matt and ryan have a chat

Added: 2026-05-27

[00:00:00]is live. Oh, there we go. I think we're live.

[00:00:02]>> You think so?

[00:00:05]>> Yeah, it looks like it.

[00:00:06]>> Okay. Okay, I'll believe you.

[00:00:08]>> Sorry we're late. Uh, what is new in your >> I um I came into this room and I went to uh click on uh you open up I saw the ecam was closed which is the tool that we use.

[00:00:21]>> Uh and so I tried to open it. I said eCam is not open. Said yeah I know I'm I'm trying to open it. And then I tried to launch it again. said eCam is not open or like eCam is not open anymore like yeah that that's the problem I'm trying to I'm starting it up ecam is not open and and so I would not >> I had to shut down the Mac and start it up again because it you know doing um >> command option escape it or command control escape to see you cancel the >> eCam's not running >> which oh wow >> was the error that was it was >> it So bizarre.

[00:01:00]>> So one thing that uh differentiates the two of us is is I used to work on Macs whereas you used to work on PCs. Uh so you're going into force quit I think whereas uh I think Windows calls it the task manager and Apple calls it the activity monitor for some reason. But in there you should be able to find your >> It wasn't I couldn't find any activity.

[00:01:21]>> I I couldn't find it just wouldn't start and it couldn't start because it wasn't started.

[00:01:28]Yeah. Yeah. Yeah. Okay. Um, so I I I suck, >> but it's this is it. The machine has been rebooted, so I don't know.

[00:01:34]>> Well, I mean, you're obviously on the thing now. Yeah. Yeah. How would you normally deal with like a lock file or a PID file? Cuz I think that's that's another one of those failure modes.

[00:01:42]Yeah. Yeah. Yeah.

[00:01:43]>> Yeah. I I don't know. I mean, if I eventually got to that point, I don't know.

[00:01:51]I think at one time at one point in time what you can do is like set the last time it was touched or something like that or like look at the um is it a time like the last time it was accessed or file modified date and let it like time out I guess in the business logic.

[00:02:04]Anyway, it's a non-trivial part of like like if you should only ever have one version of this app running. Making sure that's the case is non-trivial unfortunately.

[00:02:12]>> Yeah.

[00:02:12]>> Anyway, I have no answers for you other than to say like I I've stubbed my toe on this problem myself.

[00:02:17]>> Yeah. Yeah. I I I can't blame the ecam developers basically in this case.

[00:02:23]>> It's tough.

[00:02:25]>> So, listen, there was a lot of news recently. I'm I'm looking through kind of a like an overview of what happened between last week and this week.

[00:02:34]>> There's a lot of stuff out there in terms of business and things like that.

[00:02:37]I sent you a couple links. Did you? Um >> I saw I saw um >> I don't know if any of those are interesting to you.

[00:02:42]>> Yeah, you so you sent me a video of that from um what's his name? from 10 weird projects you need to know about >> and I had actually just watched that earlier today >> and I kept >> waiting for the you know I like by the fifth or sixth one it's like okay one of these I'm going to need to know about but so far everything is just garbage that I nobody needs to know about. I I think the one in there that might be of interest to you is honker. I I believe it's kind of like a like a wrapper in Rust for SQLite. So if you want to add SQL light to your >> project or whatever you that that seemed like a pretty handy one although I'm not like >> as far as I can tell SQLite's not that hard to integrate. No maybe >> and I mean if I'm not gonna I mean I don't know if I'd use SQLite anymore versus what is it? Terso.

[00:03:38]>> Fair. Fair. or I think there's a number of of different SQLite ones. The Auntu one was something I had looked at at one point in time. It's like distributed SQLite for um >> Kubernetes or MicroKS.

[00:03:51]>> Anyway, so of all the ones in there for for those of you keeping score, honker.dev is the one that I thought was kind of interesting.

[00:03:58]>> Um >> I mean the first one of like a GPU enabled terminals like >> with with the rat cursor that that is 3D and and rotating.

[00:04:06]>> Come on. Is this Oh god. Now, for that one, I will say they did say one of their primary inspirations was Temple OS. Have you heard of Temple OS before?

[00:04:16]Because it kind of explains why that terminal was so weird.

[00:04:20]>> Yeah. No, I mean I Okay.

[00:04:22]>> No, I don't think I do.

[00:04:24]>> You probably don't need to. Just Just know that Temple OS would be like the brainchild of one person trying to make their own operating system, which like listen, I think we've all had that fever dream before, but spent their entire life doing so. And I think they may have been um I'm trying to think of the the right term here. I I guess religious is probably a term I'm looking for, but hence the Temple OS name. It's kind of you know grounded in in religion as well. It's interesting. Like if uh I I think to me the the name I remember the name if there's a saying like my body is my temple. It feels like my OS is my temple is is kind of what that was of temple OS. was their way of I I know I'm sorry that is my pneummonic device to remember this operating system which is very odd and I will say the terminal definitely hits that that sort of cringe slash weird slash >> why did you even make this uh note right yeah I would agree with >> so the guy who wrote it passed away eight years ago >> uh oh wow I thought it was just before co okay wow >> um well co it was a while ago I know it was really good at keeping track of time obviously.

[00:05:38]>> Wow.

[00:05:39]>> But so from first principles the person made the entire operating system and all the accutra so all the accessories you'd want like a like a notepadesque a music player etc etc. I there may have been like a like a Bible I shouldn't be laughing sorry but like a Bible study guide as well. there few apps that were like um maybe primitives or or building blocks in that operating system that were more uh religiously aligned than other operating systems that I've seen before.

