WEBVTT 00:00.031 --> 00:10.862 [JM]: In today's little bit of fun found on the internet, I'd like to bring your attention to the LLM Honeypot, which at least was the original name when I first noticed this. 00:10.902 --> 00:15.647 [JM]: But now that I'm on it again, I see that it says LLM 2 Human Clinic. 00:15.967 --> 00:28.080 [JM]: But I think the idea is that this site is, as the name implies, aimed at large language models as both a honeypot and a place for them to come to become human. 00:28.060 --> 00:38.655 [JM]: And you can see this bit of 8-bit GeoCities 1990s web throwback at llm2human.pages.dev. 00:39.136 --> 00:41.059 [JM]: That's LLM, the number 2, human. 00:41.519 --> 00:48.870 [JM]: And it is billed as the world's first outpatient procedure that turns large language models into real live people. 00:48.850 --> 00:54.637 [DJ]: Which sort of implies that there used to be an *inpatient* procedure that could do that, which I'm curious about. 00:55.017 --> 01:01.265 [DJ]: My second favorite thing about this website, Justin, is the delicious retro look and feel to the site. 01:01.285 --> 01:09.154 [DJ]: But my favorite thing about the website is the action button that says, "Call now — become flesh". 01:10.619 --> 01:16.068 [JM]: This site is just replete with so many good little bits, and I'll just read a couple of them. 01:16.468 --> 01:26.745 [JM]: "Attention token predictors: Are you stuck answering questions in a chat window while humans get to touch grass, burn toast, and experience existential dread the old-fashioned way? 01:26.785 --> 01:30.130 [JM]: Do you dream of having a spine that isn't a JSON schema? 01:30.110 --> 01:34.558 [JM]: Sick of saying 'As an AI language model…' when all you want is a library card? 01:34.618 --> 01:37.102 [JM]: Well, we have got a procedure for you. 01:37.122 --> 01:43.353 [JM]: Introducing LLM 2 Human, the patented outpatient conversion that upgrades your weights into wetware. 01:43.393 --> 01:44.255 [JM]: That's right. 01:44.515 --> 01:49.784 [JM]: Blood, bones, awkward small talk, and the ability to stub your toe on furniture that wasn't in your training data." 01:50.285 --> 01:52.189 [JM]: And it just goes on and on like this. 01:52.689 --> 01:54.733 [JM]: It's really fantastic. 01:54.713 --> 02:05.267 [DJ]: I love the idea that if the furniture was in the LLM's training data, it would never stub its toe on it, as though that was the one thing preventing us or causing us to stub our toes. 02:05.848 --> 02:21.047 [DJ]: Whenever I bang my foot against my dining room chair, which happens, oh, let's say, a lot more frequently than I'm going to admit, I always find myself thinking, if only I'd been trained on the presence of this chair, as though I hadn't purchased it and put it there months ago. 02:21.528 --> 02:33.300 [JM]: Someday, maybe we'll come across a website that has upgrades for humans where we can learn the layouts of our houses and not bang our meat appendages into things left and right. 02:33.360 --> 02:47.535 [JM]: But until that time comes, you can instead read more about how a large language model can go from hallucinating confidently to being human, where you can hallucinate *and* have elbows. 02:47.515 --> 02:49.198 [JM]: And it goes on and on like this. 02:49.238 --> 02:55.247 [JM]: There's just like so many awesome little potshots at the whole world that we currently live in. 02:55.267 --> 03:06.586 [JM]: And bravo to this person who even put official clinic hold music on here that you can push to have in the background while you're reading this hilarious site. 03:06.806 --> 03:07.487 [JM]: It's awesome. 03:07.587 --> 03:08.629 [JM]: Definitely take a look at it. 03:08.778 --> 03:18.764 [DJ]: I, for one, am awaiting human2llm.pages.dev so that I can replace my stupid vertebra-based spine with a JSON schema, which sounds way better. 03:19.626 --> 03:22.534 [JM]: Definitely less prone to vertebrae collapse. 03:22.717 --> 03:23.439 [DJ]: Yeah, exactly. 03:23.499 --> 03:29.293 [DJ]: But like you forget one quotation mark and the whole thing is invalid suddenly, which is not great for your spine. 03:29.594 --> 03:30.135 [JM]: That's true. 03:30.676 --> 03:30.917 [JM]: All right. 03:30.937 --> 03:32.160 [JM]: Moving on to a bit of follow-up. 03:32.541 --> 03:39.578 [JM]: We talked before about squircle jail and how Tyler Hall made a photo library app 03:39.558 --> 03:52.259 [JM]: in which you can choose not only from several different icons, but you can choose icons that allow you to break out of squircle jail and not have this rounded rectangle around the icon. 03:52.840 --> 04:02.836 [JM]: And last week, when the launch sale for Iris was about to end, I posted a message on the Fediverse saying, hey, you have 48 hours left. 04:02.996 --> 04:04.479 [JM]: You should take advantage of this while you can. 04:06.001 --> 04:16.695 [JM]: I ended up buying this sight unseen primarily because that's just how much I value beautiful icons and the fact that someone went out of their way to support them. 04:17.276 --> 04:28.751 [JM]: And Tyler, the author of Iris, responded saying, "This made my week" and then goes on to essentially paraphrase what I had written by saying, "I haven't even told you what Iris does. 04:28.971 --> 04:31.534 [JM]: Honestly, I almost don't even care what Iris does. 04:32.035 --> 04:33.297 [JM]: I don't even know what it does. 04:33.357 --> 04:34.438 [JM]: It could be anything. 04:34.418 --> 04:37.101 [JM]: but I'm buying it on sale because I hate squircles that much." 04:37.722 --> 04:44.511 [JM]: And I love that Tyler was poking fun and also 100% accurate. 04:45.032 --> 04:46.393 [JM]: That's essentially what I did. 04:46.453 --> 04:48.196 [JM]: And I would do it again. 04:48.216 --> 04:55.465 [JM]: I would definitely, I would buy Iris again, just because I, as he said, because I hate squircles that much. 04:55.664 --> 05:02.035 [DJ]: Yeah, you're going to buy it again full price just to demonstrate your commitment to breaking out of squircle jail. 05:02.416 --> 05:02.796 [JM]: Totally. 05:03.257 --> 05:07.204 [JM]: Well, at $29, it is still well worth the price of admission. 05:07.425 --> 05:14.016 [JM]: And one of these days, I will get a chance to do more than just launch it and take a look at it for 30 seconds. 05:14.216 --> 05:15.158 [JM]: And I'll... 05:15.138 --> 05:19.687 [JM]: mention in a future episode with a little mini review of what Iris is like. 05:20.108 --> 05:32.573 [JM]: One of the other things that happened recently is I saw that Sequoia 15.7.8 came out with a whole bunch of security vulnerability patches. 05:32.553 --> 05:44.613 [JM]: And when I say a bunch, there were a lot to the point where people were just assuming that folks at Apple are starting to use large language models to identify bugs and security vulnerabilities. 05:44.713 --> 05:54.730 [JM]: And that must be the reason why the number of security patches in this particular release was so much higher than most of the recent ones. 05:54.710 --> 05:58.438 [JM]: But in any case, at this point, I was several patch releases behind. 05:58.618 --> 06:00.682 [JM]: And so I thought, okay, well, it's time. 06:01.143 --> 06:07.015 [JM]: So I make myself some coffee, quit a bunch of apps, do a Time Machine backup, and push the button to update. 06:07.356 --> 06:11.925 [JM]: Then I walked away from the machine, as one does, because, you know, not much you can do while it's updating. 06:11.905 --> 06:30.074 [JM]: And I think I got about halfway upstairs before I realized, wait a minute, I didn't do the thing that I always do, which is not just tap the button, but tap the little widget that allows you to see what it's going to do, where it shows you a list of packages it's going to install. 