WEBVTT 00:00.031 --> 00:12.446 [JM]: Today's bit of meaningless fun is Dog Spinner, a site where you can turn a crankshaft and spin a dog because that is something that you really need to do in today's troubled world. 00:13.127 --> 00:14.408 [JM]: Have a little bit of fun on the web. 00:14.909 --> 00:19.374 [JM]: This is a fun place that you can go and spin a dog around and make it chase his tail. 00:19.454 --> 00:22.197 [JM]: And as usual, the link will be in the show notes. 00:22.418 --> 00:24.280 [JM]: I highly recommend that you go and check it out. 00:24.360 --> 00:25.721 [JM]: It's a fun little project. 00:25.862 --> 00:27.664 [JM]: I don't know who did this, but hats off. 00:27.744 --> 00:28.124 [JM]: Bravo. 00:28.325 --> 00:28.705 [JM]: I dig it. 00:28.685 --> 00:35.482 [DJ]: I just want to double check has the dog consented to this. Like, when you turn the crank and the dog spins around, is the dog having a good time? 00:35.685 --> 00:48.482 [JM]: Like most good fun things like this, if you spin the dog around fast enough and long enough, I do believe you get some feedback that sounds like it might be having a good time. It's up to you to decide. 00:49.255 --> 00:49.736 [DJ]: Okay. 00:50.036 --> 00:50.757 [DJ]: Dog spinner. 00:50.817 --> 00:51.077 [DJ]: Sweet. 00:51.137 --> 00:54.741 [DJ]: I'm looking forward to next show where we find, I don't know what, cat spinner. 00:54.801 --> 00:56.763 [DJ]: There was already the fish thing, right? 00:56.803 --> 00:57.404 [DJ]: Draw a fish. 00:57.885 --> 00:59.226 [DJ]: And now there's spin a dog. 00:59.587 --> 01:05.393 [DJ]: So we need another one, which will be, I'd say confuse a cat, but I'm pretty sure that's a comedy bit from like 30 years ago. 01:05.433 --> 01:09.277 [DJ]: So we have to do something else to some other animal. 01:10.799 --> 01:11.680 [DJ]: Employ a parrot. 01:11.880 --> 01:12.341 [DJ]: How about that? 01:12.761 --> 01:14.643 [DJ]: I'll vibe code that for the next show. 01:14.664 --> 01:16.005 [JM]: Can't wait. 01:15.985 --> 01:21.632 [JM]: Another link I wanted to direct everyone to is much less fun, but perhaps slightly more useful. 01:21.792 --> 01:33.407 [JM]: And that is the "Opt Out October": daily tips to protect your privacy and security resource from the Electronic Frontier Foundation, or EFF. 01:33.808 --> 01:39.655 [JM]: It looks like there will be a total of 22 tips, 13 of which as of this recording are already live. 01:40.096 --> 01:45.503 [JM]: And this is a great resource for establishing good digital hygiene, 01:45.483 --> 01:49.930 [JM]: for opting out of a bunch of things you probably don't want to be opted into. 01:49.970 --> 01:54.999 [JM]: It's just a great resource for protecting your privacy and security. 01:55.019 --> 01:57.623 [JM]: And I highly recommend checking it out. 01:58.024 --> 02:08.301 [JM]: I imagine that there will be some things in here that I will learn I try to stay on top of all things privacy and security related, but 02:08.281 --> 02:11.792 [JM]: I think it's impossible for anyone to be fully on top of all these things. 02:11.812 --> 02:20.097 [JM]: So I'm glad that they are putting this information in here and giving people a way to find out about ways that they can protect themselves and their loved ones. 02:20.363 --> 02:42.979 [DJ]: Yeah, I'm looking forward to going through those because I tend to be pretty diligent with my online privacy and security, but I probably also get a little lax over time, and then it's good to have a reminder to go back and maybe clean up, if I can, some of the places where I don't need to have that company know my email address or etc. 02:43.280 --> 02:45.563 [JM]: Indeed, that link will also be in the show notes. 02:45.683 --> 02:48.727 [JM]: So have a look and share with the people you care about. 02:49.168 --> 03:02.204 [JM]: All right, following up on the topic of managers handing generative software tools to their employees and telling them here, use this to make your job better, faster, more productive, etc. 03:02.525 --> 03:08.072 [JM]: And also following up on last week's discussion of the concept of how these tools 03:08.052 --> 03:12.319 [JM]: may or may not have considerable productivity improvements. 03:12.800 --> 03:28.187 [JM]: An executive at Facebook who is in charge of building the company's metaverse products has told employees that they should be using generative software tools to go five times faster, not 5%. 03:28.167 --> 03:38.322 [JM]: With the idea being that these programmers should be using these tools to work five times more efficiently than they are currently working, not just using it to go 5% more efficiently. 03:38.362 --> 03:46.574 [JM]: It is really hard for me to imagine anyone using any tool — it doesn't matter what the tool is — and having it improve their productivity by 500%. 03:46.594 --> 03:49.939 [JM]: I just don't see how that's possible. 03:49.919 --> 04:00.513 [JM]: But if you asked me, "Hey, I just heard that some company told its employees that we have an expectation that you should use generative software tools to improve your productivity by 500%." 04:00.573 --> 04:05.820 [JM]: And you asked me, which company said that, there is no doubt that my answer would be Facebook. 04:07.341 --> 04:08.163 [DJ]: Right, right. 04:08.183 --> 04:10.105 [DJ]: You're saying this is very on-brand. 