because back in 1980, neural nets had. Were having a kind of a resurgence of interest. And in fact, and I got interested in them. I knew a bunch of the people involved. I started looking at them and got interested. Didn't make any progress at the time. And then over the years that followed, I kept on seeing these, these papers that had these really complicated things in them, and it's like, why do we care? This is never going to work. You know, is it supposed to be like a brain? Is it supposed to be a piece of engineering? I can't. Well, it's sort of somewhere in between. It's kind of. I, I was not a believer, so to speak. And when people told me, you know, we're going to make something amazing out of this, I was like, really? You know, and I was like, you know, I think there are, there are different directions. And, you know, as it's turned out, I was wrong about that, that it is the case that when you build a big enough neural net, you can get it to do interesting things. It's, it's some, It's a case which I, you know, I'll add that to my inventory. I like to remember places where I got things wrong. You know, I'll mention another one that sort of interesting. It's like when you see a new technology coming along, you know, do you jump on it or not? And my one, you know, sometimes I'll, I'll jump on these things and say, this is cool. Like, I remember 1990 thinking, you know, I see virtual reality demonstrations. I think this is cool. I try and get people at the company to develop a bunch of VR technology around our Wolfram language technology stack and so on. We do a bunch of Things. And of course it was a bust at the time. So now maybe five years ago or something, I'm like, wait a minute, VR is back. We should be doing things. And some of the, because I have a bunch of longtime employees, some of those people had been there for the previous time and they're like, wait a minute, you told us this in 1992 or something and it was a total bust. Prove to us that it isn't going to be a bust today. And well, I wasn't completely able to do that. And we all know that this is still a somewhat up in the air question, but I'll tell you another example. This is probably 1991 time frame. People are talking about, a few people are talking about the web. And I'm like, you know, I've seen gopher, I've seen waste, I've seen these other sort of Internet based communication things. Okay, the World Wide Web, that's another one of these, you know, is it really better than Gopher? Is it really better than whatever? I don't really know. Is it really exciting or not? And I think I was, was at least three months that I kind of lost through not, you know, not being assuming because there'd been previous iterations that have been kind of less than, less than amazing that this would be a similar example. And of course it wasn't. Now, you know, once you realize that this is really going to work, then it's worth sort of jumping in and really trying to understand it. And I, I try to do that, but that was a, that was for me an interesting case because I really, I was probably three months behind, maybe more even a bunch of other people because I really had been sort of, I don't know, I got a vaccine or something against thinking about this by the fact that I'd seen previous iterations of the technology. This is a pretty common thing that you see a technology developing happen with neural nets as well. You see a technology developing, it's really not working out very well. And then suddenly for some reason or another, it takes off and one is often perhaps slower off the mark to see that. If one has watched the developments of the technology, unless one knows this effect and unless one has sort of cataloged the fact that this is a mistake I've made before. And so I, so I try to do that. The, it's, Let's see, look, in terms of this whole business about discouragement from other people. I mean, I think the, the question is at some level, you know, do you believe more in what they think or what you think. And somehow I, I sometimes feel a little bit bad because I sometimes feel it's pretty. It's sort of when I'm saying, look, I'm going to do this thing, I'm pretty sure it's going to work. And all these people are telling me it's not going to work, it's not going to work. It's sort of disrespectful of me to just completely ignore their opinions, but that's what I often end up doing. And you know, it's somehow it, it feels like, you know, everybody's telling me this, I should be listening to it. But actually I do believe in myself enough to say, you know, that, that I'm going to ignore that. Now. Now sometimes I can say to myself, from sort of my own experience, look, this person, you know, this person is telling me this because they made that mistake for this particular reason. I have to say one feature of. And it's terrible that I say this because I'm off spouting, giving you all advice about things, but I have to say one of the things about advice that's always complicated is that the person giving the advice, they've had a certain life experience, they are giving advice to somebody with different life experience, different characteristics and so on. And the, the thing that has, you know, I, I have seen it many, many times of people giving advice to me and so on. They say, oh gosh, you should set