[00:06:07]>> Interesting.

[00:06:07]>> It's pretty interesting. Yeah. Yeah.

[00:06:10]>> But it it did have really weird um color schemes, ways of rendering >> and I I feel like the spinning rat terminal which I think is is it rat rat t was a terminal name?

[00:06:24]>> I don't know. You you don't even remember the first one. I think it was the rat terminal.

[00:06:28]>> It was rat terminal. Yeah. Yeah.

[00:06:30]>> Ratty.

[00:06:31]>> Yeah. Ratty. Oh, I I thought you meant like in the the terminal in Temple. I I don't know.

[00:06:36]>> Oh, I have no idea.

[00:06:37]>> But you were talking about the first thing in the video was the rat. Yes.

[00:06:40]>> Yeah. With the the rotating rat in the in the terminal, which listen, I think maybe if you're trying to build a Rust terminal, this might be interesting to look at as inspiration. If you want something that's 3D accelerated, they're using a library as well. Sorry, I'm trying to find it right now. Ah, I'm so sorry, but there is a library that they're using inside that allows for the 3D rendering. So, if you for some reason have the need to make a terminal uh like bend, turn, and warp, >> I think Ratty might be a very good source of inspiration.

[00:07:15]>> I I I don't have that.

[00:07:17]>> Yeah. No, I don't.

[00:07:17]>> Not something.

[00:07:21]>> Uh, other than that, I've been working with something called Symphony from OpenAI.

[00:07:25]>> Okay. So, you did mention that. What was >> Okay. Did I mention that last week as well? Sorry.

[00:07:30]>> I don't think so. I don't know.

[00:07:32]>> Okay. So, Symphony is essentially an opinionated uh take on orchestration. So, Asian orchestration. Okay. I've mentioned in the channel before, I've got a bunch of pies that I'm playing around with.

[00:07:43]They're running some weird version of BSD right now. But basically what I'm trying to do is say with um like virtual agents what what could I run? How could I orchestrate these different CPUs? I think they're one tops like they're not they're not very um strong in terms of inference but there are many of them which means that 15 of them would be 15 tops which is nothing to sneeze at. It's okay. I think what OpenAI is doing is saying instead of having everyone try to build their own orchestration why don't we let you know where we got to essentially with theirs. And what I found interesting is they got pretty far down this path. I'm not saying you should use Symphony or even build something like Symphony. But I thought what you might enjoy is the fact that their their agent process that they ran for a while to make this work landed on Elixir. And I think we've talked about this before in terms of primitives that are in languages and how things like uh should have already forgot but elixir there's a few other of these um like fail forward kind of frameworks where like you accept that your task may fail and you you move on which is like very alien to me. So not really garbage collection just like you you jettison the process and move on which is a a different way to approach programming but I think what they were saying is that >> because the primitives that they have are things like issues or tasks and they're often single threaded so you send it off single thread and it comes back elixir seem to to fit that mold for them very well. Previously, I think we talked about Gleam as well as as different ways of of doing this, but basically I I think like this is a pretty interesting take, but right >> my main complaint with it is it requires linear, which is is fine. I I just like I'm not a linear user. I I don't pay for I don't know if you >> I used it for a little bit because I mean they have a free level. I mean, if you're not doing >> Yeah.

[00:09:35]>> I forget what >> it's like a one user.

[00:09:37]>> Yeah. And I so I was playing around with it, but I didn't see any value in continuing to use it.

[00:09:44]>> Yeah. The other thing that it does a lot is use GitHub. And it uses GitHub. I'm going to use the wrong term here. I think it's work trees as opposed to workspaces, which limits the amount of work you can do in one repo basically.

[00:09:57]So you can only have I think it's like one work tree per main >> or is kind of how it works. or if you've got like multiple uh forks or branches or whatever, you're fine. But if you're working on main basically trunk based developments, a little bit harder.

[00:10:11]>> But yeah, anyway, um it's on their blog.

[00:10:14]I think it's a little bit older in case you've not seen it. Yeah, April 27th, >> but um I mean it's a month old. It's not that old.

[00:10:21]>> Uh but it's one of those things I looked at and kind of read and then, you know, had Claude look through it for me and see if there was anything I can learn in there.

[00:10:27]>> And I came back to because I feel like this is the right shape for agent coordination. And I think this is where >> like a lot of people are are landing on what would we talked about mailboxes and things like that or like task planning.

[00:10:41]What would that look like? Is it a combine board? I think we had talked about autoplot which now is a different name, but what does that look like? And I think what they're saying is we're just going to like shell out to whatever that word is and just use as an API and then allow you to use that as UI >> whereas I think a lot of people are trying to tie that UI into their agent orchestration, >> right?

[00:11:02]>> And it feels like um the agent orchestration has one main consumer which is agents, >> right?

[00:11:08]>> And what we're trying to do is wrap like a human interface around that orchestration layer. And whether or not the agents need a conbon board is is maybe not even up for debate. They they probably don't basically they just need a >> but you as the viewer, >> right?

[00:11:22]>> Yeah. Like how do you how do you ration rationalize about all the agents >> and I swear there's a segue in here I'm I'm working on. So like if you've got all these agents that are uh becoming more and more intelligent, then the harness that you need to apply to them becomes more difficult. And I say this because I did watch the video that you sent me that has a bunch of swearing in it about AGI, I think. Or maybe it feels like one engineer is wrestling with um whether or not AGI exists and whether or not it's important, I guess, to the work dayto day. So So I watched it, but I had I had thoughts. Um >> I never heard of this guy before.