04:10.085 --> 04:14.230 [JM]: The most on-brand of on-brandedness from Facebook, for sure. 04:14.290 --> 04:23.803 [DJ]: This struck me as one of these statements that is predictable, but also obnoxious, in part because it seems like rhetoric to me. 04:24.163 --> 04:37.020 [DJ]: I strongly doubt whether Facebook is going to use whatever quantitative metrics of developer performance that they have and fire everyone where those metrics don't increase by 04:37.000 --> 05:04.476 [DJ]: five times, because I think it's extremely unlikely... No, that's not true — I think it's completely impossible to achieve that outcome. So I don't think they actually want that outcome. I think they're making this statement for for the purpose of making a statement, right? "We're on the cutting edge of technology, and we really believe that this particular one is going to massively improve productivity." But like what are they actually doing inside the building, I wonder. 05:04.456 --> 05:11.666 [DJ]: Once it is implemented, I would love to hear from Facebook engineers, not that I know any, about like what is it actually like? 05:12.147 --> 05:30.632 [DJ]: And maybe it is the worst case scenario that we've sort of discussed in the past where like management is kind of blindly assuming that they can say, "Okay, we expect you to hit this productivity metric that's totally untested and unmoored from reality and there will be terrible consequences if you don't." 05:30.672 --> 05:33.957 [DJ]: And now everyone is overworked and scared for their jobs and like... 05:33.937 --> 05:44.029 [DJ]: Again, you and I both tend to have a fairly low opinion of Facebook and what their working culture seems like it's probably like. 05:44.329 --> 05:54.501 [DJ]: And so I guess it wouldn't surprise me to learn that it turns into this crazy code sweatshop of like, "Why aren't you getting five times as much work done?" 05:55.082 --> 06:02.931 [DJ]: But I strongly doubt that the reality will be anything really like what's being stated here, in part because... 06:02.911 --> 06:25.813 [DJ]: Well, yeah, frankly, it just seems unrealistic and you can't merely declare that like everyone's got to go five times faster. What does that mean? And, no, like that's impossible. So if you work a five-day work week, however much work you would get done in one entire week, you must now do that in one day. And maybe Facebook knows something I don't. They do have a lot more 06:25.793 --> 06:35.806 [DJ]: experience with cutting-edge technology and a lot more money, but my personal experience using generative software to help write software has been, eh... 06:37.107 --> 06:42.154 [DJ]: Sometimes it does feel like a force multiplier, especially for certain types of tasks. 06:42.194 --> 06:53.708 [DJ]: Like certainly I can add test coverage to code much more quickly than I could by hand, which is nice, but could I get the amount of software development I do in a week done in a day? 06:53.688 --> 06:54.851 [DJ]: No. 06:55.332 --> 07:00.686 [DJ]: And as you said, like, I can't really imagine any conceivable tool that would allow me to do that. 07:01.147 --> 07:02.391 [DJ]: It just sort of doesn't make sense. 07:02.431 --> 07:06.381 [DJ]: Like it's not aligned with the reality of what it's like to build software. 07:06.631 --> 07:29.584 [JM]: It sounds like the very typical hyperbole that we often hear from people in the tech world, where someone pushes this impossibly high target with the idea of like, "Okay, well, if we get everyone shooting for up here, okay, we may only get one-fifth of the way to that target, but that's further than we would have gone if we hadn't set the target so high to begin with." 07:29.564 --> 07:34.293 [JM]: Sometimes I feel like that's what this kind of language is supposed to inspire. 07:34.353 --> 07:38.901 [JM]: It's supposed to just get people to shoot for the stars so that they go further than they otherwise would. 07:39.463 --> 07:40.965 [JM]: And I don't know, I don't buy it. 07:41.326 --> 07:43.991 [JM]: I just don't think that in this particular case, it's going to have that effect. 07:44.432 --> 07:45.253 [DJ]: No, agreed. 07:45.293 --> 07:53.108 [DJ]: Not in this case, but I am actually quite a proponent of that as an internal management philosophy, but (a)... 07:53.088 --> 07:57.617 [DJ]: I wouldn't say it out loud. Like I wouldn't say it in public that "We are definitely going to do this." 07:58.138 --> 08:02.205 [DJ]: And (b), when I've done this in the past as a manager, I make it clear that's what I'm doing. 08:02.546 --> 08:10.641 [DJ]: The article we were looking at here just kind of makes it seem like a diktat, like "Thou shalt go 5x faster or else by using these tools." 08:10.621 --> 08:27.849 [DJ]: Whereas the, let's say "healthy" way to set an ambitious target, not in the hopes that you will definitely one hundred percent reach that target, but that it will force you to go a little further than you would have otherwise is that you kind of make it clear that that's what you're doing. 08:28.010 --> 08:34.580 [DJ]: Like this thing is, and people use phrases like "north star" or direction or, you know, guidepost or something like that. 08:34.701 --> 08:34.941 [DJ]: Like... 08:34.921 --> 08:41.390 [DJ]: This is a thing we're trying to reach because the process of trying to reach it will have a bunch of positive outcomes. 08:41.670 --> 08:48.559 [DJ]: So for example, you could say we're trying to reach a point where everyone is actually like more productive by a multiple. 08:49.080 --> 09:01.177 [DJ]: It may not actually be possible for us all to be five times more productive, but if in general our software development practices end up being 20% more productive than they were, that's still a win. 09:01.237 --> 09:03.780 [DJ]: And maybe that's the more realistic outcome. 09:03.760 --> 09:14.737 [DJ]: The thing is, you end up with, especially if you're a company as high profile as Facebook, you end up with news articles with the headline, "Facebook tells workers to use AI to go 5x faster." 