up your company this way. And it's like. And I know they've set up a company that's been very successful that was set up that way. And it's like that worked for them. And they'll say, you should do it this way. Well, it wouldn't work if I tried to do it that way. I'm pretty sure it wouldn't work for me because I'm different from them. And it's hard when one gives advice to people to sort of take oneself out of the picture, so to speak. I mean, given that I know this effect, I try pretty hard to do that and to not sort of say, well, you know, sometimes I'll say, this is what I would do in this situation. But, you know, you have different characteristics. Let me see if I understand your characteristics and can say, you know, how this would work for you, so to speak. And I think that when one does try to give advice, I mean, I think that the step one is try to understand the person you're giving the advice to rather than take the thing which you know from your own experience and just try and sort of foist it on that person. And, you know, so I think that's, that's a. But, but that's something where it's the same thing with people telling you, oh, your project is not going to work. They have a model for projects that work. Let me give you an example. I mean, there are, there are projects. So I'm thinking one particular where if I'd been doing this project, it would have failed because this project, to succeed needed a bunch of kind of basically endorsement, marketing, kinds of things that are completely not what I am into, know how to do, et cetera, et cetera, et cetera. So even though the technology, you know, I'm perfectly, you know, I understood that technology isn't going to go someplace unless it can get some kind of sort of public endorsement like marketing around it. And so for me, the advice would have been, if I was giving the advice to myself, it would have been, that's not the right direction to go in because there's a piece which won't allow you to get there from here. But to the people who are doing this, they managed to make it work because they managed to do something which was completely outside of my kind of domain of expertise to execute. I mean, and I couldn't really assess whether that was something they were going to be able to execute. I mean, that's another thing that I find is that in domains where. So a common thing when you're hiring people, for example, there are certain kinds of positions. We've hired people for these positions a zillion times. And I can tell very quickly, is this going to be a fit? Is this not going to be a fit? But then there are positions where we've never had one like that before. It's really hard. And usually at the beginning, I'll get it wrong 50% of the time. And because you just don't quite know what you're looking for. And it's kind of a. And I think. Well, in any case, I would say that this thing about kind of other people giving you feedback, it's always very hard to disentangle. I will have to say, about the story of feedback. I. I have to admit that I do collect sometimes some of the more outlandish things people say to me about me, so to speak. And one of the things I've noticed about them is about the sort of outlandish things is that they are almost always vastly more about the person saying it than they are about you. It'll be some, you know, you're terrible. Because. And the because will be something about them. Not really. It's something which they have in mind. They're thinking about it and they project it onto you. And it's kind of a same as the, as the advice thing. As. Same as the. The kind of, you know, encouragement, discouragement kind of thing. So, you know, I think that. That now, you know, there are situations. How do you know if you're deluding yourself? How do you know if you think something but it's just nonsense? Well, for me, the best way that I have to answer that is I try and explain it to people. I even try. I try to maybe I try and write it down. You know, I'm sure there are things where I've realized as I'm talking to you all in these live streams, I've realized this is the thing I've often believed, but it's actually nonsense. And now that I'm explaining it, I realize that it's nonsense. And so, you know, I think one of these things that gets bad is when you have this thing that is like, it's the, you know, the crown jewels. It's the thing that is really the center of what one's doing. And I never talk about that thing. I'm talking about everything around it. It's like we have this amazing sort of magic thing and. Or this magic idea or this magic way of thinking about things. And I don't talk about that. I only talk about the wrapping around it. That's usually a mistake because quite often that thing in the middle that you thought was the amazing thing that you've never actually bothered to talk about or really sort of clearly think about, turns out that's where there's a problem. But you thought it was such a, you know, a core important thing. But by the way, you see this in fields of science all the time that people have foundational assumptions and they are like, everybody knows it. It's part of the accepted wisdom. And, you know, those are the places if you want to go attack some field of science or make progress in a field science that. That is sort of outsized progress. Going