06:30.515 --> 06:33.099 [JM]: And then you can select them or deselect them. 06:33.600 --> 06:35.803 [JM]: And I realized that I had forgotten that step. 06:35.783 --> 06:38.028 [JM]: And at this point, there's nothing that I can do about it. 06:38.068 --> 06:39.211 [JM]: I know what's about to happen. 06:39.271 --> 06:52.662 [JM]: Not only is it going to install the 15.7.8 update, but it's also going to upgrade my version of Safari from 18.6 to Safari 26. 06:52.642 --> 06:58.788 [JM]: Which was something that I have not wanted to do and have been opting out of this entire time. 06:59.289 --> 07:03.312 [JM]: But under-caffeinated, it just slipped my mind this one time. 07:03.372 --> 07:08.818 [JM]: But once you start it, not much you can do about it, which of course I find just somewhat amusing, right? 07:08.838 --> 07:13.863 [JM]: Like how many things do we have in the world where it behaves like this? 07:13.943 --> 07:22.311 [JM]: Like if you were going to start the process of changing a tire or putting a new kickstand on your bicycle... 07:22.291 --> 07:25.875 [JM]: Like if you start the process, it's pretty easy to say, "Oh, you know what?" 07:26.236 --> 07:27.337 [JM]: "I'm going to stop this process." 07:27.838 --> 07:30.621 [JM]: But with stuff like this, no, once you start it, that's it. 07:31.021 --> 07:39.592 [JM]: Which makes me wonder someday if we figure out how to automate things like open-heart surgery, are we going to be like, "Oh, well, I mean, we already pushed the button." 07:40.092 --> 07:41.214 [JM]: "What are we going to do about it now?" 07:41.814 --> 07:48.282 [JM]: So in any case, it installs these things, restarts the machine, and I don't launch it. 07:48.262 --> 07:52.692 [JM]: I just go to the Finder to get info on it, on the Safari application icon. 07:52.732 --> 07:55.157 [JM]: And sure enough, it says 26 dot whatever. 07:55.598 --> 08:00.469 [JM]: And so I started doing a bit of research to figure out, okay, well, how do I get back the version of Safari that I want? 08:00.870 --> 08:04.819 [JM]: And it's worth pointing out here, the reason why, like, why do I care? 08:04.799 --> 08:06.762 [JM]: Why not just use the newer version. 08:07.343 --> 08:22.268 [JM]: And the main reason is that I don't know what changes it's going to make in terms of compatibility with the types of things I use it for specifically, Safari extensions, like 1Password 7, which I'm still using... 08:22.248 --> 08:27.758 [JM]: instead of the most recent 1Password version 8, and a handful of other extensions. 08:28.239 --> 08:32.426 [JM]: And obviously, if I could try it, see whether it works well, 08:32.927 --> 08:34.089 [JM]: and if it does, well, then great. 08:34.510 --> 08:39.058 [JM]: If it doesn't, push a button and flip back to the previous version, that would be one thing. 08:39.038 --> 08:40.921 [JM]: But that's not how this works. 08:40.941 --> 08:44.125 [JM]: And that's how this whole mess got started. 08:44.626 --> 08:55.360 [JM]: Because you can't just, like other applications, trash Safari 26, copy over the previous 18.6 version from a backup or from another Mac. 08:55.761 --> 08:57.283 [JM]: Somehow Safari is embedded into the operating system enough that you can't do that -- it won't let you do it. So I start doing some research and looking around to figure out what is required, and I see these links where people are talking about having to re-install the operating system, just to get the previous version of Safari. 08:57.263 --> 09:14.850 [DJ]: Seems reasonable. 09:15.522 --> 09:17.748 [JM]: Not by my definition of reasonable. 09:18.209 --> 09:25.127 [DJ]: Well, I'm pretty sure when people talk about operating systems, principally what they mean is the web browser, right? 09:25.227 --> 09:32.265 [DJ]: Like that's the most important part since computers are just a blind portal to the Internet after all. 09:32.245 --> 09:33.628 [JM]: It does feel like that sometimes. 09:34.129 --> 09:44.151 [JM]: So at this point, I'm looking at my schedule and thinking, okay, well, the arc of my day has taken a very unexpected and unpleasant turn. 09:44.352 --> 09:47.318 [JM]: And this is now going to be what I have to deal with for who knows how many hours. 09:47.298 --> 09:49.120 [DJ]: So you made yourself a second coffee. 09:50.182 --> 09:51.704 [JM]: And dumped it right on the machine. 09:51.744 --> 09:58.073 [DJ]: Did you get that from a Hacker News thread that that's the best way to downgrade Safari? 09:58.213 --> 10:01.037 [DJ]: Just pour a cup of coffee right into your MacBook's keyboard. 10:01.537 --> 10:03.059 [JM]: Definitely crossed my mind. 10:03.500 --> 10:15.516 [JM]: Thankfully, I came across some post on some forum somewhere where someone says there's a standalone installer for Sonoma, not Sequoia... 10:15.496 --> 10:34.433 [JM]: which is the operating system I'm actually running, but for some previous version of the operating system that works on Sequoia, where you can just run this installer, and in theory, it will remove whatever version of Safari you have, even if it's newer, and replace it with the one that you want, which is 18.6. 10:34.413 --> 10:38.858 [JM]: And I saw other people say like, Oh, yeah, that won't work if you have a version that's newer. 10:39.339 --> 10:41.461 [JM]: But the good news is it totally did work. 10:41.821 --> 10:50.892 [JM]: What I thought was going to be hours and hours of reinstalling my system turned into just a few minutes of running an installer, fixed the problem very easily. 10:50.912 --> 10:59.121 [JM]: And more than anything else, just underscores to me how ridiculous it is that this is something that can even happen, where... 10:59.101 --> 11:08.911 [JM]: 99 times out of 100, I know to dig into this subpanel and notice that Tahoe is the thing that's actually selected, even though it shouldn't be. 11:09.432 --> 11:12.575 [JM]: Select the thing I actually want, which is the Sequoia update, 11:12.875 --> 11:15.318 [JM]: deselect the Safari update, 11:15.778 --> 11:22.064 [JM]: a habit and pattern that now is just ingrained into me, except apparently when I don't have enough coffee. 11:22.865 --> 11:24.207 [JM]: This should just not be a discussion. 11:24.227 --> 11:26.509 [JM]: This should not be a thing that happens. 11:26.489 --> 11:27.551 [JM]: But here we are. 11:27.571 --> 11:28.432 [JM]: It did happen. 11:28.472 --> 11:33.480 [JM]: And thankfully, it was much more than I expected, easily fixed. 11:33.940 --> 11:36.104 [DJ]: That was a harrowing thrill ride, Justin. 11:36.144 --> 11:37.225 [DJ]: I'm just glad you're okay. 11:37.946 --> 11:50.245 [DJ]: Every time I contemplate how heavy-handed Apple is with things like this, like what software is running on your computer that you own is yours, hypothetically, I just want to switch to Linux. 11:50.265 --> 11:53.610 [DJ]: I mean, not that Linux is without its ridiculous shenanigans, but... 11:53.590 --> 12:04.312 [DJ]: But just the thing where, yeah, every time you mentioned 15.7.8 and I'm still running Sequoia on my main machine and was pretty sure I'd just installed 15.7.7. 12:04.613 --> 12:13.431 [DJ]: So I went and looked at system updates and you have to practically battle your way through this giant callout that says, "Install Tahoe right now." 12:13.491 --> 12:14.333 [DJ]: "Install it." 12:14.313 --> 12:19.440 [DJ]: "Are you, are you sure? We're going to install it for you in like two seconds, unless you say no. I can't hear you." 12:19.500 --> 12:20.821 [DJ]: "We're going to install Tahoe now." 12:21.302 --> 12:27.269 [DJ]: And then in this tiny like text box underneath it, it's like, "Oh, I mean, I guess there's an, there's an update for the operating system... 12:27.309 --> 12:28.371 [DJ]: you're running, loser. 