09:14.777 --> 09:15.778 [JM]: Absolutely. 09:15.818 --> 09:30.561 [JM]: Following up on last week's discussion of investment in generative software tools and its implications for the broader economy, I wanted to touch on a few other points that came up since we recorded that. 09:30.541 --> 09:38.996 [JM]: One question that's on a lot of people's minds right now is, "Why is the United States economy still holding up?" 09:39.016 --> 09:40.719 [JM]: Like what is keeping it going? 09:41.080 --> 09:46.970 [JM]: Because the manufacturing industry is not doing particularly well because of punitive tariffs. 09:47.491 --> 09:49.995 [JM]: Payroll numbers are not particularly strong. 09:50.677 --> 09:55.485 [JM]: And consumer sentiment is at levels not seen since the Great Recession. 09:55.465 --> 10:05.138 [JM]: And yet, despite these warning signs, there's not even anything remotely resembling an economic implosion yet. 10:05.479 --> 10:15.392 [JM]: But if you look at some of the numbers, investment in information processing equipment and software is about 4% of GDP right now. 10:15.793 --> 10:21.901 [JM]: But it was responsible for 92% of GDP growth in the first half of this year. 10:22.442 --> 10:23.143 [DJ]: Whoa. 10:23.123 --> 10:26.671 [JM]: Yeah, GDP excluding these categories, 10:26.971 --> 10:35.209 [JM]: so excluding investment information processing equipment and software, grew at 0.1% in this year so far. 10:35.493 --> 10:37.777 [DJ]: I don't know, Justin, that's pretty close to zero. 10:38.458 --> 10:46.493 [DJ]: So there's this one sector essentially producing all economic activity or economic growth anyway, and not to rush ahead. 10:46.553 --> 10:52.764 [DJ]: And then if you look at that sector, I mean, it's basically like just a couple of companies that are experiencing all of that, right? 10:52.804 --> 10:59.256 [DJ]: Like it's not like there's hundreds and hundreds of firms that are contributing to this disproportionate amount of growth. 10:59.523 --> 11:00.004 [JM]: That's right. 11:00.044 --> 11:05.494 [JM]: This sector is essentially responsible for all of the GDP growth this year. 11:05.514 --> 11:10.623 [JM]: And that sector is comprised of a handful of companies, as you accurately point out. 11:11.144 --> 11:18.156 [DJ]: I feel like most financial advisors advise against this level of concentration in one's portfolio. 11:18.187 --> 11:29.698 [JM]: Yeah, and if you look at the economy as a portfolio, it's real hard to diversify your risk if all of your hopes and dreams are pinned on a handful of companies. 11:30.119 --> 11:36.625 [JM]: To the point where The Economist writes, "Look beyond AI and much of the economy appears sluggish. 11:36.985 --> 11:39.508 [JM]: Real consumption has flatlined since December. 11:39.948 --> 11:41.330 [JM]: Jobs growth is weak. 11:41.350 --> 11:47.276 [JM]: House building has slumped, as has business investment in non-AI parts of the economy." 11:47.717 --> 11:58.652 [JM]: As far as I know, companies that are heavily invested in this generative software sector have accounted for 80% of the gains in US stocks so far in 2025. 11:59.153 --> 12:07.465 [JM]: And in fact, more than a fifth of the entire S&P 500 market cap is now just three companies, Nvidia, Microsoft, and Apple. 12:07.445 --> 12:12.372 [JM]: And two of those are basically huge bets on generative software. 12:12.752 --> 12:24.468 [JM]: So in summary, there is a lot riding on the question of whether a generative software bust will crater the economy. 12:24.949 --> 12:26.010 [JM]: And I guess we'll find out. 12:25.990 --> 12:38.164 [DJ]: Aside from these economic reasons that we need GDP growth and this is the only place it's happening, so we have to come up with ways that AI software companies are the wave of the future because they have to be — because nothing else is happening. 12:38.545 --> 12:39.165 [DJ]: There's that. 12:39.225 --> 12:51.019 [DJ]: But then there does seem to be this narrative of the reason to keep pouring money into companies that are making glorified autocomplete, to call back to other discussions of this topic, 12:50.999 --> 13:01.651 [DJ]: is because at some point, somehow, what is currently like large language model powered software turns into something that utterly alters the fate of the entire human race forever. 13:01.691 --> 13:16.448 [DJ]: There's this weird idea, it's almost as though everything else is on hold while we hope that our AI savior is soon to arrive, which is, I'm not saying that's actually the case, but I'm saying the science fiction short story writes itself. 13:16.768 --> 13:17.649 [DJ]: Well, it doesn't write itself. 13:17.669 --> 13:20.432 [DJ]: You use an LLM to generate it, but you know what I mean. 13:20.783 --> 13:26.197 [JM]: What could we be doing with all the money that's currently being invested in generative software? 13:26.217 --> 13:28.362 [JM]: I think it's an interesting question to ask, right? 13:28.783 --> 13:37.645 [JM]: If we took the trillions of dollars that this industry says they need and that are already being raised and invested and spent in the sector... 13:37.625 --> 13:44.038 [JM]: If we took all of that money and invested it in something else, I wonder whether we might be better off. 13:44.098 --> 13:50.671 [JM]: But again, that's just something we're going to find out because maybe there is some incredible breakthrough in our future. 