to find those foundational pieces that nobody bothered to think about is a really good strategy because those are places people say, oh, yeah, that was solved 100 years ago. We don't have to think about that anymore. Well, actually, turns out that piece isn't really right. And, you know, it's. It's unprotected. It's kind of like you, you know, nobody's really. Nobody's doing things with it, you can go in and you can potentially make progress. And I think that it's the same kind of thing in other areas. Okay, the question here, okay, memes are saying ideas can be great. Implementation will define how it goes over. I absolutely agree. There are raw ideas that are really valuable and good. And once you see that idea, it's kind of like, yes, we can go in that direction, but actually executing and implementing really hard, a lot of work. Lots of. And I suppose for me personally, I sometimes minimize that because I suppose I, you know, I've. I've been involved in implementing and executing things for a long time, and I've worked with a lot of very capable people doing those things. And so for me, it's kind of like inject the idea and then this machine that, you know, I've been involved in building sort of takes over to execute it. And, and sometimes, you know, you, you realize that's actually, it's actually harder than. I minimize the difficulty of that. So, yes, it's, it's a. And, and that is, you know, when I see startup companies and so on, and it's like, well, that's a very good idea. I say, and then the people fail, and they fail through sort of bad execution, and sometimes they fail. It's, it's like whenever you're doing something like a startup company, there are zillion decisions that have to be made, and some of them you can get wrong, but if you get enough of them wrong, it's just not going to work. And I think people sometimes, yeah, I mean, it's a huge challenge to execute things. And by the way, one usually gets. I mean, for example, I have a certain way of executing stuff that is kind of baked into the structure of the company I built. And if you said to me, you've got to execute the thing you're doing, but in a different way, I'd probably fail at that. And you see this a lot when companies try and work with other companies and there's two companies and they have rather different kind of ways of executing things, and they're trying to glue those together. It often doesn't work because it's kind of like, well, for one person, it's kind of like, well, we assume this is sort of, you know, it's, it's a free ride from this point because we know how to get that done, you know, and I don't know what it would be in some companies. It might be, you know, you've got a good, you know, graphic design you know, whatever department, it's like you're going to have this thing, it's going to look nice, whatever happens. But for other. And you don't really have to think about that, so to speak. But for another company, it's like that's a big effort. You got to hire a design agency and it's a whole complicated thing and you mess up on that because it's, you know, one side can be just assuming, oh, that's easier or something like software quality assurance, you might say, you know, a company like ours has a very well developed effort in that area and it's kind of a standard thing. You're making some pieces of software, we're going to run it through a whole software quality assurance process. But you know, to somebody else it's like really, you know that, you know, we'd assume that people had that probably, or we might assume that and if it turns out it wasn't true, you end up with something very bad happening. So yes, Mark suggests you should find a producer to create a dramatic retelling the history of the second law. That would be fun. Yeah, that would be fun. You know, I think it's somehow, you know, I've been involved in a few movie like things and so on. For some reason that's a domain in which execution is, seems to be just an incredible challenge. I mean there are so many, you know, here's an idea for a script, here's a, you know, here is a script, here's a whatever. And it's like the, the, it seems to be an industry where the distance between the idea and the actually it's a movie, it got made is, is just unbelievably long. And I know again, speaking about the sort of the process of channels for execution, there are people, producers, things like that who just have a, you know, they've got a scheme for doing it. And if you say there's another, another thing that's always interesting to realize is when, when you have some kind of scheme for getting stuff done and then you say, well, I've got this new thing and it doesn't quite fit that scheme, but I'm going to try and run it through the same organization, it's usually a fail because the, the, you know, to, to sort of, you know, if there's some organization that works in a particular way, you know, it, you know, I don't know, you're publishing a book. Publishing companies I don't know about these days because I haven't used them in a while. But it Always used to be the case that publishing was an unbelievably sort of structured, regimented business where, you know, there'd be a contract for a book and the contract would be, you know, printed. Now there's a little bit of a game being played there about like, oh, you