WEBVTT 00:00.031 --> 00:04.903 [JM]: Before I mention a handy tool that I think everyone should know about... Dan, quick question for you: 00:04.923 --> 00:12.460 [JM]: How do you find out, if you do find out, when an artist that you like has released new music? 00:12.821 --> 00:17.753 [DJ]: Well, I admit that I don't follow that very closely. 00:17.733 --> 00:26.371 [DJ]: Probably these days, the way I find out is through my streaming service of choice, which at the time of this recording is Qobuz. 00:26.732 --> 00:27.553 [DJ]: Yes, that's right. 00:27.874 --> 00:31.321 [DJ]: I just appreciate being given an opportunity to say their silly name. 00:31.802 --> 00:34.628 [DJ]: But yeah, generally it would be on... 00:34.608 --> 00:35.709 [DJ]: that app. 00:35.789 --> 00:48.865 [DJ]: I would open it up, and among various things on the dashboard, it might say, if you follow a given artist, like, hey, this artist has a new album out, or if you just go to that artist's profile, it will generally show you their latest thing. 00:49.345 --> 00:57.915 [DJ]: But typically, I am not really more intent than that on knowing when artists come out with albums these days. 00:57.895 --> 00:59.898 [DJ]: I'm not sure I ever was, honestly. 01:00.178 --> 01:09.753 [DJ]: When I was really into Nine Inch Nails in high school and university, I don't know that I was scouring the headlines to learn when their latest album came out. 01:09.833 --> 01:19.448 [DJ]: I would eventually learn about it, and then I would buy a copy at HMV, which I believe was a store that sold things called compact discs. 01:19.698 --> 01:29.892 [JM]: As someone who grew up listening to music albums, there have been a number of times throughout my life where I have heard a song, say on the radio, and thought, "Whatever happened to these people?" 01:30.252 --> 01:38.704 [JM]: "Did they ever release any new music since the decade or more that has passed since the last time I remember one of their albums being released?" 01:39.284 --> 01:49.418 [JM]: And I will look it up and see that in the intervening years, they've released a half dozen albums that I have never heard of and did not know existed and thought, 01:49.398 --> 01:50.479 [JM]: Well, that's a waste. 01:50.499 --> 01:55.684 [JM]: I could have been listening to all that music over the last decade or decades, as the case may be. 01:55.744 --> 02:03.711 [JM]: And so there have been a number of times early in my software development life where I thought, I should build this thing. 02:03.771 --> 02:09.777 [JM]: I should build something that notifies me when artists that I like releases new albums. 02:10.077 --> 02:15.882 [DJ]: I find it a little hard to believe that you didn't know that Taylor Swift released a sophomore album. 02:15.902 --> 02:16.823 [DJ]: I'm just saying. 02:16.837 --> 02:22.172 [JM]: Can't say I'm a Swifty, so I don't think I would have known it in any case. 02:22.304 --> 02:29.997 [DJ]: I just like the idea of you following the most popular artist in the world and being like, wait, when did she put out all this other stuff? 02:30.017 --> 02:42.136 [JM]: And like many software people, I never really got around to scratching that itch and building this thing in part because before I actually said, "Okay, that's it, 02:42.156 --> 02:45.682 [JM]: I've had it," and rolled up my sleeves and built it. 02:46.143 --> 02:47.485 [JM]: I found a tool 02:47.465 --> 02:53.134 [JM]: that does the very thing I wanted to build and thus did not need to build it, which is great. 02:53.575 --> 02:55.718 [JM]: Because in the end, I didn't really want to build it. 02:55.758 --> 02:56.920 [JM]: I mean, yeah, that's fun and all. 02:57.321 --> 03:04.232 [JM]: But to have found precisely the thing I want without me having to build it myself is a huge win. 03:04.653 --> 03:09.160 [JM]: And that tool is called... And I really have no idea how to pronounce this. 03:09.721 --> 03:12.105 [JM]: I'm sitting here hesitating... 03:12.085 --> 03:20.776 [JM]: wondering, I have never spoken this thing aloud despite having used it for years, but it's "Muspy". 03:20.796 --> 03:25.762 [JM]: I mean, it's a portmanteau of music and Python, presumably. 03:25.802 --> 03:28.646 [JM]: So you decide how to pronounce that. 03:28.846 --> 03:33.892 [JM]: I'm just going to keep calling it Muspy, which again, sounds ridiculous, but here we are. 03:34.293 --> 03:38.498 [DJ]: So, I mean, for starters, Qobuz, but... 03:39.153 --> 03:47.062 [DJ]: But also, if what they wanted was a portmanteau of music and Python, I think they missed the extremely obvious "SoundSnake". 03:47.182 --> 03:48.483 [JM]: It was right there. 03:48.544 --> 03:49.324 [DJ]: Right? 03:49.785 --> 03:53.489 [DJ]: I mean, maybe there probably already is another piece of software called that. 03:53.669 --> 03:58.274 [DJ]: But strangely enough, it tells you when your favorite snakes have released new albums. 03:58.895 --> 03:59.516 [JM]: Oh, great. 