[00:11:59]>> What was Mo Bitson or Mo What was the guy's name? up here.

[00:12:06]>> Um I thought Oh, Moar.

[00:12:08]>> Bitar.

[00:12:10]>> Um >> he's not bitter. He's >> bitar. Sorry.

[00:12:13]>> And and the comments uh basically the the three comments on the hacker news post where this was added uh were like, "Oh, another AI is bad video." Like I don't know if he was saying that, but >> that's truth. Yeah. Um yeah, but it was it was interesting that um you know you see so many people saying that AGI is close or AGI is here >> and his perspective was well is AGI even possible well first off what is it and what's the definition of it and is it >> by any definition is it ever going to happen not just is it going to happen within a finite period of time um Correct.

[00:12:57]>> And uh and then going down the rabbit hole of trying to figure out what it is and like I am never going to deal with this topic again. This is stupid.

[00:13:08]>> Yeah. So uh I will say and again this is me doing >> four semesters of philosophy studies. So like a very junior uh approach to this but um definitely more than 11 hours of research that the person has done into AGI have definitely thought about thinking a lot like a lot more than 11 hours let's just say >> and the the first part of all this is like who cares which he did touch on a little bit like who who cares if AGI >> is going to be a thing will it affect our lives >> and if it's like a one-year thing then maybe if it's a three-year thing fivey year I think the number they used was 200 years, which is like so far in the future. I mean, >> it's not who cares, but it's like it's maybe not as important to you. But like what what might be important is if you were to roll up this year into historical events, right? Those might be important in in 200 years. Or if you set the start to a framework that's picked up and used. So I I don't think the defeatism in the video is really well warranted either because there are strategies if you think that something's >> going to happen that you can attack it.

[00:14:17]And this is one of those things that I've brought up a lot. It's Rocco's basilisk. So if if robots or AGI or or some sort of nonhuman thinking is going to become more powerful, popular and use on Earth. The question is what do you do? Do you do you stop that from happening or do you help it from happening with the understanding that when it does happen that your actions will be weighed as a result, right? And and I think I feel like that's a more interesting approach to to this because they the who cares I guess test to the argument >> is hard to pass basically if if you start from there and whether or not we have AGI soon or not and and whether or not intelligence matters or or will we define intelligence and I think he was even saying like some scientists had said that they're not sure uh dogs for example think or or have intelligence and I feel like that's more semantic. So what what you call something and what actually happens are are both important but one of them is the key here and and what's happening that reasoning is what we actually care about and whether or not a like a tool using >> mammal is is intelligence is less important than if you gave them a tool and they could use it. So for example I think we've proven that ravens can use tools to >> like add rocks to a bottle in order to pull something like that that kind of thing. That's important because when you're dealing with an AI agent or a team of agents or what have you, it's important to know what the scope is or or what the capabilities are of what they're doing. And I I think he really hit a pretty big point in there that he then shoved under the rock unfortunately is that like this matters to things outside of coding, but it might actually matter for the coding agents we're using right now because they're general purpose. They're not really purpose-built. even though they they do really well at coding, they're also doing like NLP basically. So or NLQ, sorry. So we're giving them natural language and they're using that natural language to do things as opposed to maybe vectors or what have you in terms of like how they would actually go through the neural network to work.

[00:16:16]Anyway, all this to say it felt like one of those big uh >> like oh no, you know, existential crisis and then like oh I can't do anything about it so I'm moving on. Yeah.

[00:16:26]>> And it it was long and I feel like um it might be important to people who are like wrestling through the same thing basically >> and and wondering whether or not like intelligence quote unquote intelligence matters or if it's more the mechanism of of what they're doing that matters and what that impact is. But I think the the brass tax here is that like right now models can't quote unquote >> think. They they can roleplay thinking or or do the next thing and make it look like thinking and and for multiple steps of like what we actually do for reasoning. It could approach how we we think about things, but it's not the it's not the same. And maybe it doesn't matter if it's the same as as human intelligence as long as we can use it, which I think is sort of the conclusion at the end of the video, which is probably like, you know, this meeting could have been email. This email could have been a text.

[00:17:13]>> Not to downplay it. It's it's very important to think about these things. I think it is important. I'm not fantastic at metaphysics or philosophy, but I do think thinking about movies, just like the Matrix, even just thinking about like whether or not you're in a simulation is a good exercise to think about once in a while. Like, can you prove that we're not in a video game?

[00:17:30]Like, that's a really difficult thing to do, >> right? Yeah.

[00:17:33]>> Anyway, >> the Matrix isn't a documentary inside the Matrix. Yeah.

[00:17:37]>> Is it not? Like, that's very difficult to prove.

[00:17:40]>> No.

[00:17:42]>> Uh oh, shoot. We get a lot of comments here.

[00:17:45]>> So, we started off with Jay Topham asking about Hermes agent. Um, or some people like to say Hermes, but I like to say Hermes because that's what the guys at Noose Research like to call it.

[00:17:59]>> Um, >> that's a N O >> N O US. Um, >> what do you think about Hermes? Because I I actually have played with it now.

[00:18:07]>> I have too. Uh, and I've I've deleted OpenClaw because I uh didn't find it as useful. Um, I I don't know. Oh, how did that come out?

[00:18:19]I got ink everywhere.

[00:18:21]>> Sprink over there.

[00:18:22]>> I don't know. My My pen just decided to start bleeding ink.

[00:18:28]>> Are you using a fountain pen by chance?