can't change the contract, it's printed. Well, actually, I remember one time with a big publishing company, we were working with them, deal didn't end up working in the end, but, but basically took this contract and we crossed out the whole contract, just every page, just like, nope, nope, nope, and started again. And they weren't as freaked out as, as, as they might have been. It was kind of like, well, we tried it on, you know, just giving you a printed contract, but wasn't going to work. But, but the publishing industry, at least back in Those times now, 25 years ago or something now was very, very regimented. As in, you're making a book, we're going to pay you X amount of money as an advance on royalties. You're going to get X percent on the sales of the book. We're going to take a formula where you can take X number of dollars that was spent on the advance and that X number of dollars is equal to, numerically equal to the number of dollars we're going to spend on marketing the book, which is numerically equal to the amount it's going to cost to print the book. I think there's a small factor there actually, but there's a well defined formula and if you say, well, I've got a book and this happened to me because I made a book that didn't fit those parameters, you'd say, I've got this book, it's going to be successful. Here's the evidence, it's going to be successful. But it doesn't fit those parameters. That industry at the time at least was completely incapable of working with that and turned out. My big book, New Kind of Science, which we ended up running through our own publishing company, was extremely, as a pure publishing venture, independent of its intellectual characteristics, was an extremely successful publishing story which, you know, the, the Publishers Weekly magazine, you know, picked up about. Yes, these people have succeeded even though they didn't go through the traditional publishers and so on. But you know, it was a case where you just couldn't, you know, round peg, square hole, whatever, it just wouldn't fit. And often you find situations where people say, but it's obviously a good idea, you know, you should be doing this, this is a good thing to do. But it just won't fit. And you know, we see that a bunch in partnering with other companies, companies partnering with us and so on. There's a natural way that a company works and if you align with that, fantastic, it's going to work out. If you don't align, it's one of these things where, yes, it may be a better idea, but. But you're probably going to have to do it yourself because you're just not going to be able to fit it through that channel, so to speak. Let's see, Tori is asking what strategies can be used to break down research projects into tasks that can be distributed to collaborators in order to leverage being part of a group that is challenging. You know, I think it's again, it's one of these things where you got to know the people you're working with. The better you understand the people you're working with, the better you can give them pieces that they'll actually be able to deal with. And when you give them piece they can't deal with, it just never ends well. It's always, they take it on, they don't take it on. You think they're going to do it, they don't do it. The whole project goes off track because it's because you're relying on them, but they couldn't get it done, et cetera, et cetera, et cetera. Again, it's sort of an art to assess the people you're working with to get to know them well enough to try to figure out what their strengths are. And. That's an important thing. I realize, as I'm talking to you guys, one exercise that I have for that is we've been doing this annual summer school for 20 years now and also a summer camp for high school students for a decade or so now. And every year there'd be, I don't know, total of maybe 150 people or something. And you know, one of my jobs is to come up with original projects for all those people to do. And it's a very interesting problem because you have to make some assessment of these people and try and give them a project that's actually going to be a fit that can actually be done in two or three weeks. And you know, it's like I've now seen the results of that thousands of times of, you know, did you want to successfully assess the person? Did one give them a project of the right size? Was it a project that had the right kind of on ramp or not? And as I think about it, that for me, has been very educational in understanding this problem. I mean, it's also a thing that I see a lot at the company. Although the characteristics of projects are a little different than these kind of research project types of things. So I can't claim I'm always successful. I think I have a very good sort of batting average for the summer school and summer camp. But it's something which is definitely an art to be able to figure out how do you take that piece? And, you know, and sometimes what you'll realize is there isn't somebody I can delegate this to. Sorry. It might be nice, but it's just not delegatable. The actual team we have doesn't have anybody in it that is capable of taking this particular thing. I'm going to have to do it myself. If I know how to. How to do it, I better just do it. I think that in, you know, some things that I've seen tried