13:51.172 --> 13:55.701 [JM]: And maybe all of this money will, in the end, have been invested well, or maybe not. 13:55.681 --> 13:58.164 [JM]: I'm thinking not, but we're going to find out. 13:58.785 --> 14:10.579 [JM]: And if it turns out it wasn't invested well, I do wonder how could it have been invested instead that might have been better for the global economy and the human species. 14:10.719 --> 14:14.764 [JM]: But we'll probably not find out the answer as to what happens in that alternative timeline. 14:15.244 --> 14:20.811 [DJ]: No, that's going to be another science fiction story that you'll have to generate using a large language model. 14:21.263 --> 14:22.224 [JM]: Okay, moving on. 14:22.264 --> 14:28.893 [JM]: So in other news, Apple has introduced some updates to a few product lines. 14:29.274 --> 14:42.391 [JM]: They've introduced the M5 system-on-a-chip or SoC, and they are using the M5 in a few new products, including the MacBook Pro, the iPad Pro, and the Vision Pro. 14:42.411 --> 14:45.215 [JM]: A lot of pro things getting the M5, it seems. 14:45.656 --> 14:48.199 [DJ]: Which is weird because the M5 is not itself a pro chip. 14:48.415 --> 14:53.905 [JM]: Well, it doesn't yet come in Pro and Max variants, to be fair to the M5. 14:54.306 --> 15:00.137 [DJ]: It is just weird to me that the MacBook Pro has the M5, but not the M5 Pro, because the M5 Pro isn't out yet. 15:00.397 --> 15:03.984 [DJ]: And then when the M5 Pro comes out, it'll be in the MacBook Pro, but not the iPad Pro. 15:04.024 --> 15:05.987 [DJ]: The iPad Pro will just have the normal M5. 15:06.408 --> 15:07.470 [DJ]: It's all very confusing. 15:07.737 --> 15:11.123 [JM]: I think Apple's overloading of the term "pro" is not helping here at all. 15:11.544 --> 15:26.792 [JM]: But what is happening is that unlike the launch of the M4 and other SoCs, this particular time around the regular M5 is not being introduced along its cohorts, the M5 Pro and the M5 Max. 15:26.772 --> 15:30.957 [JM]: Presumably because of some kind of production issue or delay. 15:30.977 --> 15:46.097 [JM]: No one really knows why, of course, but presumably once whatever issues are sorted out, we will see the M5 Pro and M5 Max chips available in the MacBook Pro at some future date. 15:46.381 --> 15:48.664 [DJ]: Didn't they used to make a phone called the Pro Max? 15:49.204 --> 15:51.827 [DJ]: Like didn't the pro iPhone used to come in two sizes? 15:52.187 --> 15:54.069 [DJ]: So there's like the iPhone Pro and the Pro Max. 15:54.170 --> 15:54.830 [JM]: No, it still does. 15:55.111 --> 15:56.492 [JM]: No, the Pro Max still exists. 15:56.812 --> 15:57.333 [DJ]: Does it still? 15:57.874 --> 15:58.234 [DJ]: Okay. 15:58.614 --> 16:01.718 [DJ]: So I wish they would also make a system-on-a-chip called the Pro Max. 16:02.138 --> 16:04.181 [DJ]: And it's like a Max, but there's like two of them. 16:04.681 --> 16:05.963 [JM]: You just described the Ultra. 16:06.343 --> 16:06.743 [DJ]: Yeah, I know. 16:06.783 --> 16:09.947 [DJ]: But it would be more fun if they named it the same way as they named the big iPhone. 16:10.367 --> 16:11.549 [DJ]: So it's the Pro Max chip. 16:11.849 --> 16:14.532 [JM]: So you want the Ultra, but you want it to be called the Pro Max. 16:14.765 --> 16:15.546 [DJ]: Exactly. 16:15.586 --> 16:16.647 [DJ]: That's what I want. 16:16.907 --> 16:19.551 [DJ]: I want the naming to be even sillier and more confusing. 16:19.591 --> 16:24.817 [DJ]: Did you know that after a recent change, you can now watch Apple TV on the Apple TV on the Apple TV? 16:25.277 --> 16:26.018 [DJ]: Because that's awesome. 16:26.378 --> 16:27.480 [JM]: That is a thing that happened. 16:27.520 --> 16:27.880 [JM]: It's true. 16:29.502 --> 16:32.065 [DJ]: I mean, listen, naming things, naming things is hard. 16:32.085 --> 16:32.786 [DJ]: We all know this. 16:33.006 --> 16:39.894 [DJ]: But still, it feels like Apple has limited themselves to such a small set of possible words that everything like they have no choice now. 16:39.974 --> 16:42.677 [DJ]: Everything has to be "pro" or "max" or "TV". 16:42.957 --> 16:43.438 [DJ]: And that's it. 16:43.705 --> 16:51.753 [JM]: In typical fashion, these things were not announced via some special event or streamed video announcements. 16:51.833 --> 17:00.662 [JM]: These were, I think, just released in the form of press releases because they are more or less specification bumps to existing products. 17:01.203 --> 17:11.333 [JM]: If you have a M4 MacBook Pro or an M4 iPad Pro, there's very little reason for anyone to rush out and get the M5 variants of these. 17:11.313 --> 17:23.010 [JM]: These are largely useful if you are on say an M1 or an M2 and you think, "Okay, well, I want all of the advancements that have been made in between." 17:23.070 --> 17:29.760 [JM]: But the difference in terms of going from the M4 to the M5 for the iPad Pro or the MacBook Pro, is not particularly significant. 17:29.740 --> 17:42.022 [JM]: If the Vision Pro were a super useful product that lots of people used, then you might, I suppose, consider going from the M2 variant of it to the newer spec-bumped M5 variant of it. 17:42.102 --> 17:45.167 [JM]: But I don't know anyone personally who owns a Vision Pro. 17:45.147 --> 17:50.960 [JM]: Nor do I imagine that anyone's going to rush out and replace their M2 Vision Pro with an M5 Vision Pro. 17:51.040 --> 18:00.642 [JM]: But I think it is good that they finally updated this product with a more current SoC, even if I don't see a whole lot of future for that particular product line. 18:00.925 --> 18:04.832 [DJ]: Well, I guess the future of it is unknown to some degree, but it is. 18:04.932 --> 18:11.243 [DJ]: So it's nice to see they're continuing to invest in it because at least it means they didn't make one of these things and then go, "That was a mistake." 