[00:18:30]I mean, that does.

[00:18:31]>> Of course, I use a fountain pen.

[00:18:34]>> What kind of a nerd would he be without a fountain pen? uh while while he's cleaning up the the inky mess. I'll say that what I liked about Hermes agent is that uh that the scope of what it does can be made very small as opposed to at least when I used open claw it seem to want to pull down everything and use skills by default. What I liked about Hermes is you can say like and I think you can now for openclaw is like no skills library, no default skills added.

[00:18:59]Uh you can say like no web access by default which is fantastic. So it felt to me like those rational defaults uh were in place whereas when I used open claw at least there was there was a wizard to get started but it did not feel secure after I started at all.

[00:19:13]>> One of the things I like about one of the things I hate about openclaw is that I couldn't come up with a video about how to how to use openclaw because >> the installation is different every time you use it. Not just between different versions, but every time you if you wipe your machine, because I wiped it a dozen times trying to like make a a repeatable video, I could not make a repeatable video because the experience was different every single time. Um, and not a dozen. I probably did it two or three dozen times. It was different every time. And >> doing the same thing with Hermes, it was the same every single time. It was the same process, repeatable. And that is what comes from I mean it is a more mature you know because Hermes was started months before openclaw was started which I didn't realize until pretty recently that you know they they it was an internal tool they were using and then they thought oh people seemed to like this thing let's just turn it >> open source instead >> and as they evolved at the same time it feels like noose was learning from some of the failures of open glass I saying failure here but like some of the uh the mistakes. Yeah. Yeah. And I one of the things I like is that >> um you start off with no skills or minimal skills. Um but as you start doing things, it says, "Well, I'm going to I'm going to make that into a skill."

[00:20:38]Okay. And and so it makes it into a skill. And so now it's it's researched it once and now it doesn't need to research it each time it runs it because it's already figured out and made a good still for it. Um, and so I've done that with a number of things and it's like this is fantastic. Um, >> so I like that.

[00:20:57]>> I like the timer and schedules for for some reason I I don't know what it was with Open Claw, but I could not get things to fire on. Maybe that was just me, but that's that's kind of like my minimal thing is like I could set up Naden to to get a schedule. I could use cron and mess up the computer schedule or a VM or what have you. Or I could use an agent that actually understands time >> and it it seems to just work. It's not setting up like Apple playlists and doing anything in the background. It's it's quite handy.

[00:21:25]>> Yeah. Yeah. I found that repeating tasks actually repeat more reliably in Hermes than they did in OpenClaw. I mean, I could say, "Do this every day at 5:00 a.m." And it says, "Great. I've set that up." And then you come back to it, it's like there's nothing set up. There is nothing in any anywhere.

[00:21:45]>> It has done >> I've done it haven't done what wrote files to temp a lot of times I I would find and some of it like access they didn't ask for or something.

[00:21:57]We did have one comment here from Juan uh Alberto is asking about AI agents doing work and I think what you're saying is that uh a lot of people are talking about AI agents working 247. I I'm I'm certainly not there yet. I don't know Matt if if you are, but what I do want is like once or twice a day do do something. Yeah.

[00:22:18]And that's kind of what I've achieved with Hermes or whatever you want to call it.

[00:22:23]>> But I mean there are definitely people who have set like uh they have in their heartbeat uh thing to you know every every you know five minutes or 30 minutes or whatever your heartbeat is running at you know do this research task. And so it's it's not running 24/7.

[00:22:38]It's just doing a task over and over and over and over again. But if you're doing that task over and over and over again, each one of those uh initiations is using the model to figure out what it needs to do, >> right?

[00:22:52]>> And so you're get spending tokens so fast. I mean people even if you're using Olama uh which is so much cheaper, you can go through your entire plan super quick if you've got this. I mean, and so if you're going to do a regular task or a scheduled task, it's so much more efficient and faster and better in like every way if you're pushing that off into an N8N or some other >> workflow solution versus having Hermes try it figure it out every time or or open claw or any of these tools.

[00:23:27]>> U I will admit to having made that mistake just just last week. So there's a new model that came out and I was like, "Oh, this is gonna be fantastic."

[00:23:34]I got Ola Cloud. This is going to be great. So I started using I think it's Deepseek Pro. Deep See Pro you will you will sorry you will definitely hit the limits of Ola Cloud if you're using Deep Seek Pro for something like a five minute timer because I think that's exactly what I was doing and you'll hit it. But I think Matt's kind of hit a point. It was like once or or got a point once it once you hit that utility it's better to to move that off something like an end to end task. prove prove the t prove the point and the the goal but once you've like seen that it works over and over again great make it repeatable in something like again it could be n or make or zap year or whatever you tool you want to use but use those tools for doing that and and some people will say oh I'll just build an application great build an application to do that if you don't want to use I think that's the more difficult appro approach to that but >> an API yeah >> uh um Sean Sean also said that they're using OpenCloud. Sorry, they went from OpenCloud to Hermes and now they're using Kimmy K 2.6. That That's what I went back to. Kimmy K 2.6 seems to be the best value right now. Uh I'd love to be proven wrong, but I've I've tried some of the newer models that work great. They just seem to be very token hungry. The outputs tokens in particular. Uh very um interesting.

[00:24:50]>> This is like the most comments I I mean, I'm like scrolling through all these comments I haven't seen yet because this like keeps going on and on. It's amazing. because we're talking about I don't I feel like this is what happens.

[00:25:01]>> Juan Alberto Unamuno and Sean don't understand how you know people don't comment on our show. It's it's this is not >> normal.