and I don't know how well they work is. Is one is you've got three or four people and they work with each other as well as working with whoever's leading it. That. That can sometimes work. I. I don't know. I. I think I know my children at various times have had, you know, projects they do with other. They're now grown up, but they were in the past. They would have projects they would do, you know, with other kids. And, And I'm always like, I'm always saying to them things like, you'll end up doing the project yourself, so to speak. And they're like, no, no, you're way too cynical. No, no, no, no, no. And then they're like, oh my gosh, this is terrible. These people. I have to do twice as much work to even drag them along to get a piece done. Now, truth is, in many of those cases, the project that was picked was one that wasn't particularly suitable for the whole group. It was suitable often, I will admit, for, you know, a kid of mine picked a project that was good for them, so to speak, and kind of led the group in that direction. And even though the group, you know, might have, you know, there might have been a project that was more. More suitable for the group as a whole. But. But, you know, I think that's the. If it's your project you're kind of saddled with. If it's a project you really want to get done, it's your project. You're the one who has to fill in all the pieces that cannot be delegated to the people that you have found to work on that project. If you're in a different situation where. Well, I don't know. I'm probably not the expert on being delegated to. Even if I, I do make the complaint with some regularity about things at our company. The, the reverse delegation phenomenon, which is, you know, like, like we're doing something and it's like, well, the solution is that I, you know, do a bunch of work and it's like we have a strategy and then at the end of the strategy it's, well, then I do a bunch of work and I'm always. It's, you know, sometimes it's the right answer, but I'm always, I'll always make. Often make snarky comments about reverse delegation when that's what happens. All right, maybe one more question here and then comments here and then I should wrap up. See. Well, Phil is asking here, do you think a career working on AI tools for use in science and mathematics is likely to have a big positive impact? Depends what you mean by AI. I mean, at some level, the things I've been working on for most of my sort of life can be thought of as quotes. AI tools for science and mathematics. If by AI you mean things like neural nets, I think there are pieces that are sort of more user interface type pieces where that's going to be very valuable. I think there are. It's a complicated story where, okay, so, you know, neural nets kind of are a way of sort of. They're. They're a model of the world that seems to be a good match with the model that we humans make of the world. And they seem to be a good match with kind of sort of smoothing things over in a nice way. If, you know, they don't necessarily expect the unexpected, but when things are kind of smooth, they'll smooth it over in a nice way. You know, you have that picture of the, you know, you know, the alligator with a lamp for its tail or something. And you'll be able to get an image generation system to smooth over the alligator and the lamp for its tail, and it'll sort of make a decent seam between those two. And there are plenty of cases where when you're doing protein folding, let's say you can get something where each individual piece, oh, yeah, it looks just like this protein from here or there, but you've got to smooth those pieces, pieces together, fit them together, and that's a place where neural nets can be really useful. And I think that similarly, in some sense, there are places where that smoothing could be done in many different ways, but there are ways that are very human ways to do it. And that's, I think what's happening in part with these large language models is that we're sort of putting pieces together in ways that are very natural, seem natural to us humans partly because we, us humans have neural nets inside our brains. And so we're kind of used to that way of things working. Now if you say, okay, I'm going to give you a math problem to solve and get the neural net to solve the math problem, it's just not the nature of what it does. I mean particularly not the kinds of neural nets that have been trained as they are for something like ChatGPT where they're feed forward neural nets basically. And it's just like you've got this complicated network with, you know, 175 billion weights or whatever it is and you're at every step you're feeding something in, it's rippling through, you're getting an answer, maybe you're feeding it, once you've got an answer, you're feeding it back and going on. But fundamentally it's not, it's something where you're just, it's like a giant, not quite a lookup table, but it's kind of a segmented version of a lookup table and a function evaluation kind of thing. And that's not the nature of what we have learned to do in more sophisticated computation that involves going back and reusing pieces of data that we've used before and so on. It's a different kind of computation. And I think so if you say, are we going to be able to do these things in a way