04:01.118 --> 04:01.939 [JM]: Who doesn't want that? 04:02.479 --> 04:05.903 [JM]: I don't remember when I started using this (web) app. 04:05.883 --> 04:10.730 [JM]: But it has delivered exactly what I wanted for years. 04:11.251 --> 04:22.106 [JM]: And the way you use it is you add artists to a list of favorites and you get notified via email when that artist produces new album releases. 04:22.407 --> 04:26.052 [JM]: And that's really as complicated as it gets. 04:26.032 --> 04:32.740 [JM]: There's also an RSS feed if you prefer to get your information that way instead of, say, via email. 04:33.141 --> 04:40.230 [JM]: And as I said before, this is not only written in Python, it also uses Django and is open source. 04:40.810 --> 04:47.559 [JM]: If you go to the repository, you'll see that the last commit was 12 years ago. 04:47.579 --> 04:51.324 [JM]: And the commit message for that says: "Sync with production" 04:51.344 --> 04:52.545 [JM]: So... 04:52.525 --> 05:02.003 [JM]: the question, and this is unanswered, and one of these days I should reach out to the author of this tool and just ask questions that are going to come up right here and now. 05:02.023 --> 05:07.553 [JM]: And one of those is, is that commit 12 years ago really the last one? 05:07.613 --> 05:08.695 [JM]: Like, is the... 05:08.675 --> 05:17.336 [JM]: code that's running on muspy.com the same as the code that was last allegedly touched here in this repository 12 years ago? 05:17.396 --> 05:22.248 [JM]: Because the commit message "Sync with production" leads me to think the answer to that question is "No". 05:22.288 --> 05:26.278 [JM]: It leads me to think that the code's been changed and 05:26.258 --> 05:32.266 [JM]: and deployed to production, just not synchronized with the code that we are seeing here in GitHub. 05:32.586 --> 05:36.731 [JM]: And I love seeing things in recent-ish commit messages. 05:36.812 --> 05:42.098 [JM]: And by "recent", I mean, 14 years ago, that say "Update manage.py to Django 1.4". 05:42.799 --> 05:53.633 [JM]: So it's, on some level... I kind of hope that the software running on the web site is more recent than the 12-year-old code that's running here in this repository, because 05:53.613 --> 05:57.538 [JM]: from a security perspective, it's probably not great if that's the case. 05:58.038 --> 06:06.649 [JM]: And if they exfiltrate my list of favorite artists, like, oh, no, it's not exactly the highest, you know, quality... biggest risk payload. 06:07.270 --> 06:08.591 [DJ]: Are you kidding? 06:08.691 --> 06:13.978 [DJ]: But just imagine like criminals across the world could pretend to have your taste in music. 06:14.659 --> 06:15.880 [DJ]: That's right. 06:15.900 --> 06:18.383 [DJ]: That's the most dangerous form of identity theft of all. 06:18.844 --> 06:20.246 [JM]: Artist preference doxxing. 06:21.147 --> 06:21.607 [DJ]: That's right. 06:21.687 --> 06:22.168 [DJ]: Yeah. 06:22.148 --> 06:23.950 [DJ]: Yeah, that's actually scarier. 06:24.391 --> 06:31.821 [DJ]: You know, someone posts online, it's like, hey, did you know Justin likes... insert name of super embarrassing artist for you to like here? 06:31.861 --> 06:34.485 [DJ]: So... the current version of Django isn't 1.4. 06:34.685 --> 06:38.490 [DJ]: That was my big takeaway from that whole conversation. 06:38.810 --> 06:40.352 [JM]: We're at 6.0 right now. 06:40.412 --> 06:44.578 [JM]: So yeah, been a while since 1.4 was released. 06:44.727 --> 06:47.992 [DJ]: It is always strange to come across repositories like that. 06:48.253 --> 06:52.700 [DJ]: And there's always, I always feel this note of dismay, like, oh no, is this software abandoned? 06:52.960 --> 06:56.887 [DJ]: But then on the other hand, maybe I think, but maybe it was perfected 12 years ago. 06:57.268 --> 06:58.249 [DJ]: Doesn't need to be changed. 06:58.590 --> 06:59.591 [JM]: Could very well be. 06:59.631 --> 07:01.875 [JM]: And either way, as much grief... 07:02.159 --> 07:14.893 [JM]: as I'm giving this repository for last being touched allegedly 12 years ago, I am deeply grateful to the person who created this project, who continues to run it. 07:15.294 --> 07:17.400 [JM]: I've received a lot of value 07:17.380 --> 07:18.621 [JM]: from it over the years. 07:19.042 --> 07:28.714 [JM]: And if you, like me, want to know when your favorite artist has released new albums, then go to muspy.com and create an account and add some artists. 07:29.174 --> 07:37.905 [JM]: And then you too will be surprised to see that, oh, this artist that I enjoyed listening to when I was a teenager still makes music today. 07:37.925 --> 07:40.928 [JM]: It's a nice thing to be notified about when it happens. 07:41.389 --> 07:41.709 [JM]: All right. 07:41.810 --> 07:46.375 [JM]: In other news, shortly after we talked about the... 07:46.355 --> 07:55.251 [JM]: M5 Pro and Max variants of the MacBook Pro that was released last week, Apple announced the MacBook Neo. 07:55.732 --> 08:01.663 [JM]: And the gist of the MacBook Neo is that it is a lower cost 08:01.643 --> 08:15.322 [JM]: notebook computer that is powered by an iPhone system-on-a-chip, or processor, and not a Mac-specific system-on-a-chip like the M1 through M5. 