[00:25:12]>> This is not normal. I mean maybe we can get used to this the new normal. This is great. I love this.

[00:25:18]>> And I think the last one from Sean is is kind of about skills, right? So listen, there was a new MCP uh I think it was release candidate or at least the spec release candidate came out today where they're talking about changing things a lot in the MCP server which will help >> but in the meantime I think the skills are are probably where you want to go.

[00:25:34]So as you start realizing that things are repeatable as Matt said it should suggest making skills to you but I use a skill creator skill which is redundant but uh quite handy to be able to like just create I have not asked it to create any skills. It has gone ahead and said I've created a scale like great >> cuz I saw that I ran the same thing tweing.

[00:26:01]>> Oh, >> that's being attacked by spiders.

[00:26:02]>> There's another one.

[00:26:04]>> Oh, how many?

[00:26:06]>> Just two. And they're like the daddy long legs. So, I mean, they're not >> there's Are they Do they Oh, there's a third.

[00:26:12]>> They don't bite.

[00:26:13]>> No, >> they don't bite.

[00:26:15]>> No, but that's uh more than I ever noticed were in here. Daddy long legs or dandy long legs. I I actually don't know.

[00:26:22]>> Oh, I I I've only heard daddy longlegs, but >> it's called a spider. It is daddy long legs. I I >> In fact, it's interesting that uh somebody I I commented uh recently.

[00:26:38]Somebody was talking about locks and I said, "Oh, there's a place near me that uh sells locks on a bagel, but with also with uh salmon or sorry, locks is salmon, but uh with um Oh, it was a avocado toast. Avocado toast with locks on top and pickled onions."

[00:27:00]>> And and the person and somebody responded like liquid oxygen. It's like I never knew that locks also meant liquid oxygen. So anyway, >> alternate meanings for not >> looking on Wikipedia, it also suggests that, you know, under the disambiguation section, liquid auction could be locks.

[00:27:19]>> So anyway, >> I I think what they're saying is that uh they generally miss the live shows and they they're happy to both be here at the same time, which is like, hey, >> that's great. Amazing.

[00:27:28]>> That's great.

[00:27:29]>> Uh so normally folks would be talking about uh things you can eat, uh spiders.

[00:27:34]>> Yes.

[00:27:35]>> Possibly kaying. Yeah. Today we're talking about Hermes. I >> kaying though. I wish I could I wish I could I wish I could pick up the camera and point it into the garage because there's been in the middle of our or on one of the bays of the garage there's been this kayak on the ground like not on the ground but on a on stretchers uh for months and we bought a kit to to to you know uh uh haul it or hoist it up to the ceiling. We bought this like in November. We finally installed it over the weekend and oh my god, it is life-changing seeing this thing up in the air. Anyway, sorry, distraction.

[00:28:17]>> It's all good. Did you name your boat Adams? Is is that how we're running the boat?

[00:28:22]>> No, >> you've not named your craft yet.

[00:28:24]>> I have not named my craft.

[00:28:26]>> Um, which is fun. I don't know if that helps you name that other yet.

[00:28:32]>> Taco Cat. I don't know. Um, >> there you go. That's a pretty good name for a boat.

[00:28:37]>> Well, since because of the uh I don't know, maybe um Matt uh Inman would be offended that I >> because it was as it's one of the game the name of a game that you know, Exploding Kittens created and >> Matt it lives nearby and >> pretty close. Yeah. Fair enough.

[00:28:56]>> So, I don't know if I can take that name.

[00:28:59]>> You you mentioned a Mat and and I just wanted to talk again about Matt.

[00:29:03]who I I think we I I may have maligned last week. I felt that. Matt PCO's YouTube's actually quite good and he released a couple videos this week that I would >> even before the AI even before the AI craze. I mean, his videos were always great.

[00:29:17]>> I'm so sorry. I'm like I'm going back and and watching all the videos because I'd only seen one on I think it was uh not Roast Me, Quiz Me, Roast Bake Me, Grill Me, Grill Me. Sorry, I'm so bad.

[00:29:28]The The Grill Me skill, which is actually really handy and I I have been using. I started watching some more of his videos because he's got >> I think Grill Me with Docs is what we talked about last time, but he's actually got a skills repo that's full of great skills if you want to get started on on like building your own or like >> starting point or like looking how you would build a skill that can work with different agents for example >> as opposed to just cloud code. I anyway I'm I'm smitten very good channel. Yeah, there there was um for a long time there were three coding YouTube channels that I paid attention to. There were like his and uh is it Harrington or or you know um >> okay >> uh reddish reddish hair guy older guy um and then uh one out of I feel like it was out of Germany called Fun Fun Function. Did you ever see that?

[00:30:19]>> Don't I don't think he does. Huh?

[00:30:24]>> Jack Harrington. Jack Harrington. Yeah.

[00:30:27]Um, fun function. I don't know if he's still I think he's stopped for a long time. I don't know if he's still doing it, but uh they were great. Super entertaining videos.

[00:30:37]>> Fun function.

[00:30:38]>> Yeah.

[00:30:38]>> Uh, JavaScript total.

[00:30:41]>> Yeah. I mean, the right language. Yeah.

[00:30:44]>> Uh, last video is one year old, but yeah.

[00:30:46]>> Yeah, I think they they were released from last year.

[00:30:48]>> Yeah.

[00:30:49]>> Nice. These look like Oh, I've seen this one on uh functional programming. I already >> Okay, right on. Hold on.

[00:30:56]>> So, >> there's always good tech YouTube out there. But anyway, Matt Matt Pox, another Matt space.