that's just trained to do it, training seems to be quite incompatible with the kind of irreducible computation that's involved in doing sort of deeper science, math type computations. So I think that's, I'm not optimistic about that direction, but there are pieces even of that where it's like a very typical example is something where there's a search going on. Let's say you've got a algebraic expression, you're searching for the simplest form of that algebraic expression, or you've got a variety of things like that where it's a question of searching for an answer. Well, is a neural net going to be useful as a way to say try going this way rather than trying to go that way? Maybe. I mean we've done experiments on this for years without that much success. I think the place where it's going to be Useful is where the direction you want to go is informed by what humans would do. If it's informed only by what the sort of the pure, pure computation, pure science, pure math would do, I think it's less likely to be useful. But if it's basically you want a thing that will automatically be an automatic proxy for that's the thing a human might do, that's the decision a human might take, then it's a reasonable thing to do. And in terms of what that means for kind of, well, that's one direction. There's a whole other direction which is can you. I mean, we've done experiments for more than a decade on programming with natural language. I mean, I even had this post that I wrote in 2011 called Programming with natural Language is actually going to work, which was based on a bunch of examples we had at the time. Turned out that the getting it to really work is still, still in the future. But we could see that it was, it was. There was a direction where it was, it was going to be possible to make it work. I think that will be interesting. I don't think it's as important for the sort of the deep, you know, frontline research folk, because those folk, I would like to believe would learn tools like ours as a way to make progress rather than just saying, well, I want to be, you know, talking to my AI to get it to figure out my science project. Because I think that I like to believe that with computational language and so on, we have a really good notation for people to take the thoughts they're having, commit them to something which can be made computational and then really zoom forward with it. So I mean, it's complicated this, you know, the, the. I think there are particular cases where sort of the things we've learned from neural nets and so on, the cases where it's. It's sort of be the proxy human. There are cases where it's kind of be a better way of filling in and smoothing things than we've had before. And that's the case maybe with protein folding. It's probably the case with a bunch of things in computational chemistry. It's probably the case a bunch of other things where it's kind of like, can you find a way to. It's not quite a search problem, but can you find a way to sort of smooth these pieces together better than we'd be able to do before? And I think a bunch of the methods that have been invented for neural nets may be very useful for that. It may be kind of a You know, just as the neural net revolution of 10 years ago or so now was enabled by GPUs that have been invented for things that weren't for doing computer graphics and things like that, they were not invented for doing neural nets, but they could be used for that. So similarly there are things that are being invented for doing large language model, you know, linguistic stuff that can be used for, for other purposes. I mean that's even the, the sort of, the, the way large language models. The fact that large language models can be used for protein folding is pretty interesting. It probably is telling one something both about proteins and about language. I was actually just exchanged piece of mail with a friend of mine who's a sort of leader in the protein folding world in which was talking about why things like ChatGPT work. And the fact that I think what it's really showing us is that there are these aspects of sort of human language, these semantic building blocks of human language which we didn't quite know were there, of which one is kind of the stuff I was mentioning earlier with Aristotle and the logic as a building block of linguistic meaning and so on. And so I was making the quip that in proteins there are these different sections of proteins which are pretty well understood, of which the two most obvious are alpha helices and beta sheets. Alpha helices I think were, were discovered by Linus Pauling. And so I was making some quip about how Linus Pauling was sort of the Aristotle of the protein folding world. Just as Aristotle identified in human language these kinds of building block structures that turned into syllogistic logic. So Linus Pauling identified in proteins these kind of building block structures like alpha helices and so on that were sort of structural structured features of proteins. All right, I should wrap up here, but thank you for joining me and thanks for asking lots of questions that make me think about how I should do things and I hope this is useful to you guys and it's, it's useful to me. So thanks and look forward to talking to you all again another time. All right, bye for now.