08:15.662 --> 08:26.537 [JM]: So the MacBook Neo is powered by an A18 Pro, which if memory serves is the system-on-a-chip that debuted with 08:26.517 --> 08:28.081 [JM]: the iPhone 16. 08:28.722 --> 08:33.653 [JM]: And this MacBook comes in a single form factor: 13 inches. 08:34.154 --> 08:39.346 [JM]: And the main headline-grabbing feature of this MacBook is the price. 08:39.887 --> 08:44.678 [JM]: It starts at $600, or $500 08:44.658 --> 08:45.961 [JM]: for educational pricing. 08:46.321 --> 08:51.211 [JM]: And I believe this is the lowest-price laptop that Apple has ever made. 08:51.652 --> 09:00.709 [JM]: I think they had to make a number of interesting innovations in order to get the price down that low. Everything from changing how 09:01.010 --> 09:10.563 [JM]: they make their aluminum enclosure, to finding a lower-priced screen, built-in camera, microphone, speakers. 09:10.603 --> 09:28.148 [JM]: I think in every single aspect, they found less expensive ways of supplying components that are clearly not at the same level as their more expensive notebook computers, but good enough for the vast majority of people who need a computing device. 09:28.128 --> 09:34.840 [JM]: One of the things that they had to do to make this all work is only offer two ports. 09:35.421 --> 09:38.928 [JM]: And if you need to charge, that means you only have one port available. 09:39.388 --> 09:41.793 [JM]: And those two ports, by the way, are not equal. 09:42.334 --> 09:50.288 [JM]: One of those two ports is believe it or not, a USB 2.0 port with a maximum throughput of 09:50.268 --> 09:57.620 [JM]: 480 megabits per second, which is ancient and very slow by modern standards. 09:57.960 --> 10:11.922 [JM]: And there's no indication by the way, as to which of the two ports is the good one and which is the ancient one you should really only use if you are say connecting like an external keyboard or mouse or some other peripheral device. 10:11.902 --> 10:24.377 [JM]: And if you were to, say, try to plug in a 4K monitor into the slower port, my understanding is that a little piece of software is included that pops up on your screen. 10:24.417 --> 10:34.028 [JM]: Well, not the monitor screen, but the built-in screen that says, yeah, you're going to need to plug this thing into the other port, which is not the... 10:34.008 --> 10:50.700 [JM]: greatest user experience and something that presumably will be rectified in future iterations of the MacBook Neo, where it's not limited, say, by the capabilities of an iPhone processor like the A18. 10:50.680 --> 10:59.556 [JM]: And presumably, even if they use subsequent iPhone SoCs in the MacBook Neo, this limitation will likely go away. 10:59.757 --> 11:09.094 [JM]: But for now, it's a necessary compromise in order to deliver the product that Apple was trying to ship at the price that it is being offered at. 11:09.074 --> 11:21.848 [JM]: And I think it's great that Apple is doing something that no one really thought they ever would and that Apple more or less said they wouldn't do, which is to target this end of the market. 11:22.308 --> 11:29.076 [JM]: And when I say that Apple said they wouldn't do, I'm referring to quotes where they said things like, "We're not going to ship junk." 11:29.096 --> 11:38.406 [JM]: And I think the implicit thing that I'm saying here is that up until now, I don't think that Apple felt like they could ship a notebook at this price 11:38.386 --> 11:42.072 [JM]: that wasn't junk, while still making it somewhat profitable. 11:42.512 --> 11:47.300 [JM]: And who knows how much money they're making on this particular product. 11:47.640 --> 11:51.706 [JM]: I imagine the margins are nowhere near their more expensive products. 11:51.987 --> 12:00.640 [JM]: But I think it's great that they're trying to deliver a Mac at this price level, because otherwise, the answer has been well, "Just get an iPad." 12:01.061 --> 12:02.503 [JM]: "If you don't have 12:02.483 --> 12:08.453 [JM]: you know, $1,000, $1,100, $1,200 to spare, then just go get an iPad for half that price." 12:08.754 --> 12:23.479 [JM]: And I like that the answer now is that you don't have to do that, that you can get a notebook Mac at a price that's comparable to, say, a similarly specced or at least, you know, low-ish end iPad. 12:23.459 --> 12:26.787 [JM]: And the question I keep asking myself is why? 12:26.827 --> 12:28.612 [JM]: Like, why is Apple doing this? 12:28.833 --> 12:35.750 [JM]: Like, what do they get out of targeting a segment of the market that seems very much not what they normally do? 12:35.770 --> 12:39.780 [JM]: It's not high margin, high profit per unit. 12:39.760 --> 12:42.783 [JM]: And I think the answer that I've come up with is twofold. 12:43.043 --> 12:52.212 [JM]: One is to expand the overall market for people who are trying to buy a laptop computer and who would normally reach for Windows. 