[00:31:02]>> Yeah, >> there's there's this this Matt and then there's a few other mats that are popular in the space.

[00:31:08]>> Yeah, did really like his videos. I I think as you're as people are moving more towards skills or uh as anthropic might say like describing your job in in markdown as we get more of that I don't know knowledge base created I think we're making skills better and as a result agents are getting better because skills really do help.

[00:31:32]You mentioned you mentioned GitHub earlier and uh I did the uh there was there was a breach uh >> last week. One of the the the comment I loved the most about that was um somebody did a tweet x post whatever um and uh it was essentially uh um oh my god the the the hackers found enough up time GitHub had enough up time to do the breach.

[00:32:03]>> Oh oh oh that hurts.

[00:32:07]>> That's amazing. I have a mighty fallen.

[00:32:10]I I loved GitHub for years for years for years. It it feels bad for them to be the the enemy in the current timeline or or like a like a a bad player in the current timeline, but I think a lot of people are moving to Codeberg is is the one that you had mentioned last week or the week before.

[00:32:27]>> Yeah. Yeah. I mean, >> seems like it's increasing.

[00:32:30]>> I mean, the benefit of things being mostly on GitHub was that was a discovery platform. you could go there and find >> whichever tool you wanted. And now >> we don't have one.

[00:32:42]>> Uh it's a bunch of different ones. You know, there's Bitbucket that's existed forever, but I don't know if it's ever was discoverable. Um and then there's that Codeberg and another one.

[00:32:54]>> Oh, Gy fair to GitLab.

[00:32:56]>> Huh?

[00:32:56]>> GitLab.

[00:32:57]>> Oh, GitLab. Yeah. Yeah.

[00:33:00]>> But but Gy, I think. Yeah, they've got the hosted version or the self-hosted version. I'm running self-hosted. I think a few people are as well.

[00:33:07]>> A few people. Yeah, >> maybe millions.

[00:33:10]>> Yeah, >> they've got millions of downloads. Git seems to be quite popular. I will say the command line t command tea is very handy for which is like a big reason why I started using it is because >> just like the gh command generally your agent's going to understand how to use the t commands. We'll figure it out real quick which is nice. Did you see anything about had you heard anything about Bixon mania?

[00:33:37]>> No. Bixonomania.

[00:33:40]>> Bixonomania. Um a disease or an affliction uh from blue light from screens that cause sore eyes and a number of other things. If you were to, you know, describe your symptoms to an AI, it would often respond, well, you know, it could be ask your doctor about ask your doctor about bixon mania and it was fake.

[00:34:06]>> Okay, that's googling this like this. So apparently this researcher saw that um uh had had thought that the uh different AIs were training on you know they're training as as much stuff that they can get including preprints.

[00:34:23]>> Um and apparently preprints are like uh they described it as the tabloids for academia. um you know like you could put anything you want into these these preprints and >> publish it >> right and so it didn't matter what I mean they would accept all kind of garbage as long I sometimes you might have to pay to put it in there or they just >> they'll just take anything um >> some journals are pay to play anyway yeah >> sure but these were like with I mean post anything you want um and so this this research or this author this uh journalist u I think uh said Um she created her own university like fictitious university and then created preprint for that university and made it look kind of official. Um but then print posted this this article to it.

[00:35:12]>> Um and in the article in the in the in the um paper uh towards the end it says you know we want to uh thank our um you know all the people that supported us including the Lord of the Rings. Um, special thanks to the um to the science lab on the enter uh USS Enterprise and to Starfleet Academy. Um, special thanks to all these other people. And at the beginning it said, you know, all the 10 people who are cited on this uh as authors on this paper do not exist. Um, and yet all these AI engines use that in the training and is actually recommend or giving that as an answer. So it was amazing. So, I I was on um a Discord the other day and I just come back from walking the dog and somebody mentioned this Bixon Mania as like evidence or like a a proof point to something else somebody else was saying. Um and I had just on the dog walk I was just listening to the there's a a podcast called Science Quickly by Scientific American.

[00:36:16]>> Um and uh they were that's the topic they were talking about. It was It's such a amazing story. Yeah.

[00:36:24]Um, >> did you hear any news about the major AI lab saying that we need more AI fiction where it's not the quoteunquote bad guy in the story? because a lot of the problems that they're having, >> including I think I think it might have been anthropic actually, they where they had a test where multiple engineers were threatened with blackmail by the LLM agent >> and they were trying to figure out why this happened and and apparently they they found the source, one of the the sources and it's that a lot of the fiction that's being consumed by the agents includes the agent as a bad guy or the AI quote unquote as a bad guy in the story.

[00:37:02]>> I don't know. Do we need to start creating good AI fiction?

[00:37:05]>> I mentioned a book that I was reading and I just finished a few days ago called uh Infinity Gate.

[00:37:12]>> Um where it was, you know, jumping to alternate earths uh and they're like a bazillion of them and and a large number of them are actually ruled by this uh the pan dominion that rules all these different earths. Um, and uh um and but this researcher on on our Earth uh uh uses the AI to help her figure out how to do these jumps. She's just kind of discovered. And so in that the AI from her perspective is the good guy, >> but from this pandemium, the the the AI is the bad guy because it's helped her.

[00:37:57]>> God damn it.

[00:37:58]>> I like that we saw it on stream. I didn't want to say anything.

[00:38:00]>> Oh, fuck. I'm >> so sorry.

[00:38:02]>> Oh, it's all right. I just uh >> literal jump scare.