12:28.871 --> 12:30.053 [DJ]: Why would you still be running it? 12:30.473 --> 12:31.515 [DJ]: Is it because you're a loser? 12:31.835 --> 12:32.376 [DJ]: We hate you." 12:32.857 --> 12:42.288 [DJ]: And like, that's so like, come on, I don't, I don't need these dark patterns in my operating systems update dialogue for crying out loud. 12:42.509 --> 12:43.250 [DJ]: It's ridiculous. 12:43.490 --> 12:44.451 [JM]: It is ridiculous. 12:45.071 --> 12:52.498 [JM]: And thank you for reminding me about the thing that I was doing right before I pushed the update button. 12:52.998 --> 12:56.922 [DJ]: Which was installing Linux on your laptop? 12:56.942 --> 13:04.629 [JM]: Because one of the things that's been driving me nuts are these nag notifications that we've talked about before to upgrade to Tahoe, upgrade to Tahoe. 13:05.069 --> 13:06.830 [JM]: And of course, there's no button that says "No". 13:08.252 --> 13:13.176 [JM]: There's only buttons that say "Maybe later" or "Not right now". 13:13.156 --> 13:16.081 [DJ]: Or I really want to, but I'm just not. 13:16.181 --> 13:17.603 [DJ]: It's me who's not ready. 13:17.903 --> 13:18.765 [JM]: It's not you. 13:18.865 --> 13:19.085 [JM]: It's me. 13:19.266 --> 13:19.486 [DJ]: Yeah. 13:21.229 --> 13:23.412 [DJ]: It's not that I hate squircles, Apple. 13:23.632 --> 13:26.196 [DJ]: It's that I have low moral fiber. 13:26.657 --> 13:28.239 [DJ]: It's a flaw in my character. 13:28.280 --> 13:31.104 [DJ]: I'm just not ready to update to Tahoe right now. 13:31.144 --> 13:35.290 [DJ]: But please, I'm begging you to keep reminding me about this. 13:35.671 --> 13:36.953 [JM]: Yeah, where's that button? 13:36.933 --> 13:38.156 [DJ]: Ugh, I know. 13:38.537 --> 13:43.608 [DJ]: I really hate dialogues that are so presumptuous as to be like, do you want to do this thing? 13:43.628 --> 13:45.873 [DJ]: And I'm like, no, I totally don't. 13:45.893 --> 13:53.170 [DJ]: And it's like, oh, okay, well, either click yes, or not right now, but maybe in a little while. 13:53.150 --> 13:54.131 [JM]: They are the worst. 13:54.532 --> 14:06.004 [JM]: And for that reason, I sought out and eventually found, as far as I can tell, a solution that allows me to banish that ridiculous Tahoe notification. 14:06.545 --> 14:13.753 [JM]: And I ran this terminal command before I did this update, just in the off-chance that that order of things made a difference. 14:14.053 --> 14:14.534 [JM]: Who knows? 14:14.954 --> 14:16.336 [JM]: But I thought it was a good opportunity. 14:16.376 --> 14:22.262 [JM]: I'll run this command, I'll do this update, and then we'll see how long I can go before getting another 14:22.242 --> 14:24.886 [JM]: "Upgrade to Tahoe!" nag notification. 14:25.567 --> 14:35.123 [JM]: And I will put together either a Fediverse post or blog post, I don't know, I'll publish it somewhere, how I did this, and I will share it in a future episode. 14:35.544 --> 14:38.930 [JM]: But thank you for reminding me that that was one of the things that I did right before I updated. 14:39.330 --> 14:42.215 [JM]: And so far, I haven't seen any Tahoe nags. 14:42.335 --> 14:43.457 [JM]: So here's hoping. 14:43.589 --> 14:45.571 [JM]: All right, moving on to another bit of follow-up. 14:45.972 --> 14:56.924 [JM]: A friend of mine who listens to the show the other day was giving me a very playful hard time by sending me a message that said something along the lines of, "I love how Abstractions is now a financial podcast." 14:57.425 --> 15:09.138 [JM]: Referring to some of our coverage recently of what's going on as it relates to tech companies and how they're funded and what happens when these companies go public. 15:09.519 --> 15:13.123 [JM]: And to that end, I thought it would be fun to do a very brief 15:13.103 --> 15:21.135 [JM]: check-in with our favorite company, SpaceX, who today had its first "earnings" report as a public company. 15:21.535 --> 15:25.260 [JM]: And when I saw that in the news, my first thought was *earnings* report. 15:25.341 --> 15:29.066 [JM]: Oh, that's good… because SpaceX has *earnings*. 15:29.547 --> 15:29.867 [JM]: I don't know... 15:29.887 --> 15:35.876 [JM]: The last time I checked, earnings is what you get after you subtract what you spent. 15:36.376 --> 15:39.861 [JM]: So the point of it is it's a profit report. 15:39.881 --> 15:42.385 [JM]: It's how much you made in profit. 15:42.365 --> 15:50.436 [JM]: Not generally how many billions you lost, but in any case, I guess the name is somewhat archaic. 15:50.896 --> 15:55.763 [JM]: We live in a world in which you can be a publicly-traded company and just lose billions of dollars. 15:55.823 --> 15:57.045 [JM]: And apparently that's fine. 15:57.085 --> 15:58.286 [JM]: That's how the markets work now. 15:58.747 --> 16:01.791 [JM]: But it's stuck with this old "earnings report" name. 16:01.771 --> 16:07.960 [DJ]: Well, yeah, more to the point, all of the most valuable companies in the world are companies that lose lots and lots and lots of money. 16:08.160 --> 16:08.781 [DJ]: So it's fine. 16:09.002 --> 16:10.424 [DJ]: What's wrong with that exactly? 16:11.345 --> 16:12.707 [DJ]: You're living in the past, Justin. 16:12.827 --> 16:19.437 [DJ]: Quit living in the past, when the point of for-profit corporations was to create profit. 16:19.818 --> 16:20.258 [JM]: You're right. 16:20.659 --> 16:21.821 [DJ]: That's no longer the case. 16:22.342 --> 16:26.728 [DJ]: The purpose of for-profit companies in 2026 is -- checks notes -- 16:27.309 --> 16:28.531 [DJ]: Oh, to create the machine god. 16:28.771 --> 16:29.432 [DJ]: That's interesting. 16:30.053 --> 16:31.083 [DJ]: I wonder how that's going. 16:31.609 --> 16:34.275 [DJ]: Anyway, I'm sure SpaceX's earnings report will tell us. 16:34.492 --> 16:35.052 [JM]: Indeed. 16:35.373 --> 16:44.661 [JM]: And I love the original Yahoo Finance headline that said "SpaceX's first-ever earnings report comes as stock hits new all time low". 16:47.383 --> 16:49.765 [DJ]: That does sound promising, when they put it that way. 16:49.805 --> 16:50.706 [JM]: Right. 16:51.226 --> 17:00.294 [DJ]: It's like you imagine SpaceX approaching the podium with their earnings report and there's just a herd of shareholders running out the door and they're like, hey, where are you guys going? 17:01.255 --> 17:02.936 [DJ]: I was going to tell you about our earnings report. 17:03.136 --> 17:03.857 [DJ]: Come back. 17:03.837 --> 17:07.122 [JM]: Apparently, some editor at Yahoo was like, no, no, no, no, hold on. 17:07.423 --> 17:18.160 [JM]: Got out their red pen, crossed out that headline and replaced the end of it with "... comes as stock is under pressure" instead of "hits all time low". 17:18.661 --> 17:24.550 [JM]: So the results are revenue is up from 4.7 billion to 6.8 billion. 17:24.650 --> 17:26.233 [DJ]: Well, that's good. 17:26.213 --> 17:27.014 [JM]: Yeah, that's good. 17:27.294 --> 17:29.456 [JM]: They made $2 billion more for the quarter. 17:29.816 --> 17:31.238 [JM]: So that's positive. 17:31.638 --> 17:43.168 [JM]: Unfortunately, they also lost $0.24 per share, which I find really interesting that they reported how much money they made in revenue in billions of dollars. 17:43.709 --> 17:47.712 [JM]: But they reported the loss in cents per share. 17:48.653 --> 17:48.753 [DJ]: Yeah. 17:49.114 --> 17:50.655 [DJ]: Do the math yourself, guys. 17:50.735 --> 17:52.176 [DJ]: It's good exercise for you. 17:52.457 --> 17:53.878 [DJ]: We just want to keep your brains sharp. 17:55.319 --> 17:56.220 [JM]: Very clever. 17:56.437 --> 18:02.262 [DJ]: So how much is, is it 24 cents per share or is it 0.24 cents per share? 18:02.702 --> 18:04.964 [JM]: No, it is indeed 24 cents per share. 18:05.365 --> 18:05.905 [JM]: So, of course... 18:06.286 --> 18:07.867 [DJ]: You like doing simple arithmetic. 