12:52.573 --> 13:01.682 [JM]: This offers them an opportunity to get something that is really, I imagine, quite better than what they would otherwise be able to get. 13:01.782 --> 13:04.705 [JM]: My understanding is that if you were to go out and try to buy a 13:04.685 --> 13:16.025 [JM]: a notebook that runs Windows for $600, the thing you would get comparatively is junk in terms of the overall fit-and-finish and user experience. 13:16.045 --> 13:17.848 [JM]: I'm not talking from personal experience. 13:17.868 --> 13:21.715 [JM]: I'm just talking about from what I have read and what I've heard other people say. 13:21.695 --> 13:26.001 [JM]: So I think by expanding the market, they're going to be able to reach a whole new audience. 13:26.521 --> 13:39.639 [JM]: And who knows, those people over time, as they progress throughout their life, may find that at some point, actually a MacBook Air is a better successor when it's time to replace their Neo or maybe even a MacBook Pro. 13:39.940 --> 13:46.268 [JM]: And so Apple gets to win that customer earlier in that person's life and hopefully get 13:46.248 --> 13:51.276 [JM]: take them along in terms of moving them up to higher profit margin items. 13:51.676 --> 14:00.270 [JM]: And the other reason that occurs to me is, and this is one I'm not particularly jazzed about, is services revenue. 14:00.830 --> 14:03.835 [JM]: I wonder whether this is a way of getting people 14:03.815 --> 14:13.232 [JM]: people into Apple's ecosystem, they're buying a product that really doesn't provide very much profit for Apple, because the Neo doesn't have, I imagine, great margins. 14:13.653 --> 14:26.076 [JM]: But then they're able to sell them, say, iCloud storage, or a Creator Studio subscription, or Apple TV, or all the other various things that are quite high margin for Apple. 14:26.056 --> 14:46.957 [JM]: But regardless of their motives, I think it's cool that Apple continues to innovate in the Mac space and that they did it in a way in which isn't targeting the very wealthy with $6,000 - $7,000 monitors and instead is producing a great Mac for what you're spending. 14:47.338 --> 14:48.300 [JM]: I think it's a strong move. 14:48.280 --> 14:56.859 [DJ]: Having this Mac come out very soon after the Creator Studio bundle announcement does seem like a natural fit. 14:56.939 --> 15:06.660 [DJ]: Because if you think about those Creator Studio apps, I think they're the kind of apps that most people would use on a computer, like a laptop computer, not on an iOS device. 15:06.640 --> 15:14.432 [DJ]: And so I think your point about services revenue is probably a big part of the value proposition of this, which is okay. 15:14.853 --> 15:22.745 [DJ]: We've rolled out a service that is mostly going to be purchased by Mac users as opposed to iPhone and iPad users. 15:22.765 --> 15:27.253 [DJ]: So how do we get there to be more Mac users becomes an important question. 15:27.333 --> 15:29.877 [DJ]: And this is a great way to do that. 15:30.177 --> 15:33.222 [DJ]: So it does seem like a natural fit. 15:33.202 --> 15:43.397 [DJ]: Even if neither you nor I really love many of the other incentives that Apple seems to have taken on in their pursuit of growing services revenue. 15:43.664 --> 15:50.917 [JM]: I also think it's great that they released this computer in more colors than Apple has released in a while. 15:51.378 --> 15:58.211 [JM]: I still feel like there's plenty of room for them to add more color just across their product line everywhere. 15:58.271 --> 16:03.360 [JM]: You can get it in the usual silver and black, 16:03.340 --> 16:04.442 [JM]: they call it indigo, 16:04.542 --> 16:08.328 [JM]: it basically just looks grayish, like a dark, dark gray. 16:08.788 --> 16:16.500 [JM]: But in addition to those rather standard Apple hues, it comes in a yellow and pink finish. 16:16.981 --> 16:21.147 [JM]: And I don't think the yellow is for me, but I could see the pink one being pretty popular. 16:21.346 --> 16:37.043 [DJ]: I don't know if Apple has abandoned the 24-inch iMac that they came out with a few years ago, because I can't remember the last time that thing got a revision, but the colors they introduced for that computer really made me excited. 16:37.023 --> 16:43.430 [DJ]: And then they proceeded to never ship any other products in interesting colors like that. 16:43.550 --> 16:45.412 [DJ]: So come on, guys. 16:45.853 --> 16:47.094 [DJ]: Like, I agree with you. 16:47.234 --> 16:47.615 [DJ]: I wish... 16:47.715 --> 16:54.142 [DJ]: like, supposedly Apple is the best company in the world at this point at, like, aluminum manufacturing. 16:54.202 --> 17:06.456 [DJ]: And please give us more colorful products because, yes, they are, especially as you go higher end, sadly, the more money you pay for an Apple product, the more boring-looking it is. 17:06.436 --> 17:07.699 [DJ]: And that's that's a shame. 