[00:38:06]>> Oh, >> I'm so sorry cuz your light was hitting it so we could all >> Worst dream ever.

[00:38:15]>> Oh my god.

[00:38:17]>> I'm going to pee my pants.

[00:38:20]That was so good. Are you okay? Do you need a second?

[00:38:24]>> Yeah.

[00:38:25]>> Uh did you not just describe Rick and Morty to us? So, what am I missing?

[00:38:31]>> Are there more spiders?

[00:38:33]>> I'm looking around. I mean, now I'm Now I'm really paying attention.

[00:38:37]>> They don't bite. You You'll be fine.

[00:38:38]>> I know. They don't bite. They just make these webs that are all over the place.

[00:38:42]But oh my god, that Well, he he's not going to make any more webs. That one is gone.

[00:38:52]>> A spiderb.

[00:38:55]So, you actually saw it in the on the >> Yeah, sorry. I was like, should I say I was like, h maybe maybe it'll just like float around. It'll go away anyway, notice. I'm just like, you know, don't don't alarm your co-host.

[00:39:07]>> Let him let him melt down on >> Oh my god.

[00:39:13]Wow.

[00:39:14]>> You poor.

[00:39:16]>> Anyway, um Okay.

[00:39:19]>> Yes. Rick and Morty, Portal.

[00:39:22]>> Yeah. What about Rick? Huh? That's what it sounds like to me. It sounds like the Rick and Morty, Portal Dun. And they do have Dominion that I think >> Oh, yeah.

[00:39:29]>> Uh rules some of the the pan galactic.

[00:39:34]>> Okay.

[00:39:35]>> And their their TV, you know, the Rick and Morty TV.

[00:39:38]>> I've never watched it.

[00:39:40]>> I know it's um uh >> it's a good exercise to think about Dan Harmon.

[00:39:45]>> Yes.

[00:39:46]>> Writes and created his story circle around it.

[00:39:50]Um, >> would somebody else in the chat said they went, "Ah, >> oh my god."

[00:39:59]Had you seen I Here's another interesting thing I saw. Uh, BFI um is some some feature of uh Wi-Fi routers.

[00:40:11]>> BFI.

[00:40:12]>> WiFi.

[00:40:13]>> Okay.

[00:40:14]>> Uh BFI.

[00:40:16]>> Okay. Yeah, I knew about Beam former.

[00:40:18]>> Yes. Um, and apparently some researchers have >> exploit >> figured out are able to use it to identify the people in the room.

[00:40:30]>> Oh yeah, this has been used for a minute.

[00:40:32]>> Oh, has it? Without any electronics.

[00:40:36]>> Oh, >> no. I mean, the the person doesn't need to have electronics. It's just how their body uh how the bodies absorb different signals.

[00:40:45]>> Interesting. And then I mean it looked like so I don't know if they could say you know they could track like me from from house to house to house but rather they know that these three people are typically in the house >> where you are in the house >> and so therefore >> uh Matt is in or not in the house as long as you know what subset of people exist. Yeah. So there are really >> it's not about >> it's not about uh you know Matt has a special signature and you can identify him anywhere in the world.

[00:41:18]>> I don't think >> Matt occupies space and blocks Wi-Fi signals. Yeah. Yeah. Yeah. And and kills spiders.

[00:41:24]>> Pretty amazing.

[00:41:25]>> Pretty amazing.

[00:41:27]>> Um I don't have much else to say about that, but I just saw that. That was cool.

[00:41:32]>> I I think I chocked that up when it first came out. I chocked it up to like u people could snap your snoop your keystrokes by using like a really long range like sound thing and >> wasn't that >> hear the frequency and like keystrokes.

[00:41:46]>> The movie Sneakers they I think was the first time I saw where they were like uh recording what they were saying based on the vibrations on the windows.

[00:41:55]>> Yeah.

[00:41:55]>> Um >> Yeah. And I thought that was fake when I was you >> just just sci-fi.

[00:42:01]>> God, it's ridiculous.

[00:42:03]So, I I I'm I'm still looking around.

[00:42:07]>> Um, so I um uh you know, for for I I often had to wanted to make videos that were a little bit more had a little bit more sound in it than just my voice, you know. So, have music but >> not too much music, you know, music in the right places. But, I mean, it's a skill I don't have to be able to mix that in. but also around sound effects.

[00:42:33]And so at one point I, you know, kind of bought into the idea of like really putting more effort into the sound effects, even though I can probably just like come up with, you know, the five or six kinds of sounds and then just use those five or six in the videos. But for some reason, I was enamored with this tool called Audio Design Desk.

[00:42:54]>> Okay.

[00:42:54]>> And I actually subscribed to it for a little while. um and uh or not I subscribed to it and then kind of forgot about it and then would use it every now and then and would make an attempt to use it. Um but then uh it it looked like they were slowing down as a business and weren't really sure what was going on and then as of maybe two months ago, three months ago, they just fully disappeared. Um, so if you have their uh software, it just it it doesn't stop working, but it can't, you know, connect to its servers and so it won't uh get new stuff. And that was the whole thing. You get new sound effects.

[00:43:37]So I mean I do I need the 100,000 sound effects that I have? No, >> they seem to be a lot. Yeah.

[00:43:43]>> Yeah. I don't need this. But um but they had this all these sound effects in this this uh format called the add sound I think it was called and or add audio and add library. So a library had multiple audio files in it.