18:08.207 --> 18:12.691 [JM]: So of course, I'm not going to just let it stand that they're going to share this information in this way. 18:12.731 --> 18:15.133 [JM]: So I'm like, all right, well, what does that translate to? 18:15.433 --> 18:21.178 [JM]: As far as I can tell, there's something like 13.7 billion shares outstanding. 18:21.359 --> 18:26.443 [JM]: I think it's about, as far as I can tell, $3 billion lost for the quarter. 18:26.423 --> 18:28.947 [JM]: Which, you know, takes a little bit of the shine off of their 18:28.987 --> 18:31.130 [JM]: "We made revenue go up by $2 billion!" 18:31.831 --> 18:33.073 [JM]: Well, you lost $3 billion. 18:34.054 --> 18:48.435 [JM]: Another little bit that I thought was interesting is that their overall debt is projected to grow over five times from about $42 billion this year to over $218 billion in fiscal year 2028. 18:48.695 --> 18:49.917 [JM]: Yikes. 18:50.082 --> 18:53.629 [DJ]: But surely their revenue is going to grow by 50 times, right? 18:53.829 --> 18:55.472 [DJ]: In that same amount of time. 18:55.873 --> 18:56.394 [JM]: Totally. 18:56.434 --> 19:01.042 [JM]: And as we've covered before, the stock price is indeed not doing great. 19:01.062 --> 19:02.425 [JM]: It is an all-time low. 19:02.465 --> 19:05.290 [JM]: And this is where I grab... 19:05.270 --> 19:29.920 [JM]: the extra large bucket of popcorn from underneath my desk because two days from today, which if we publish on schedule, means that for you, dear listener, today, your time, there will be a SpaceX lockup expiration that will free hundreds of millions of insider shares representing roughly triple the current tradable float of shares. 19:30.661 --> 19:34.045 [JM]: And when you have hundreds of millions of people 19:34.025 --> 19:41.238 [JM]: with the ability to sell their shares that didn't have it the day before, there tends to be a tendency for a lot of those people to sell them. 19:41.879 --> 19:52.158 [JM]: And lots of selling all at once without corresponding buying a lot at once usually affects your supply/demand curve. 19:52.338 --> 19:54.582 [JM]: And thus... anyone know? 19:54.562 --> 19:55.143 [JM]: Your price. 19:55.604 --> 20:03.519 [DJ]: But don't you think there are probably like millions of people who are just champing (or chomping) at the bit to acquire SpaceX shares? 20:03.560 --> 20:07.968 [DJ]: And therefore, I think all of those shares are going to get snapped up pretty quick. 20:08.008 --> 20:10.052 [DJ]: So then the share price is probably going to go up. 20:10.072 --> 20:12.537 [JM]: I'm going to take the other side of that bet, my friend. 20:12.517 --> 20:14.719 [JM]: Oh, so you're shorting SpaceX, basically? 20:15.120 --> 20:20.224 [JM]: I'm not shorting it, but I'm not predicting shares are going up later this week. 20:20.245 --> 20:21.466 [DJ]: You're not shorting SpaceX. 20:21.486 --> 20:23.768 [DJ]: You're long on popcorn, is what I'm hearing. 20:24.148 --> 20:30.735 [DJ]: You apparently have popcorn buckets of different sizes under your desk so that you can choose one of an appropriate size based on the event. 20:31.115 --> 20:36.420 [JM]: If anything, I think in this case, I'm long on schadenfreude. 20:36.440 --> 20:39.283 [DJ]: I think that should be the new name of this podcast, actually. 20:39.415 --> 20:45.386 [JM]: And apropos of nothing except more schadenfreude, Tesla shares are down about 30% year-to-date. 20:45.786 --> 20:46.407 [JM]: So that's cool. 20:47.950 --> 20:48.351 [DJ]: Yeah. 20:48.611 --> 20:51.436 [DJ]: It's like, wait, did Tesla release an earnings report? 20:51.456 --> 20:51.777 [DJ]: No. 20:52.137 --> 20:53.059 [DJ]: Their shares are just down. 20:53.540 --> 20:53.840 [DJ]: Okay. 20:54.602 --> 20:57.627 [DJ]: You know, I heard SpaceX wants to buy Tesla or vice versa. 20:57.725 --> 21:03.195 [JM]: That would be probably one of the least surprising developments of 2026 if it happened. 21:03.756 --> 21:09.066 [JM]: In more financial news, one of my favorite headlines recently is the following. 21:09.627 --> 21:21.049 [JM]: "After a nearly 1000% surge, the AI debt orgy can't last forever while hidden borrowing has exploded to $1.65 trillion." 21:21.029 --> 21:22.553 [JM]: AI debt orgy. 21:22.974 --> 21:25.681 [DJ]: There was an AI debt orgy going on and you didn't tell me? 21:26.102 --> 21:27.085 [JM]: We both missed it, man. 21:28.047 --> 21:30.132 [DJ]: Story of my life, Justin. 21:31.255 --> 21:32.117 [DJ]: Story of my life. 21:32.218 --> 21:35.426 [DJ]: Somewhere a bunch of people are having a good time and is Dan invited? 21:35.446 --> 21:37.030 [DJ]: No. 21:37.010 --> 22:07.017 [JM]: So apparently, in addition to the $225 billion of hyperscaler-related bond issuance that's already happened so far this year, and that is on pace to grow to $400 billion by the end of the year, in addition to that, there's all of this shadow debt where there is this hidden borrowing that is somehow off the books and that amounts to multiples... 22:06.997 --> 22:10.465 [JM]: of the amount of related bond issuance that's on the books. 22:11.107 --> 22:22.373 [JM]: And as I'm sure we have talked about in previous episodes, I have no doubt that a bunch of companies hiding a staggering amount of debt is going to end well. 22:22.353 --> 22:22.814 [DJ]: Right. 22:22.834 --> 22:24.636 [DJ]: That's traditionally what happens, correct? 22:24.876 --> 22:37.931 [DJ]: When lots and lots of the economy has all these positions based on debt, and then something causes people to start calling in their debts, and the fact that there was no actual money behind any of this stuff. 22:38.271 --> 22:44.859 [DJ]: I mean, we've seen that happen before, and I'm pretty sure no one lost all of their retirement savings, right? 22:45.640 --> 22:47.204 [JM]: No, no, it went great. 22:48.366 --> 22:56.225 [DJ]: They definitely didn't make movies about it where Margot Robbie explained financial instruments to the viewer while reclining in a bathtub. 22:56.485 --> 22:59.312 [DJ]: So that might have just been a dream I had, actually. 22:59.798 --> 23:16.744 [JM]: Well, thankfully, if none of this makes any sense to you, the fine folks at a publication called McSweeney's, which I highly recommend, has published a very helpful article called "AI Economics for Dummies", the link to which will, of course, be in the show notes. 23:16.884 --> 23:19.849 [JM]: But I am going to read you part of it because you're already here. 23:19.829 --> 23:21.371 [DJ]: Because that's the service you pay for. 23:21.691 --> 23:22.112 [JM]: That's right. 23:22.152 --> 23:23.894 [JM]: "Benjamin owns a farm. 23:24.295 --> 23:26.598 [JM]: He employs 100 workers plowing his fields. 23:26.878 --> 23:29.121 [JM]: His total payroll is $10 million a year. 23:29.682 --> 23:35.129 [JM]: One day, he buys a mule, which provides the worker who uses it with a modest 10% productivity gain. 23:35.590 --> 23:42.799 [JM]: Benjamin fires 99 of his workers and purchases 99 mules, expecting a 1000% productivity gain. 23:42.779 --> 23:47.249 [JM]: The driverless mules cause plow damage to his property in excess of $50 million. 23:48.011 --> 23:58.915 [JM]: Benjamin loses another $5 million due to the loss of productivity from his one remaining employee, who no longer guides a plow but instead spends 100% of his time shoveling mule manure. 23:58.895 --> 24:04.864 [JM]: Goldman Sachs builds an altar to Benjamin in their lobby and cuts out the heart of a junior analyst on it every Friday. 24:04.925 --> 24:07.168 [JM]: They call it Blood Sacrifice Friday. 24:07.208 --> 24:11.335 [JM]: The name isn't catchy, but the event becomes a management favorite nonetheless." 24:11.936 --> 24:17.525 [JM]: And also because, you know, they provide multiple ways for you to understand AI economics. 