17:07.739 --> 17:08.481 [JM]: Yeah. 17:08.501 --> 17:24.478 [JM]: And you're totally right to highlight the iMac as this weird outlier in their lineup where they actually seemingly care about colors because you can still get the iMac in blue, purple, pink, orange, yellow, and green colors 17:24.458 --> 17:34.173 [JM]: in addition to the usual silver, that's a full seven different colors, if you count silver as a color, that you can get for the iMac. 17:34.373 --> 17:38.018 [JM]: And that's just unparalleled across Apple's line-up. 17:38.058 --> 17:41.423 [JM]: It's the only place where they seemingly care to offer colors. 17:41.483 --> 17:44.228 [JM]: I think that this is something they should do more of. 17:44.268 --> 17:44.608 [JM]: I'm with you. 17:45.069 --> 17:48.714 [JM]: Yeah, and it'll be interesting to see how many of these they sell. 17:48.734 --> 17:53.942 [JM]: I think also in part, their motivation could be because the... 17:53.922 --> 18:11.762 [JM]: M1 MacBook Air that they originally released I think at the end of 2020, so at this point over five years ago, up until this announcement was and maybe still is, I don't know, for sale at exclusively Walmart and 18:11.742 --> 18:22.236 [JM]: that's the only place where you could buy a brand-new Mac for, I think it's $650, just $50 more than this new MacBook Neo. 18:22.857 --> 18:39.859 [JM]: And I wonder if that particular segment sold well for Apple, but they realized like, we can't keep selling an architecture that's 5+ years old as brand new, in part because every year they continue doing that, that's 18:39.839 --> 18:47.307 [JM]: to some degree one more year they are kind of on the hook for supporting it from a software perspective, which gets harder and harder. 18:47.347 --> 19:00.422 [JM]: So I think they were at a point where they're like, "Well, we either need to just get rid of that Walmart exclusive or give them, say, an M2 or something more up-to-date to sell instead." 19:00.482 --> 19:08.631 [JM]: "Or... we could just cut out the middleman and we could take this market ourselves instead of offering it exclusively to Walmart." 19:08.611 --> 19:10.173 [JM]: And I think that's what they chose to do here. 19:10.554 --> 19:20.289 [DJ]: Yeah, I sort of wish I could be a fly on the wall inside Apple and know for just how long have they been trying to pull off a Mac using an A-series system-on-a-chip. 19:20.710 --> 19:27.180 [DJ]: Because I assume maybe all the other pieces being equal, I assume that was the main constraint that they were going for. 19:27.280 --> 19:34.551 [DJ]: If their previous cheap Mac was just using the oldest M-class chip available... 19:34.531 --> 19:42.801 [DJ]: I'm sure that for a long, long time, they must have been trying to figure out like, all right, how, when is the A series at a point where it could realistically power a Mac? 19:43.182 --> 19:45.805 [DJ]: What you just pointed out is like very obvious. 19:45.825 --> 19:51.232 [DJ]: Like as long as they kept selling the M1, they would be like, okay, we got to stop selling that M1. 19:52.133 --> 19:52.834 [DJ]: How do we do it? 19:52.994 --> 19:56.438 [DJ]: Like, how do we keep, how do we shore up the low end of this lineup? 19:56.587 --> 20:01.252 [JM]: The most Apple-like way of dealing with that conundrum would be just to cancel it. 20:01.612 --> 20:10.481 [JM]: And I'm really glad that in this case, they did the un-Apple-like thing instead of saying like, "Okay, well, we don't want to support the M1 architecture forever from a software perspective." 20:10.921 --> 20:16.467 [JM]: "So we're just going to kill it and no more inexpensive Mac exclusive at Walmart." 20:16.787 --> 20:26.277 [JM]: They instead built a very worthy successor that is in some ways better, some ways worse, than that legacy M1 MacBook Air. 20:26.357 --> 20:29.644 [JM]: But the fact that they're innovating in that area at all, to me is fantastic. 20:30.085 --> 20:44.920 [JM]: And around the same time, they also introduced an M5 variant of the MacBook Air, the iPhone 17E, and they updated the iPad Air for the M4 system-on-a-chip. 20:44.900 --> 21:09.405 [JM]: And I don't have a lot to say about these more-or-less incremental updates to these three products, other than to say, as I always say, it just makes me happy to see that Apple is giving them attention, that they continue to do the best they can to keep their product line updated over time, that they don't let any particular variant 21:09.672 --> 21:11.714 [JM]: languish for too long. 21:12.294 --> 21:18.620 [JM]: Obviously there are notable exceptions like the Mac Mini that was left untouched for years. 21:18.740 --> 21:31.772 [JM]: The Mac Pro, which, which has still been left untouched for years, which may never be touched again, which is still selling for some unholy price that I dare not name. 21:31.832 --> 21:35.996 [JM]: And I think contains an M2 Ultra processor. 21:36.036 --> 21:36.977 [DJ]: Oh no. 21:37.217 --> 21:39.259 [JM]: Yeah. Which is just criminal. 21:39.319 --> 21:40.622 [JM]: No one should buy this. 21:40.702 --> 21:42.366 [JM]: It should not be for sale, but it is. 21:42.386 --> 21:52.588 [DJ]: No, someday the floodwaters will rise, and it'll just be poor John Siracusa left perched on, like, the one mountaintop still above the waves, clutching his Mac Pro. 21:52.788 --> 22:08.568 [JM]: He does seem to be the sole, lonely bannerman hoping and advocating for the return of a rejuvenated Mac Pro, but it doesn't seem like that's in the cards. 