18:11.283 --> 18:12.125 [DJ]: "Let's just get rid of it." 18:12.165 --> 18:14.028 [DJ]: Like with my beloved iPhone mini. 18:14.489 --> 18:16.753 [DJ]: And technically they made two of those, but still not enough. 18:16.973 --> 18:18.576 [DJ]: Not enough iPhone minis, Apple. 18:18.796 --> 18:19.518 [DJ]: Just saying. 18:19.538 --> 18:22.022 [JM]: Yeah, I agree that it's good that they updated it for sure. 18:22.255 --> 18:27.422 [DJ]: Most updates are never "Oh, you should go out and trade in your thing from last year for this new thing." 18:27.923 --> 18:36.855 [DJ]: But I like seeing the sort of spec-bump updates because it at least means everyone who buys it starting now gets a better device. 18:37.496 --> 18:41.742 [DJ]: And then the way most people do it is, we kind of leapfrog, right? 18:41.762 --> 18:46.829 [DJ]: We're like, I have an M2 MacBook Air and I'm not going to buy the M3 or the M4 or the M5. 18:46.889 --> 18:51.435 [DJ]: But at some point when I want to replace my computer after maybe five-ish years, I'm 18:51.415 --> 18:56.005 [DJ]: I will get whatever is current, the M7 or M8 or something like that. 18:56.025 --> 19:07.249 [DJ]: What I like about keeping these products for a long time is that by the time you buy a new one, it really does feel like a gigantic upgrade because there's been a lot of movement in the hardware in the meantime. 19:07.469 --> 19:11.618 [DJ]: The software keeps getting worse as we continue to complain about, but the hardware much better. 19:12.037 --> 19:12.618 [JM]: For sure. 19:12.638 --> 19:22.300 [JM]: I have an M1 MacBook Pro and normally this is around the time I might think, "Ooh, okay, they're up to M5 now, maybe it's time to upgrade." 19:22.480 --> 19:31.260 [JM]: But the reality is the M1 MacBook Pro is so good and so fast that I do not feel any need to replace it. 19:31.240 --> 19:41.396 [JM]: I don't think I will feel any need to replace it until there is some other non-processor related improvement, such as an OLED screen. 19:41.696 --> 19:47.184 [JM]: That is something that I would probably leap at if it were available, say, next year. 19:47.205 --> 19:48.186 [DJ]: One can dream. 19:48.487 --> 19:52.371 [DJ]: The M2 MacBook Air is really good. 19:52.391 --> 19:57.177 [DJ]: I have my quibbles with it or whatever, but the bottom line is it's a heck of a computer. 19:57.197 --> 20:07.969 [DJ]: I kept my last MacBook Air for more than six years until like literally the battery was starting to swell and the storage drive just like died on me. 20:08.029 --> 20:11.052 [DJ]: And I was like, all right, I got to replace this thing. 20:11.032 --> 20:18.980 [DJ]: So I do tend to keep computers for quite a long time, and so I appreciate computers that last, and Apple's computers tend to do that. 20:19.420 --> 20:24.185 [DJ]: Yeah, I don't think I would replace this thing unless I decided I wanted a different kind of computer. 20:24.285 --> 20:34.876 [DJ]: I have occasionally thought I might like a MacBook Pro, especially when I do want to experiment with local large language models in particular. 20:34.896 --> 20:36.718 [DJ]: I have a couple use cases for that. 20:36.698 --> 20:53.473 [DJ]: And the bottom line is they just need a gigantic amount of RAM, ideally shared between the GPU and CPU, and if you want that, you really need a MacBook Pro in a laptop, or you need the Mac Studio, let's say, or a different computer that isn't made by Apple, but that's what we're talking about today. 20:53.913 --> 21:01.600 [DJ]: So that's one thing that has made me think I might upgrade, but otherwise, yeah, I mean, ride or die with this little laptop. 21:01.961 --> 21:04.663 [DJ]: So call me again when you release the M8. 21:05.504 --> 21:14.200 [JM]: I believe Apple did announce some improvements to the M5 in terms of the performance when running large language models. 21:14.320 --> 21:15.883 [JM]: So that could be interesting. 21:16.364 --> 21:22.215 [JM]: I think it'll be more interesting when, as you said, the M5 Pro and Max variants come out. 21:22.616 --> 21:25.521 [JM]: Those will probably be better suited to that particular task. 21:26.075 --> 21:26.877 [JM]: Okay, moving on. 21:26.957 --> 21:34.537 [JM]: The other day I was asked about Home Assistant, which I use for my home automation tasks. 21:34.557 --> 21:38.728 [JM]: And specifically I was asked, "Okay, I'm interested in Home Assistant. 21:39.169 --> 21:41.414 [JM]: You, Justin, have been telling me about it for a long time. 21:41.936 --> 21:42.818 [JM]: What should I get?" 21:42.798 --> 22:06.105 [JM]: And I immediately went to my web browser to pull up a link for the Home Assistant Yellow, which is an appliance that the folks that work on Home Assistant have produced as an official product to make it easier for folks to buy something that they can deploy without having to learn how to be a systems administrator and administer some server in their closet. 22:06.085 --> 22:13.209 [JM]: ... only to find when I went to look this up that just two days prior, the Home Assistant Yellow has apparently been discontinued. 22:13.631 --> 22:15.758 [DJ]: Insert sad Charlie Brown music here. 22:15.923 --> 22:17.486 [JM]: Sad trombone indeed. 22:17.526 --> 22:29.828 [JM]: They still produce the Home Assistant Green, which is a more entry-level device and is still, I would imagine, perfectly adequate for most people. 