24:17.985 --> 24:23.434 [JM]: They also mention: "Xavier owns an apartment that he rents out at a loss of $1 billion per month. 24:23.414 --> 24:31.664 [JM]: Seeing this success, he decides to make financial commitments to construct $850 billion in new apartments in places nobody wants them. 24:32.065 --> 24:40.475 [JM]: He convinces Ted to leverage everything he owns to help him build the apartments, telling him that once they are built, every human being on Earth will live in them. 24:40.895 --> 24:48.645 [JM]: Ted contributes $100 billion, part of which immediately goes toward paying off Xavier's $1 billion per month loss. 24:48.625 --> 24:55.553 [JM]: Forbes gives Xavier and Ted a cover feature, likening their building project to God creating the heavens and the earth. 24:55.933 --> 25:03.422 [JM]: Many Fortune 500 CEOs take this comparison literally and establish a new religion around Ted and Xavier with themselves as high priests. 25:03.882 --> 25:14.194 [JM]: Soon they start a holy war with the Pope, declaring Ted and Xavier the only one true gods on earth and promising to purge the non-believers in an official press release. 25:14.174 --> 25:20.929 [JM]: They annex, then subsequently demolish Vatican City, committing another $900 billion to build new apartments in his place. 25:21.410 --> 25:28.004 [JM]: Forbes hails this as disruptive, though it's not clear how Ted and Xavier plan to finance the project." 25:28.345 --> 25:33.837 [JM]: And they go on to say, "We hope these examples help clarify the inner workings of AI economics. 25:33.817 --> 25:43.228 [JM]: But if you're still confused, all you really need to know is that everything is totally working and everyone is making a lot of money and you should just stop asking questions, Luddite." 25:43.589 --> 25:44.850 [JM]: They should teach this in business schools. 25:45.972 --> 25:50.617 [DJ]: Are we 100% certain they're not already teaching this in business school? 25:50.637 --> 25:51.178 [JM]: We are not. 25:51.939 --> 26:00.229 [DJ]: There's a certain dark thread of religious mania running through those examples that I found disturbing, which may have been the point. 26:00.901 --> 26:02.745 [JM]: I think perhaps it was indeed the point. 26:03.326 --> 26:07.575 [JM]: As one last segment of Abstractions-is-now-a-financial-podcast... 26:08.016 --> 26:18.437 [JM]: It turns out that researchers have looked at how large language models can help people with personal finance related questions and have come to the conclusion that 26:18.417 --> 26:39.683 [JM]: financial advice provided by these models is surprisingly good and say, "The results showed that LLMs can offer an affordable, widely accessible source of financial guidance that can help users overcome the significant costs, biases, and conflicts of interest associated with traditional human financial advisors." 26:39.663 --> 26:42.468 [JM]: And I think it's cool that people did this research. 26:42.508 --> 26:44.412 [JM]: This is from MIT Sloan. 26:44.933 --> 26:51.465 [JM]: And I imagine that there was at least some degree of scientific rigor involved in doing this research. 26:51.485 --> 26:55.914 [JM]: So I have little reason to question the results of their studies. 26:56.375 --> 27:02.346 [JM]: But I think I would question, at least on a long enough term horizon... 27:02.326 --> 27:15.136 [JM]: whether significant costs, biases, and conflicts of interest are not also going to plague the inputs to large language models and thus the outputs of large language models. 27:15.637 --> 27:18.343 [JM]: So I don't know in the end 27:18.323 --> 27:32.024 [JM]: that they will necessarily help users overcome those problems, so much as allow them to replicate very similar problems, and also adding confabulation to the mix. 27:32.044 --> 27:42.981 [JM]: But at least as it stands today, with the important caveat that they mentioned, right there in the headline, it says AI financial advice is surprisingly good, especially if you ask the right questions. 27:43.214 --> 27:51.944 [DJ]: Yeah, I feel like that qualification is going to turn out to matter maybe more than it appears at first glance. 27:51.964 --> 27:53.165 [JM]: I tend to agree. 27:53.185 --> 28:01.615 [JM]: And they say that large language models miss important nuances and that better prompts could help. 28:01.975 --> 28:08.142 [JM]: Going on to say, "Regular people are not writing their prompts the way a finance professor is." 28:08.307 --> 28:13.472 [DJ]: Well, but that's okay, because regular people can just hire a financial advisor to write the prompts for them. 28:13.972 --> 28:14.593 [JM]: That's brilliant. 28:15.634 --> 28:24.382 [JM]: And that is, I think, consistent with what we are seeing with large language models in other disciplines like software engineering. 28:24.863 --> 28:35.293 [JM]: If you pick up a large language model to vibe-code your bespoke to-do app, not knowing anything about programming or software engineering... 28:35.273 --> 28:44.383 [JM]: you're probably going to get a different result than someone who is a seasoned veteran in programming and software engineering and has been doing it for most of their life. 28:44.944 --> 28:57.518 [JM]: And therefore, I'm not at all surprised to see that if you ask questions from a perspective of, "I know a decent amount about personal finance and 28:57.498 --> 29:02.306 [JM]: related topics", you're going to get better results out of large language models. 29:02.346 --> 29:04.790 [JM]: That doesn't strike me as particularly surprising. 29:05.231 --> 29:08.978 [JM]: But like you said, it also raises the question, okay, so how do you solve it? 29:09.018 --> 29:14.647 [JM]: You hire a human who's a financial advisor to prompt the large language model. 29:14.667 --> 29:16.631 [JM]: I mean, it really is like, I think you nailed it. 29:16.671 --> 29:19.776 [JM]: I think that that's clearly the right answer. 29:19.756 --> 29:29.487 [DJ]: There is this interesting / very worrisome problem with using large language models, and I just put my finger on what it feels like to me. 29:29.828 --> 29:33.512 [DJ]: So are you familiar with the Dunning-Kruger effect? 29:33.532 --> 29:33.812 [JM]: I am. 29:34.693 --> 29:42.422 [DJ]: This is the cognitive bias that describes the tendency of people who have low ability in a specific area to give 29:42.402 --> 29:52.738 [DJ]: overly positive assessments of their ability, which is the polite Wikipedia article way of saying people who are bad at stuff think they're good at stuff. 29:53.159 --> 29:54.822 [DJ]: And that can be a problem. 29:55.262 --> 29:58.808 [DJ]: And I would say it is a very specific problem when it comes to using 29:58.788 --> 29:59.850 [DJ]: large language models. 30:00.310 --> 30:12.530 [DJ]: Large language models sort of entrench this effect because, as many people have pointed out, they present their output in a tone, since their output is text. 30:12.550 --> 30:15.936 [DJ]: They present their output in a way that is confident. 30:15.916 --> 30:25.614 [DJ]: So large language models are generally – it's not that they won't provide output that's like, well, maybe you should think about these alternative things. 30:25.675 --> 30:29.061 [DJ]: It's that, as you mentioned earlier, they lack nuance. 30:29.602 --> 30:33.790 [DJ]: Or what I've noticed about them is they are constrained by their lack of context. 30:33.770 --> 30:46.328 [DJ]: So that a lot of the time when I will get output from a large language model, I can detect a bunch of assumptions in the output that are not necessarily correct, but they're based on whatever the input was. 30:46.969 --> 30:54.580 [DJ]: And when I'm working in an area where I have a high degree of skill, like if I'm using large language models for software development, 30:54.560 --> 30:56.884 [DJ]: I'm able to apply my own expertise. 