22:09.028 --> 22:22.525 [JM]: And as usual, I think part of the reason that I don't have a lot to say about these particular product line updates is that I'm generally not all that interested in this level of Apple's product line. 22:22.505 --> 22:27.235 [JM]: I'm more interested in the capabilities of a MacBook Pro, say, than a MacBook Air. 22:27.897 --> 22:34.551 [JM]: And same with the iPhone and same with an iPad, even though, generally speaking, I don't really do productive things with iPads. 22:34.952 --> 22:36.977 [JM]: But I think all three of these are great updates. 22:37.197 --> 22:37.999 [JM]: And if... 22:37.979 --> 22:47.451 [JM]: And for most people who are in the market for, say, a MacBook, an iPhone, and an iPad, I think all three of these are excellent mid-range choices. 22:47.992 --> 23:02.451 [JM]: One other bit of information that came out recently is that the Mac Mini and some servers that are being made by Apple to power Apple Intelligence are going to be made in the United States in Houston. 23:02.431 --> 23:19.697 [JM]: And this seems to be part of a broader push to manufacture more things outside of China, say, 23:19.677 --> 23:21.300 [JM]: to make certain things in the United States. 23:21.941 --> 23:33.739 [JM]: And it will be interesting to see if this is going to be some small part of their manufacturing or if it's a small part now and something that can expand over time. 23:34.300 --> 23:42.293 [JM]: I think that we probably won't see the day, at least in our lifetimes, where Apple is producing, say, 23:42.273 --> 23:45.016 [JM]: the majority of their products in the United States. 23:45.496 --> 23:59.371 [JM]: But even if they could get it from single-digit percentages, which I'm sure is where they're at, to say double-digit percentages, that would be a huge difference in terms of making more things in the United States, to the extent that that is one of their goals. 23:59.992 --> 24:11.284 [JM]: And one other thing that appeared in the headlines over the last week is that the 512-gigabyte Mac Studio has vanished from Apple's store. 24:11.264 --> 24:13.389 [DJ]: You're saying there was a daring theft, Justin? 24:14.331 --> 24:17.518 [DJ]: Who is this Mac Studio bandit and where can they be found? 24:18.561 --> 24:28.844 [JM]: I think the bandit is probably known as OpenAI, Anthropic, Google, Facebook, and Apple themselves, perhaps. 24:29.265 --> 24:30.167 [JM]: Oh, and Microsoft. 24:30.198 --> 24:32.041 [DJ]: So a criminal conspiracy is afoot. 24:32.402 --> 24:32.702 [JM]: Yes. 24:33.324 --> 24:35.688 [JM]: There seem to be several possible explanations of this. 24:35.728 --> 24:48.190 [JM]: The one that is being bantered about the most is the idea that, like everything else hardware related, the supply of RAM has become constrained because of 24:48.170 --> 24:58.607 [JM]: generative software companies like OpenAI tying up, say, 40% of global production in a single day, which doesn't leave a lot of RAM for everyone else. 24:58.627 --> 25:11.848 [JM]: I've seen other people say that a more likely explanation is that Apple themselves is trying to free up that production for, say, an M5-powered Mac Studio. 25:11.828 --> 25:28.194 [JM]: And it's also possible and maybe even probable that both things are true, that there is constrained supply and that Apple is trying to ensure that it has enough for the much higher margin Mac Studio relative to the Mac Mini. 25:28.535 --> 25:36.688 [JM]: So if Apple has to choose between shipping, say, a one terabyte RAM M5 Mac Studio that costs... 25:36.668 --> 25:47.042 [JM]: $6,000 versus a much lower margin, but still very expensive 512 gigabyte Mac Mini, it's pretty clear which of those two they're going to choose. 25:47.863 --> 26:05.967 [JM]: But I think the thing that occurs to a lot of us is what does this mean in the broader context of we computer nerds being able to do what we've always been able to do, which is to find RAM and processors and drives and 26:06.183 --> 26:07.873 [JM]: and put them in a computer, 26:08.191 --> 26:10.894 [JM]: or upgrade computers we already have with those components. 26:11.214 --> 26:12.756 [JM]: These are things we've always been able to do. 26:13.256 --> 26:23.206 [JM]: But when companies buy up all of those components and then try to sell you the services that run on those components, then at some point you don't have a choice anymore. 26:23.246 --> 26:29.432 [JM]: You don't get to choose between, okay, I'm going to build my own computer or buy my own computer. 26:29.812 --> 26:37.540 [JM]: Because this isn't just about the exceedingly few of us who like buying computer parts and slapping them together and building computers. 