22:29.989 --> 22:36.801 [JM]: It doesn't come with, for example, a Thread/Matter radio built in. 22:36.781 --> 22:44.596 [JM]: I believe the Yellow variants had more RAM, a faster processor, and perhaps some other high-end features that the Green doesn't have. 22:44.996 --> 22:56.578 [JM]: But again, I think the Green is probably useful for most people, particularly if you're just trying to get started and you want to get something up and running easily and you don't want to be spending a lot of time monkeying to 22:56.558 --> 23:07.211 [JM]: download some ISO image and flash it onto a thumb drive and stick it into... This is like the kind of work that most people, like, I've already lost them. 23:07.392 --> 23:10.636 [JM]: So that's why I think the Green is great and they can use it. 23:10.716 --> 23:13.259 [JM]: And if at some point they outgrow it, well then no problem. 23:13.459 --> 23:21.970 [JM]: At least they have something that is functional and they're able to create automations without having to do all that other stuff. 23:22.350 --> 23:25.975 [JM]: And so that's why I'm really glad that that Green product still exists. 23:25.955 --> 23:35.528 [JM]: And for folks who want something that can do more than the Green can do, because side note, everywhere I looked, the Yellow is already out of stock. 23:35.988 --> 23:47.203 [JM]: And also, even if it's still were in stock, I started to do the math and a lot of times it doesn't actually work out favorably compared to other things you could do. 23:47.470 --> 23:48.872 [DJ]: You mean in terms of cost? 23:49.253 --> 23:51.737 [JM]: In terms of what you get per unit of cost. 23:52.197 --> 23:58.487 [JM]: The Yellow was cool in that just like the Green, you get it out of the box and you don't have to monkey with it in terms of setup. 23:58.868 --> 24:05.258 [JM]: But if you don't mind doing a little bit of setup, I think that mini PCs are probably a better option. 24:05.318 --> 24:08.483 [JM]: There are a lot of really interesting products out there. 24:08.463 --> 24:12.772 [JM]: And I'll put some links in the show notes to some examples of them. 24:13.293 --> 24:30.707 [JM]: But for around 300 or so United States dollars, you can get a very capable tiny little x86 mini PC that will run circles around the Green or the Yellow in terms of processing power and other capabilities. 24:30.687 --> 24:42.784 [JM]: So for someone who doesn't mind figuring out how to get the software onto a thumb drive and then figure out how to boot the mini PC from that thumb drive and then install, it could be a really good solution for Home Assistant. 24:43.144 --> 24:44.025 [JM]: All right, moving on. 24:44.366 --> 24:59.547 [JM]: I mentioned in our discussion about generative software companies this idea that they don't have a moat because it's not incredibly difficult for other companies to produce models of roughly comparable capabilities. 24:59.527 --> 25:04.320 [JM]: But recently, OpenAI announced their GPT Store. 25:04.821 --> 25:14.548 [JM]: And I think this is OpenAI's attempt at creating this moat, because the GPT Store is supposed to be their version, I suppose, of an app store. 25:14.528 --> 25:19.696 [JM]: But essentially, it's a bunch of integrations with a whole bunch of other services. 25:19.736 --> 25:28.390 [JM]: And it's not hard to extrapolate and project what they announced with what they want the world to look like, right? 25:28.790 --> 25:37.043 [JM]: Because right now, if you want to use ChatGPT, you install their app most likely, or use it in a browser. 25:37.023 --> 25:40.508 [JM]: And you're doing that in an environment that they don't have a lot of control over. 25:40.528 --> 25:45.115 [JM]: It's happening on your computer or it's happening on your phone or some other device you have. 25:45.515 --> 25:55.269 [JM]: But to me, it seems very clear that they probably want it at some point to be on devices that they manufacture and ship. 25:55.289 --> 25:59.335 [JM]: But even if that doesn't happen, they definitely want to be the operating system. 25:59.315 --> 26:02.379 [JM]: That's the thing that to me is very clear about what they want. 26:02.800 --> 26:06.445 [JM]: They want all of the other software to be obsolete. 26:06.726 --> 26:09.029 [JM]: Because why would you need anything else? 26:09.590 --> 26:14.016 [JM]: But this chatbot, this chatbot will do anything you ask it to do. 26:14.457 --> 26:17.121 [JM]: So you don't need any of the apps on your phone anymore. 26:17.101 --> 26:19.785 [JM]: You don't need any of the software running on your computer anymore. 26:20.366 --> 26:36.229 [JM]: You could even, if you look out far enough, envision some augmented reality glasses that you're wearing where you just talk to it and ask it questions and it gives you the answers, either visually or via audio or both. 26:36.650 --> 26:38.593 [JM]: And you don't need any other device. 26:38.973 --> 26:41.137 [JM]: You've got your screen, you've got your audio. 26:41.537 --> 26:44.762 [JM]: So the hardware is already covered and they're providing all the software. 26:44.742 --> 26:48.929 [JM]: So I think this is how they think they're going to create this moat. 26:49.350 --> 26:58.826 [JM]: And this is how they think they can justify trillions of dollars worth of investment because their vision is to replace the whole industry. 26:59.227 --> 27:01.090 [JM]: They're going to replace all of it. 27:01.591 --> 27:04.876 [JM]: Apple, Google, Facebook, Microsoft. 27:05.197 --> 27:06.379 [JM]: They're going to replace the whole thing. 27:06.699 --> 27:08.843 [JM]: That's, I believe, what they're shooting for here. 27:09.127 --> 27:10.729 [DJ]: Yeah, that does seem like the goal. 27:11.230 --> 27:23.065 [DJ]: The thing that seems to be the case about this GPT Store, though, too, just to like make it explicit, is not that they want to replace all application software under the hood, but they want to make it so you don't have to run an app. 27:23.466 --> 27:23.726 [DJ]: Right. 