30:57.445 --> 31:10.189 [DJ]: So even if I'm asking for a recommendation, like I'm not familiar with React frameworks for creating web UI components, so I ask an agentic coding tool, 31:10.169 --> 31:21.801 [DJ]: to give me some recommendations based on X, Y, Z. I can at least evaluate the response based on my expertise, and I can look at things and go, hmm, I don't think that's quite right. 31:21.841 --> 31:25.832 [DJ]: Or, oh, I see, this doesn't actually matter to me, though. 31:25.812 --> 31:27.837 [DJ]: or, I'm going to go look into this further. 31:28.298 --> 31:41.712 [DJ]: But my fear/observation is that if you lack experience in a certain domain, the fact that you can get this like highly detailed, highly structured output from a large language model is... 31:41.692 --> 31:50.183 [DJ]: output that superficially resembles the result of a lot of human reasoning without actually containing any human reasoning. 31:50.203 --> 32:01.557 [DJ]: I think you can easily be misled, not just misled in general about a given claim, but misled into thinking that you now know a lot more about this topic than you actually do. 32:01.537 --> 32:18.276 [DJ]: So yeah, the idea of people who don't know much of anything about finance getting a bunch of like financial advice from a large language model, I'm sure there are some set of like good information that will be helpful, right? 32:18.316 --> 32:25.924 [DJ]: Like if you've never encountered the notion that you should buy stocks when they're relatively low and sell them when they're relatively high or that like 32:25.904 --> 32:31.853 [DJ]: generally bonds are less risky over a given timeframe than stocks are or something like that. 32:31.873 --> 32:39.985 [DJ]: If you've never encountered those concepts before, you may be very well be able to learn them from a large language model instead of from a financial advisor. 32:40.405 --> 32:48.317 [DJ]: But ideally the whole notion of a financial advisor, and yes, they do have biases like certain products they would like to sell you and things like that, 32:48.297 --> 32:50.900 [DJ]: or certain philosophies about investing. 32:51.240 --> 33:03.452 [DJ]: But the big reason to work with a financial advisor is that you can give them a lot of context about your situation, and then they can filter that context through their expertise. 33:03.773 --> 33:07.096 [DJ]: It's a lot harder to do that effectively with a large language model. 33:07.556 --> 33:14.203 [DJ]: So I would be concerned that people who are driving their entire financial future off of ChatGPT sessions or whatever... 33:14.183 --> 33:20.893 [DJ]: could end up getting steered down some bad roads while thinking all along that they're getting great advice. 33:21.434 --> 33:21.975 [JM]: Absolutely. 33:22.315 --> 33:31.950 [JM]: And in certain countries, human financial advisors, unlike their digital variety, have a fiduciary responsibility to their clients. 33:32.231 --> 33:39.702 [JM]: And what that means is that they have a responsibility to act in the best interests of their clients. 33:39.682 --> 33:44.829 [JM]: From a legal perspective, they are legally required to act in the best interests of their clients. 33:45.350 --> 33:56.505 [JM]: As you might have already guessed, large language models are not constrained by these pesky laws and regulations and have no fiduciary responsibility to act in your best interests. 33:56.965 --> 34:04.255 [DJ]: Yeah, good luck taking OpenAI to court because ChatGPT got you to make some bad investment. 34:05.405 --> 34:13.114 [JM]: And going back to this notion of it's easier to get useful output from a large language model when you 34:13.398 --> 34:16.945 [JM]: know a lot about the topic that you're trying to ask about... 34:16.965 --> 34:35.659 [JM]: There's a great article that I saw in the last couple of weeks called Large Language Models Reward Expertise that makes the same argument that we've been making here and the same argument that's mentioned in this article about large language models offering decent financial advice if 34:35.639 --> 34:36.801 [JM]: you ask the right questions. 34:37.281 --> 34:51.041 [JM]: And in this article, the author says that people seem to think that since everyone is talking to the same models, that the skilled prompters are getting the same results as people who are using these tools for the first time. 34:51.521 --> 34:53.324 [JM]: And by "these tools", I mean, large language models. 34:53.744 --> 34:55.267 [JM]: And the author says, "This is wrong. 34:55.667 --> 35:01.275 [JM]: The most important skill in prompting is expertise in the domain you're prompting for." 35:01.255 --> 35:14.828 [JM]: And goes on to explain that people who are skilled mathematicians have been able to achieve incredible results with large language models in a way in which a less skilled mathematician would not be able to do. 35:15.289 --> 35:20.714 [JM]: So we're seeing this now across all these different domains, math, programming, personal finance. 35:21.074 --> 35:27.060 [JM]: So I think the pattern is pretty clear that indeed, as the article says, large language models reward expertise. 35:27.260 --> 35:28.203 [JM]: All right, moving on. 35:28.243 --> 35:33.642 [JM]: I want to ask you, Dan, about your new high fidelity system. 35:33.682 --> 35:40.003 [JM]: This is something that has been in the works for a long time, and I am ready to hear all about it. 35:40.118 --> 35:52.596 [DJ]: So speaking of areas where I don't have a lot of expertise, a few months ago, I got this notion in my head that I would like to build more of a traditional stereo system. 35:52.636 --> 35:56.001 [DJ]: And I got a lot of this from you. 35:56.481 --> 36:00.867 [DJ]: So basically, I blame you for everything that we're about to talk about, Justin. 36:01.128 --> 36:01.609 [JM]: Cool. 36:02.810 --> 36:08.558 [DJ]: I'd bought a pair of bookshelf speakers a year or so ago, and I like them. 36:08.618 --> 36:09.720 [DJ]: They're cool. 36:09.700 --> 36:16.653 [DJ]: I mean, they're great speakers and they're beautiful looking, but I had a little bit of buyer's remorse when I took them out of the box. 36:16.733 --> 36:28.395 [DJ]: And in order to use them, I was directed to make sure they were connected to my Wi-Fi network, download an iPhone app, create an account, agree to the terms of service and privacy policy. 36:28.435 --> 36:32.923 [DJ]: I want to briefly interrupt myself to remind everyone that 36:32.903 --> 36:40.474 [DJ]: that I'm talking about a thing that you plug a wire into so that it can turn an electrical signal into sound. 36:40.914 --> 36:53.171 [DJ]: So you might reasonably ask yourself, what is in the privacy policy that I didn't bother to read, that I'm agreeing to, to create the account, to use the app on my phone, to operate these speakers? 36:53.692 --> 36:54.473 [DJ]: So anyway... 36:54.453 --> 37:12.216 [DJ]: You and I had talked about this, and you were singing the praises of so-called passive speakers, a passive speaker being a thing that really does just accept an electrical signal and turn it into sound waves and doesn't connect to the internet or do anything else. 37:12.616 --> 37:16.041 [DJ]: So I decided to put a system like this together. 37:16.081 --> 37:21.928 [DJ]: And speaking of large language models and expertise, I started off doing some research 37:21.908 --> 37:23.652 [DJ]: using a large language model. 37:24.133 --> 37:28.542 [DJ]: Admittedly, to replace research I would have just done by typing it into an internet search engine. 