26:37.520 --> 26:55.439 [JM]: Even if you're the kind of person that has no interest in that and all you want to do is just buy a prebuilt computer, even that is looking like those days could be numbered because all of those components are being allocated in data centers for companies that are trying to sell you the services that run on those computers. 26:55.840 --> 27:06.531 [JM]: And it kind of reminds me of a somewhat recent development in the United States real estate market where private equity companies figured out 27:06.511 --> 27:09.535 [JM]: that they could just buy lots of houses. 27:09.915 --> 27:31.742 [JM]: They could just buy a bunch of residential houses and then rent them to people who increasingly couldn't afford to buy the houses themselves because they're not multi-billion dollar corporations who, by buying said houses, drove up the prices of those houses, making it even less affordable for the average person to buy them. 27:31.722 --> 27:41.035 [JM]: So those people are priced out of buying the house themselves and instead are more-or-less forced to rent the house instead. 27:41.416 --> 27:44.641 [JM]: And I can't help but feel that's a lot like what's happening here. 27:44.681 --> 28:01.645 [JM]: And the implications that it has for our control over our own data and privacy is really rather significant and something that doesn't keep me up at night at this point because we still can buy computers, but... 28:01.625 --> 28:05.910 [JM]: I do wonder whether that's something we'll be able to do a few years from now. 28:06.391 --> 28:13.239 [DJ]: None of us should want a world where we are essentially the supplicants of giant corporations. 28:13.860 --> 28:24.072 [DJ]: So to the extent that we can keep making choices about what we support, you know, you typically bring up that notion of "Vote with your wallet". 28:24.052 --> 28:35.972 [DJ]: It is important to find the outliers, which is why I've become such a big fan of companies like Framework that are all about designing and selling highly repairable computers. 28:36.573 --> 28:44.406 [DJ]: And I worry about how readily companies like that will be able to stay in business when it gets harder and harder to acquire computer components. 28:44.386 --> 28:51.456 [DJ]: But for now, you know, I feel like all I can do is continue to support the things that I support. 28:51.556 --> 28:58.607 [DJ]: So there are things there, you know, hardware makers like that, that are alternatives to the giant tech companies. 28:59.127 --> 29:02.412 [DJ]: And in terms of using large language model powered 29:02.392 --> 29:11.605 [DJ]: software, I continue to be interested in open-source or open-weight models, large language models that you can run on your own hardware. 29:12.106 --> 29:26.307 [DJ]: And, you know, I think, I think we should try to invest in having our own hardware and running our own software on them because really like a world where the only computing that exists is you interact with ChatGPT through your iPhone and 29:26.287 --> 29:29.593 [DJ]: I don't think that's a great world, not from a technology perspective. 29:29.613 --> 29:40.891 [DJ]: It's not as good, in my opinion, as the world where you can run any of the vast universe of software on any of the vast universe of hardware that you like. 29:41.252 --> 29:44.137 [DJ]: That diversity is better for everybody. 29:44.658 --> 29:47.282 [DJ]: So we have to continue supporting it. 29:47.498 --> 29:55.831 [JM]: You mentioned running large language models locally on our own hardware, and that is something that you're right, we can absolutely do. 29:55.851 --> 30:05.507 [JM]: And there are models that come out with open weights, like Qwen 3.5 is a recent example, and they get better and better 30:05.487 --> 30:06.288 [JM]: over time. 30:06.748 --> 30:18.819 [JM]: But the reality is that they still lag the state-of-the-art hosted frontier models, like Claude, ChatGPT, etc, often by a year or more. 30:19.300 --> 30:33.773 [JM]: And if increasingly the RAM that we need and the processing power we need to run them locally is being sucked into the vortex of these venture-capital-fueled builders of so-called frontier models, then in the end, 30:33.753 --> 30:49.897 [JM]: it seems like unless we can somehow figure out a way of avoiding it, the future is one in which we are renting computing power from those companies and accessing them from a locked-down device that we can't really operate on its own without that service. 30:50.417 --> 30:57.027 [JM]: And in the end, the computing gets mediated and captured by these hosted companies. 30:57.007 --> 31:04.499 [JM]: Models that are going to obscure the sources of their data, and their training, and the process by which they're trained, from us. 31:04.959 --> 31:05.580 [JM]: But I agree with you. 31:05.600 --> 31:15.015 [JM]: The more that we can do to try to retain our data sovereignty and not become overly reliant on hosted services... 31:15.516 --> 31:19.622 [JM]: Those are the cards that we can play and we should play them. 31:19.602 --> 31:28.819 [JM]: Speaking of Qwen 3.5, I've been playing with Qwen 3 Coder Next and Qwen 3.5 models since they were released. 