27:23.826 --> 27:31.596 [DJ]: So one of the examples they gave in the announcement was AllTrails, which is a really cool app if you're into hiking because it has all trails. 27:31.717 --> 27:36.703 [DJ]: Well, it has this giant database of like hiking locations with all kinds of information about them. 27:36.683 --> 27:40.849 [DJ]: To use it, you have to go to this thing called a web site or this thing called an app. 27:41.350 --> 27:48.341 [DJ]: And we all saw over the last 20 years how apps kind of displaced web sites as the site of application usage. 27:48.441 --> 27:57.274 [DJ]: And I suppose for that matter, earlier on than that, web sites kind of displaced running a different type of application on like your desktop computer in a way. 27:57.254 --> 28:01.339 [DJ]: But you always had to run this program specific to the thing you wanted. 28:01.399 --> 28:07.267 [DJ]: So if you want to get an AllTrails database, you have to use the AllTrails app or website somehow. 28:07.748 --> 28:13.315 [DJ]: And instead, if I understand correctly, the goal with the GPT Store is you just chat with ChatGPT. 28:13.435 --> 28:21.085 [DJ]: And when it needs to, it will go call AllTrails and say, tell me about this thing, and then give the information back to you. 28:21.065 --> 28:47.680 [DJ]: And that feels like the model that if you scale it up in the way that you gave in your sort of hypothetical example there, that yeah, you can see a world where like to some degree application software doesn't exist anymore, because the way you get to applications is always through ChatGPT, which basically turns every third-party software company from an app maker into an API maker where the API is available to ChatGPT. 28:47.660 --> 28:51.088 [DJ]: And I hate this idea for a variety of reasons. 28:51.188 --> 29:01.331 [DJ]: But one of them is just like there are big complaints about how Apple and Google have become on the smartphone became the choke point for application distribution. 29:01.371 --> 29:04.217 [DJ]: And then there's lots and lots of controversy 29:04.197 --> 29:14.314 [DJ]: about how if you want to deploy an app on an iPhone, you have to go through Apple, you have to go through their app store, they have to approve of your app, they're going to take 30% of all your sales, period. 29:14.394 --> 29:15.155 [DJ]: You don't have a choice. 29:15.215 --> 29:21.345 [DJ]: There's no other way to run software on an iPhone except to deploy it through the Internet and have it in a web browser. 29:21.325 --> 29:28.613 [DJ]: Well, this would be the same world, except OpenAI would be taking the cut, essentially, because now they're the gateway to all of your customers. 29:29.053 --> 29:31.135 [DJ]: And the bottom line is it's just bad. 29:31.195 --> 29:36.421 [DJ]: It's bad to have any other company be the only way you can reach your customers. 29:36.882 --> 29:42.988 [DJ]: And in this hypothetical ChatGPT future, presumably people don't go to the web and search for stuff. 29:43.308 --> 29:44.450 [DJ]: They don't go to their phone. 29:44.910 --> 29:45.931 [DJ]: They don't go to app stores. 29:45.951 --> 29:46.952 [DJ]: They don't download apps. 29:46.992 --> 29:50.556 [DJ]: They just use ChatGPT as their interface to everything. 29:50.536 --> 29:54.804 [DJ]: which of course makes ChatGPT super valuable and makes everyone else totally replaceable. 29:55.004 --> 29:58.070 [DJ]: So yeah, I don't like this idea, of course. 29:58.391 --> 30:09.351 [DJ]: But as you point out, it is interesting to see, especially as a reaction to the observation that just building the chatbot is not necessarily a sustainable business. 30:09.872 --> 30:14.080 [DJ]: It's certainly not a business where competitors can't just come along and eventually... 30:14.060 --> 30:20.325 [DJ]: And so instead, their bet becomes like the other tech giants before them. 30:20.365 --> 30:24.964 [DJ]: "How do we effectively replace the platform that preceded us?" 30:25.096 --> 30:28.202 [JM]: To be clear, I don't love this vision either. 30:28.222 --> 30:29.985 [JM]: I don't like anything about it. 30:30.446 --> 30:46.817 [JM]: It's been really hard witnessing the birth of the World Wide Web and then seeing it largely supplanted by mobile apps that, as you said, have gatekeepers in terms of how that software is approved 30:46.797 --> 30:54.166 [JM]: and shipped and pulled off of app stores when organizations like governments demand it. 30:54.627 --> 31:03.657 [JM]: And so you can almost see this progression where you have this huge open thing, which is the Web, and then it gets concentrated down into essentially two companies. 31:04.178 --> 31:09.885 [JM]: And now we're talking about a potential future in which it gets concentrated down to one company, 31:09.865 --> 31:18.594 [JM]: where OpenAI now is going to be the choke-point and distribution center for all of software. 31:19.135 --> 31:34.050 [JM]: And yeah, that's not a particularly rosy-looking future in terms of if you are a fan of software, and you like some of the open characteristics of the Web, and you don't particularly love how that's been 31:34.030 --> 31:39.596 [JM]: significantly supplanted by mobile apps, this potential future isn't one that I relish. 31:39.876 --> 31:47.503 [JM]: And I definitely hope it does not come to pass, even if that's perhaps a probable outcome. 31:48.064 --> 31:48.604 [DJ]: I agree. 31:49.165 --> 31:58.054 [DJ]: And something that always does give me a little bit of hope is the Web still exists and other ways of getting software and running it still exist. 