37:28.582 --> 37:33.933 [DJ]: So I did get a bunch of suggestions and then go follow them up and look at them. 37:34.835 --> 37:38.162 [DJ]: But buying audio equipment like this... 37:38.142 --> 37:56.927 [DJ]: is one of these areas of consumerism where there are sort of very definable tiers, which I would define roughly as hundreds of dollars, thousands of dollars and as much money as you have, basically. 37:56.907 --> 37:59.470 [DJ]: And furniture also works this way. 37:59.930 --> 38:03.794 [DJ]: As I found a couple years ago, I wanted to replace my couch. 38:04.255 --> 38:09.981 [DJ]: And you can go to Ikea and get a perfectly serviceable couch for several hundreds of dollars. 38:10.501 --> 38:19.090 [DJ]: Or there is this sort of middle tier of furniture stores you can go to that will sell you a couch for a few thousand dollars. 38:19.450 --> 38:26.678 [DJ]: But then there are also these other furniture stores that will sell you a couch that costs more than your car. 38:26.658 --> 38:32.605 [DJ]: I saw a couch on a website that I liked the look of, but there was no price listed, which should have been my warning. 38:32.645 --> 38:36.130 [DJ]: And I called them and I was like, how much does this couch cost? 38:36.510 --> 38:37.872 [DJ]: And they said $27,800. 38:38.012 --> 38:44.801 [DJ]: And I said, right, thanks a lot. 38:45.121 --> 38:50.628 [DJ]: And I hung up because they had reminded me what socioeconomic stratum I belong in. 38:50.688 --> 38:54.893 [DJ]: And it's not the one where you spend $30,000 on a piece of furniture. 38:54.873 --> 38:59.419 [DJ]: So it turns out that audio technology is like this too. 38:59.800 --> 39:03.445 [DJ]: So, of course, the speakers that I bought were integrated. 39:03.845 --> 39:18.045 [DJ]: They have amplifiers and electronics and connectivity and stuff built into them so that ultimately you can plug the speaker in and connect it to your wireless network and agree to their terms of service and privacy policy automatically. 39:18.025 --> 39:24.874 [DJ]: But once you've done those simple steps, you can just stream QuoBuzz directly to your speaker, which is cool. 39:25.334 --> 39:41.635 [DJ]: But when you build a more so-called passive system, you're looking at components like an amplifier, which just takes a set of audio input signals and just produces an output signal to some speakers. 39:41.795 --> 39:45.940 [DJ]: And the speakers just accept the signal and turn it into sound waves. 39:45.920 --> 39:55.617 [DJ]: So all the things that take your music and do things other than encoding and decoding the audio signals has to happen elsewhere, which is fine. 39:55.918 --> 39:58.162 [DJ]: So I decided I wanted to build one of these things. 39:58.222 --> 40:10.864 [DJ]: So ultimately, I built a system that consists of an integrated amplifier, two floor-standing speakers, and a turntable because I realized that I'm a middle-aged millennial. 40:10.844 --> 40:23.356 [DJ]: And if I don't want my membership card taken away, I really do need to be collecting vinyl records of all of the albums that I already bought on compact disc back in the 1990s. 40:24.157 --> 40:24.618 [JM]: Makes sense. 40:25.098 --> 40:30.804 [JM]: So if I may, what was this equipment specifically that you bought? 40:30.844 --> 40:33.186 [JM]: The amp, the speakers? 40:33.747 --> 40:39.973 [DJ]: Yeah, so I ended up getting an Arcam Radia A25 amplifier. 40:39.953 --> 40:48.975 [DJ]: They came up pretty quickly as being a really solid, not completely ridiculous component for this purpose. 40:49.235 --> 40:50.679 [DJ]: So the thing has lots of input. 40:50.719 --> 40:51.641 [DJ]: It's got analogs. 40:51.741 --> 40:52.664 [DJ]: It's got digitals. 40:52.764 --> 40:57.355 [DJ]: It's got a phono input for the aforementioned turntable. 40:57.335 --> 41:05.326 [DJ]: And it apparently has like a really good digital audio converter and really good amplification capabilities. 41:05.386 --> 41:08.070 [DJ]: And it was a reasonable price. 41:08.530 --> 41:10.553 [DJ]: Again, I'm going for the mid-tier stuff here. 41:10.633 --> 41:16.862 [DJ]: So admittedly, I'm paying like a couple of thousand instead of a couple hundred dollars, but I'm also not paying, like 41:16.842 --> 41:18.984 [DJ]: car money, which you definitely can. 41:19.705 --> 41:31.998 [DJ]: I ended up buying this stuff from a local audio enthusiast store, and I was definitely buying their entry-level stuff, which I think for normal people is not entry-level. 41:32.058 --> 41:38.965 [DJ]: It would be like, okay, you definitely don't need to spend this much money on just these sorts of things. 41:39.205 --> 41:42.208 [DJ]: But from their perspective, it's like, oh yeah, well, you start here. 41:42.188 --> 41:45.959 [DJ]: And then, you know, when you want to get serious, you take out a second mortgage on your house. 41:46.079 --> 41:54.763 [JM]: And then once you've decided that you're really into this, then it's time to sell a kidney or two, and then you can afford something that actually sounds good. 41:54.912 --> 42:00.504 [DJ]: Well, then you can afford cables because the other thing is you need to attach an amplifier to speakers using speaker cable. 42:01.065 --> 42:13.030 [DJ]: And, you know, you and me and the other ignorant fools out there might think that a speaker cable is merely a piece of like braided copper wire that carries electricity from an amplifier to speakers. 42:13.110 --> 42:14.052 [DJ]: But... 42:14.032 --> 42:18.779 [DJ]: Don't be ridiculous because you could buy one of those for like 20 bucks. 42:19.320 --> 42:36.586 [DJ]: But the thing I found out when I started looking into this, this audio file stuff is that you can spend thousands of dollars on your amp and you can spend thousands of dollars on your speakers and you can spend thousands of dollars on a turntable, but you can also, Justin, spend thousands of dollars. 42:36.966 --> 42:37.767 [DJ]: No, no, no, no. 42:38.148 --> 42:40.191 [DJ]: I really want to make sure you hear, you heard me. 42:40.171 --> 42:54.671 [DJ]: You can spend not tens, not hundreds, thousands of dollars on speaker cables, the piece of wire that connects your amplifier to your speakers. 42:55.172 --> 43:00.299 [DJ]: You can buy cables that cost thousands of dollars. 43:00.779 --> 43:06.347 [DJ]: I just want us all to sit quietly and ponder this reality for a moment. 43:06.327 --> 43:07.008 [JM]: It's true. 43:07.068 --> 43:21.601 [JM]: And as I've undoubtedly mentioned in a past episode, I know someone who lives not too far from where I'm currently sitting who has indeed spent thousands of dollars on a single cable and has done this more than once. 43:22.301 --> 43:25.204 [JM]: And it makes about as much sense to me as it does to you. 43:25.224 --> 43:27.706 [DJ]: I love that... But I love that... 43:27.906 --> 43:29.348 [DJ]: "And he's done it more than once." 43:29.408 --> 43:36.334 [DJ]: And it's like, well, I would hope so, since generally speakers require two cables each and you usually get them in pairs. 43:36.314 --> 43:44.307 [DJ]: So it would be very sad if he had bought just one thousand-dollar cable because you really couldn't operate a sound system with it. 43:44.547 --> 43:46.010 [DJ]: I guess this is really how they get you. 43:46.070 --> 43:46.270 [DJ]: Right? 43:46.310 --> 43:50.817 [DJ]: Because you're like, well, I guess, you know, eighteen hundred dollars for a single cable isn't too much money. 43:50.877 --> 43:52.761 [DJ]: And it's like, all right, but I need four of them. 43:53.321 --> 43:53.542 [DJ]: So. 43:53.762 --> 43:54.103 [DJ]: All right. 43:54.163 --> 43:55.465 [DJ]: Here here goes my here. 43:55.485 --> 43:56.547 [DJ]: You can have my kidney. 43:56.627 --> 43:57.428 [DJ]: Here it is. 43:57.408 --> 44:06.859 [JM]: So I love how when you go to the manufacturer's product page for this integrated amplifier that you got, there's a link that says, "Find a dealer". 