31:29.440 --> 31:42.403 [JM]: And they are being touted as at the same level as Sonnet 4.5 from Claude, which I think came out about a year ago, maybe a little bit less than a year ago. 31:42.586 --> 32:00.450 [JM]: But it seems to me that the open-weight models play these benchmark games and they seem like when they're released, they come with promises of being as good as something that was state-of-the-art a few months ago, but then they disappoint in actual use. 32:01.111 --> 32:11.125 [JM]: And indeed, in my experience, these latest releases are good, but they do not perform at Sonnet 4.5 levels, for example. 32:11.105 --> 32:13.387 [JM]: But they are impressive for open-weight models. 32:13.728 --> 32:21.075 [JM]: And it's amazing what you can do with self-hosted models today versus what you could do just a few months ago. 32:21.516 --> 32:36.792 [JM]: I do think it's important that one doesn't believe the hype that they deliver the Sonnet 4.5 level results that people sometimes say they can because you'll probably be disappointed once you get into anything that's reasonably complex. 32:36.856 --> 32:58.238 [JM]: And it also seems that the leadership, and by leadership, I mean the technical leadership, the key people that have produced such surprising, significant improvements in these open weight models, haven't been too particularly jazzed about things going on at the very top level of their organizations. 32:58.498 --> 33:01.201 [JM]: And it seems like there may have been some key departures. 33:01.402 --> 33:01.902 [JM]: So... 33:01.882 --> 33:13.879 [JM]: I am hopeful that either they decide to go back to that organization and continue to work on producing great releases of Qwen or that they strike out on their own and make something that's even better. 33:14.340 --> 33:21.831 [JM]: Because we, as you said, we need to see lots of continued innovation in the realm of open-weights models. 33:21.811 --> 33:39.735 [JM]: Another thing that I saw that is related to this whole topic of computing components becoming scarce and more expensive is that it's affecting not just you and me as consumers, but also the other companies that aren't producing, say, generative software models, 33:39.715 --> 33:50.939 [JM]: but that are producing more traditional services like hosted servers, are also finding it hard to get the components they need at a price that aligns with their cost structure. 33:51.620 --> 33:56.290 [JM]: And Hetzner, which is a company headquartered in Germany and which 33:56.270 --> 34:06.372 [JM]: offers rented servers and cloud services has announced that they're increasing their prices between 30 and 40%, which is a huge increase. 34:07.133 --> 34:16.253 [JM]: And which, for me as a Hetzner customer, is significant in terms of thinking, okay, well, I may have spawned a 34:16.233 --> 34:19.680 [JM]: hobby server here or created this fun little project over there. 34:19.720 --> 34:27.457 [JM]: But with these price hikes, it leads me to ask the question, okay, how many of these things do I still want to keep? 34:27.797 --> 34:32.487 [JM]: How many of these do I decide to retire just to keep the cost down? 34:32.467 --> 34:37.474 [JM]: And these are all things that I am self-hosting because I love self-hosting. 34:37.895 --> 34:50.614 [JM]: And this feeds back into this question of how long can we continue to self-host when just like your average United States resident who's trying to buy a home? 34:51.135 --> 34:54.880 [JM]: At what point do we get priced out of the self-hosting market... 34:54.860 --> 35:03.695 [JM]: and are forced to use the mediated, curated service provided by the popular hosted generative software models? 35:04.015 --> 35:20.563 [JM]: And I hope that this is the kind of thing that is somewhat temporary, and there aren't too many permanent seismic shifts in the industry that companies will say, okay, there's all this demand, and they start producing more RAM and more drives and more 35:20.543 --> 35:24.770 [JM]: these other components so that those prices can come back to a reasonable level. 35:25.452 --> 35:31.642 [JM]: And Hetzner will be able to, for example, not have to resort to increasing their prices or maybe even reduce them. 35:32.043 --> 35:40.678 [JM]: Maybe we can just buy a few thousand MacBook Neos and string them all together and create our own makeshift server farm. 35:40.658 --> 35:48.389 [JM]: And then people can host their WordPress blogs on our MacBook-Neo-powered server farm. 35:48.709 --> 35:54.878 [DJ]: I think that'll work great as long as we don't accidentally plug those cables into that USB 2 port. 35:54.898 --> 35:55.980 [JM]: Definitely don't do that. 35:56.701 --> 35:58.203 [JM]: All right, everyone, that's all for this episode. 35:58.223 --> 35:58.984 [JM]: Thanks for listening. 35:59.024 --> 36:04.171 [JM]: You can find me on the web at justinmayer.com and you can find Dan on the web at danj.ca. 36:04.592 --> 36:08.758 [JM]: Reach out with your thoughts about this episode via the Fediverse at justin.ramble.space.