31:58.234 --> 32:03.379 [DJ]: And I'm hopeful that that will continue, even if it only continues in the margins. 32:03.359 --> 32:19.950 [DJ]: I really do feel myself over the last few years, especially, and it feels like it's accelerating, becoming just weirder and weirder as far as like computing and technology and software goes because I'm so unhappy 32:19.930 --> 32:33.540 [DJ]: with this sort of consolidation that's basically all about one or two or three companies just acting in their own relentless self-interest with gigantic consequences for all of human civilization. 32:33.600 --> 32:38.952 [DJ]: Generally not positive consequences, by the way, in case you just got here and haven't been paying attention. 32:38.932 --> 32:43.018 [DJ]: For planet Earth's new listeners, it hasn't all been good. 32:43.078 --> 32:57.719 [DJ]: So I really want to take all of my attention and resources, because that's really all I have the control over, and invest them literally or figuratively in the places where technology still inspires me. 32:58.179 --> 33:03.707 [DJ]: So things like the World Wide Web and open-source software and platforms. 33:03.687 --> 33:07.811 [DJ]: And there's really interesting open source hardware out there now, too. 33:08.132 --> 33:15.999 [DJ]: You know, as much as I do love the laptops that Apple makes, my next laptop is probably not going to be a Mac. 33:16.080 --> 33:24.768 [DJ]: It's probably going to be from one of these other companies that has a different set of practices and a different sort of relationship with consumers. 33:25.389 --> 33:31.315 [DJ]: Because, yeah, I see this kind of stuff and, OpenAI announces GPT Store, and I just go like, "That sucks." 33:31.295 --> 33:35.582 [DJ]: Like I'm not happy about it, especially like... I work for a software company that makes software. 33:35.602 --> 33:41.191 [DJ]: I don't really want to become a essentially sharecropper to this one company. 33:41.211 --> 33:43.294 [DJ]: It's bad enough with the mobile app stores already. 33:43.835 --> 33:49.264 [DJ]: Mobile app stores at least let you still have a brand that you can express in your mobile application. 33:49.785 --> 33:54.332 [DJ]: But the thing that ChatGPT is proposing here, like you don't even have a brand anymore. 33:54.632 --> 33:58.078 [DJ]: No one even needs to know you exist because they're just using ChatGPT. 33:58.244 --> 34:00.848 [JM]: No, it just becomes generative software serfdom. 34:01.289 --> 34:01.729 [DJ]: Exactly. 34:01.789 --> 34:02.250 [DJ]: Yeah. 34:02.430 --> 34:09.381 [DJ]: And like for now it says, I answered your question using AllTrails, but eventually who cares as long as it answers your question correctly. 34:09.761 --> 34:27.508 [DJ]: I'm not a fan of this at all, but all I can do about it is, you know, again, like just invest my own personal relationship with technology in the places where some degree of openness and I don't know, "freedom" feels like a loaded word, but something like that still flourishes. 34:27.640 --> 34:28.261 [JM]: Absolutely. 34:28.301 --> 34:43.850 [JM]: It doesn't take too much imagination to envision a world in which OpenAI or some other company, if OpenAI is not the one that ends up achieving what we're describing here, where they just pull like an Amazon Basics and say like, "Okay, AllTrails, guess what?" 34:44.271 --> 34:48.619 [JM]: "We developed in-house our own version of your API. 34:48.599 --> 34:50.722 [JM]: "And we don't need AllTrails." 34:50.762 --> 34:51.904 [JM]: "We don't need you anymore." 34:51.944 --> 35:03.743 [JM]: "We're now going to just Amazon-Basics you right out of the equation." And just continue to do that one sector at a time, to the point where they really are owning all of software at that point. 35:04.144 --> 35:05.967 [JM]: But I'm with you that 35:05.947 --> 35:15.561 [JM]: we live in a world right now where there's still people who are passionate about things that for you and I and some other people like us that are niche. 35:15.581 --> 35:32.545 [JM]: And as long as those little corners of the world exist and continue to exist, then it really won't matter too much, at least to me, what companies like this do and how much of the global market for software they 35:32.525 --> 35:38.653 [JM]: suck into their vortex because it won't affect too much my day-to-day life. 35:38.954 --> 35:46.704 [JM]: Because I'm with you that I don't know how long I'll continue to use Apple products if they continue down the path that they seem to be going down. 35:47.245 --> 35:53.474 [JM]: I have become more and more curious about alternatives, even if I don't really seriously consider them. 35:53.634 --> 35:56.398 [JM]: I'm certainly more interested than I have ever been. 35:56.418 --> 36:01.705 [JM]: And depending on how the next few years go, it could be a very significant 36:01.685 --> 36:07.815 [JM]: shift in terms of how I spend my time and the tools that I use to do the things I want to do. 36:07.875 --> 36:16.710 [JM]: And thankfully, there's plenty of great open source and more open hardware to run that open-source software. 36:17.151 --> 36:17.371 [DJ]: Yeah. 36:17.391 --> 36:24.283 [DJ]: And I look forward to continuing to share it right here alongside our links to web sites where you can spin a dog around. 36:25.006 --> 36:25.748 [JM]: All right, everybody. 36:25.768 --> 36:26.650 [JM]: Thanks a lot for listening. 36:26.670 --> 36:27.833 [JM]: We hope you enjoyed the show. 36:27.853 --> 36:33.827 [JM]: You can find me on the web at justinmayer.com and you can find Dan on the web at danj.ca. 36:34.288 --> 36:38.357 [JM]: Share your thoughts about this episode via the Fediverse at justin.ramble.space.