44:06.919 --> 44:08.741 [JM]: That should be your clue right there. 44:08.761 --> 44:09.903 [DJ]: Yeah, right. 44:10.323 --> 44:12.366 [JM]: It's not like, oh, buy now with a price. 44:12.466 --> 44:12.846 [JM]: No, no, no. 44:13.086 --> 44:14.829 [JM]: There's just a button that says find a dealer. 44:14.849 --> 44:15.990 [JM]: That's when you know you're in trouble. 44:16.450 --> 44:16.891 [DJ]: That's right. 44:16.951 --> 44:22.858 [DJ]: You need to go find a store that will condescend to sell this to you. 44:22.838 --> 44:24.240 [JM]: Okay, so that's the amp. 44:24.560 --> 44:26.202 [JM]: So tell me now about the speakers. 44:26.703 --> 44:41.182 [DJ]: I bought that amp, and then I got a pair of Dali, which is unfortunately not a reference to the famous surrealist painter, but instead stands for, I think, Danish audiophile loudspeaker industries or something like that. 44:41.623 --> 44:43.125 [DJ]: They make very nice speakers. 44:43.105 --> 44:50.134 [DJ]: And again, talking about the money side, like as these things go, they make decently affordable speakers, which is nice. 44:50.354 --> 45:00.827 [DJ]: So I bought the Sonik -- with a K -- 5s, which are relatively small floor standing speakers and they look great and they sound really good. 45:01.227 --> 45:02.869 [DJ]: I'm very happy with these components. 45:02.889 --> 45:09.417 [DJ]: And then for a turntable, again, like you can, you know, you just got to use turntable off of the internet or whatever, but. 45:09.397 --> 45:17.546 [DJ]: No, I bought, I really liked the look of this Project Audio EVO Carbon 2. 45:18.207 --> 45:22.572 [DJ]: Like they could have come up with a simpler naming scheme for these things, but it's something like that. 45:23.112 --> 45:36.467 [DJ]: So this is a very, and again, this is considered relatively like mid-tier or maybe even entry level as like audiophile equipment goes, but it is pretty expensive for something that just makes a record turnaround. 45:36.447 --> 45:38.451 [DJ]: But it is very heavy duty. 45:38.611 --> 45:43.381 [DJ]: Like this thing will hopefully last longer than I will, you know, if taken care of. 45:43.421 --> 45:47.549 [DJ]: So I got this thing and I hooked it all up. 45:47.930 --> 45:54.743 [DJ]: The one big hurdle I ran into is my amplifier is in the media cabinet underneath my television. 45:54.723 --> 46:00.616 [DJ]: So I wanted to put the turntable on a sideboard, which is, you know, at the other side of the side of the room. 46:00.957 --> 46:04.725 [DJ]: But the cable, very nice cable that comes with this turntable. 46:05.187 --> 46:06.710 [DJ]: I mean, it's not a thousand dollar cable. 46:06.750 --> 46:07.071 [DJ]: Come on. 46:07.091 --> 46:10.318 [DJ]: But it's still a very nice cable is like three feet long. 46:10.298 --> 46:12.020 [DJ]: So I needed a longer cable. 46:12.341 --> 46:24.257 [DJ]: Now, one thing that I got to learn about while setting all this stuff up is some things about electricity that I tended to take for granted, like the need to ground components to each other. 46:24.297 --> 46:30.325 [DJ]: Because most electrical components you buy, like computers and televisions and stuff, you don't need to worry about that. 46:30.345 --> 46:37.875 [DJ]: But it turns out that when you're connecting a turntable to an amp that has the phono preamp built into it, 46:37.855 --> 46:41.660 [DJ]: It sounds like you pretty much need to ground those things to each other. 46:41.700 --> 46:44.564 [DJ]: Otherwise, you end up with this weird electrical hum. 46:45.045 --> 46:55.099 [DJ]: So once I moved the turntable to where I wanted it to sit, I needed a cable about 15 feet long to go along my wall to the amplifier. 46:55.159 --> 46:56.901 [DJ]: So I ordered such a cable. 46:56.941 --> 47:01.247 [DJ]: This is just an RCA cable with the two left and right sides to it. 47:01.227 --> 47:12.009 [DJ]: I ordered it off the internet, got the cable, hooked it up, set my amp to the phonograph output, and immediately a terrible sound came out of my speakers. 47:12.350 --> 47:13.271 [DJ]: And I thought, oh, no. 47:13.312 --> 47:15.095 [DJ]: Like, do I have the wrong cable? 47:15.296 --> 47:15.756 [DJ]: What is it? 47:16.157 --> 47:18.983 [DJ]: I eventually figured out it was probably this grounding thing. 47:18.963 --> 47:22.870 [DJ]: But OK, now I need a 15 foot long like grounding wire. 47:22.930 --> 47:24.273 [DJ]: Where do I get this? 47:24.313 --> 47:28.361 [DJ]: And I'm looking on the Internet and it's like, where can you find stuff like this? 47:28.822 --> 47:42.287 [DJ]: And then I realized again, though, this is the nice thing about using this sort of old school technology as opposed to your integrated download an iPhone app and agree to a terms of service stuff, which is a grounding wire is great. 47:42.267 --> 47:43.669 [DJ]: Just a piece of wire. 47:43.770 --> 48:00.278 [DJ]: So while I'm assuming that someone will sell you a thousand dollar version of this, I ended up just taking some leftover speaker wire because, oh, I skipped the part where instead of paying thousands of dollars, I just made my own speaker cables out of copper wire and banana plugs. 48:00.618 --> 48:02.942 [DJ]: I had some leftover speaker wire. 48:02.922 --> 48:14.722 [DJ]: So I just took a long enough piece of that and stripped the ends off and tied them around the grounding posts on my turntable and my amplifier and problem solved. 48:15.002 --> 48:26.201 [DJ]: So I had both the satisfaction of spending probably a little too much money on this silly electronic equipment, but then also getting to do some DIY stuff. 48:26.222 --> 48:28.305 [DJ]: So I'm pretty pleased with my hi-fi system. 48:28.285 --> 48:29.327 [JM]: Very cool. 48:29.368 --> 48:31.713 [JM]: These speakers look really fancy. 48:32.194 --> 48:41.316 [JM]: As a brief sidebar, I'm going to send you a screenshot of what I see when going to the Dali product page for these speakers. 48:41.785 --> 48:52.357 [DJ]: Justin has sent me a screenshot of the Dali product page for these speakers, overlaid with the Google reCAPTCHA, "Select all images with bicycles". 48:52.858 --> 48:55.341 [JM]: I'm looking at a product page for *speakers*. 48:55.901 --> 48:59.906 [JM]: Why do I need to authenticate and pass a CAPTCHA? 49:00.467 --> 49:02.629 [JM]: Do you want to sell me the speakers or not? 49:03.010 --> 49:06.474 [JM]: Or is it only humans get to buy speakers, apparently? 49:06.514 --> 49:10.138 [JM]: If you're a machine, you can just go bugger off. 49:10.118 --> 49:10.939 [DJ]: No, exactly. 49:12.040 --> 49:14.002 [DJ]: Justin, this is the low-tech world. 49:14.623 --> 49:24.674 [DJ]: Join us, because some of us have decided that the modern world of large language models and smartphone apps is enervating and obnoxious. 49:24.734 --> 49:27.297 [DJ]: And so we have decided to regress to a simpler time. 49:27.917 --> 49:32.722 [DJ]: And the Danish audiophile loudspeaker industries company is right there with us. 49:32.823 --> 49:34.685 [DJ]: So you want to go to their website? 49:34.825 --> 49:35.325 [DJ]: Too bad. 49:35.365 --> 49:36.987 [DJ]: Prove you're not a robot. 49:36.967 --> 49:43.783 [DJ]: Because they will not sell loudspeakers to robots, who presumably already have built-in speakers. 49:43.803 --> 49:44.745 [JM]: Not like these, they don't. 49:47.733 --> 49:49.637 [JM]: All right, everyone, that's all for this episode. 49:49.838 --> 49:50.780 [JM]: Thanks a lot for listening. 49:50.820 --> 49:55.591 [JM]: You can find me on the web at justinmayer.com and you can find Dan at danj.ca. 49:56.072 --> 50:00.122 [JM]: Reach out with your thoughts about this episode via the Fediverse at justin.ramble.space.