[00:43:57]>> Many sounds. Yeah. um and you would download the whole library and you'd have this library but but now the thing is out of business but seems to be out of business and the um but uh you know so I was looking for a way to uh reverse engineer or or not reverse engineer but at least be able to pull this files >> like in a in in the right way. Um, and there was a video by Wes Boss pretty recently, okay, about >> him um trying to figure out how to use uh his amaran lights that use um >> what is that sinus?

[00:44:36]>> But can he figure out the MQTT or something?

[00:44:39]>> Yeah. Yeah.

[00:44:40]>> Yeah. could he figure that out and then start and he did.

[00:44:44]>> Um and so he basically he basically hand it said um uh whichever tool whichever thing whether it was claude code or one of the others um basically get had access to the camera so it could look watch to see when the light was on and said >> yeah yeah yeah >> figure this out. This camera will show you when the light is on and when it's off. Here's the how you can get to the camera. figure out how I can control it and and to, you know, even like change the lighting and change everything. And so it took the that the agent um a number of hours and it he ended up having to pay like $80 in tokens to to uh to now he's got this this thing that works. Now I don't know if he can >> Okay, I'll I'll pay his 80 bucks. I don't know if he could distribute that in any way because >> Yeah, it might be difficult because it's a proprietary protocol that you kind of bit bash. Yeah.

[00:45:38]>> And so distributing it might not be >> something he can or should ever do. I don't know. Um but you know I had that idea in mind like well hey can whenever agent can you figure out this format >> and it likely does this search and says well rather than you know figuring it out how about you just use this tool that does it for you >> and so there's this other tool that extracts the data uh called shadow ta shadow tag >> um and it's a kind of a media player media library but also imports add audio audio files. Well, the person the founder of Audio Design Desk is the creator of this tool.

[00:46:24]>> So, I was like, okay, so it must be legal. I mean, >> yeah, >> right.

[00:46:32]>> I don't know.

[00:46:34]>> I'm using the guy who created the tool to extract the data. So, now I've got these >> 100,000 sound effects, but as usable audio files, which is great.

[00:46:45]So it was it was in I mean it's the guy it's the guy's code the the guy who wrote it and so and he's that Yeah. So I thought that was interesting.

[00:46:53]>> Did I ever tell you my reinvent story about uh emergency sound systems?

[00:46:57]>> No.

[00:46:58]>> Did Did you know Matt that uh cop cars and ambulances and other emergency vehicles the sounds that they make is up to the operator to define? So even though they may sound similar in in in regions, >> apparently these are all completely different systems that like like in in countries and like languages and stuff like it's going to be different everywhere. And one company that happens to be somebody I met at reinvent I think possibly even a data dog client they were telling me about how they had to reverse engineer all of these formats.

[00:47:33]Now uh imagine Matt if you can a year not unlike 2018 that we were describing earlier a year before her AI how would you reverse engineer a siren sound or an ambulance turns out it's like very very difficult to figure out like just just what went into the sounds and how they work some of them are like bit maps some of them are like text some of them have like just all types of weird stuff because you could do whatever you want >> and they said that prior to this it wasn't that decoding the sounds was the biggest problem is that like if they couldn't use the sounds then it was something that would not like stop a merger but it' be like part of a merger discussion if they were going to be acquiring a company which tech they were using because it may be more or less difficult depending on they've got anyway this is a a task that essentially was a nightmare for years years and years and years and I think I met them in 2024 and they were like once I found Janai and put on this problem I could not believe it and it it it takes a long time. Sometimes it costs us a lot of money, but it is absolutely worth it because not only can we go down and like reverse engineer, but then we can go back and test it and like record a file and then use multimodal extraction to figure out if the cos are like further away. And they said it was basically a game changer for their industry which is inundated with with code like that. When they were talking about that, all I could think about was cobalt because I think one of the things about cobalt is like once you've compiled the cobalt code and shipped it, >> you just you trash the source because >> it's shipped and it's working and you do like you don't I think it was like a hacker news comment like you don't have storage space to >> write down the code. So the binary is the living artifact and it's it's living for years like 30 40 years. Anyway, this this is my story about uh >> random binary. I know we're close to time. My check need to leave at time.

[00:49:24]>> No, >> you always have an appointment like >> not this time. Uh my co-host is on vacation by the way. Okay. So, yes.

[00:49:32]>> Did you see that Ferrari has a new car?

[00:49:36]>> I did. Fully electric.

[00:49:39]>> Fully electric.

[00:49:40]>> Mhm.

[00:49:41]>> Have you seen a picture of it?

[00:49:44]>> No, I have not. Is it?

[00:49:45]>> It's horrible. It is like the first time that Ferrari has ever made a vehicle that looks absolutely awful.

[00:49:53]>> Oh, it is not.

[00:49:54]>> I mean, it looks It is actually I saw one one uh picture that had >> um had uh the Ferrari next to a Tesla. I mean, they look exactly almost exactly the same car, and I'm not a fan of the design of Tesla. Um, so it looks horrible. Now, if you like Tesla cars, then it looks great. But, um, it's awful.

[00:50:23]>> It does not look good.

[00:50:24]>> And even in the in the dashboard inside is like, oh yeah, it looks like two iPads because it's >> the dash looks okay. Yeah, but I'm, you know, the interior looks fine. I'm not a fan of the exterior of this car. Is it luch luch?

[00:50:37]>> Yeah, I think so. That's my Italian.

[00:50:40]Yeah. Yeah, it it seems like a great idea. It was made and manufactured in Italy, which is awesome.

[00:50:49]>> So, it's not like a BYD platform that they're using, which again is just what everybody seems to be doing in Europe.

[00:50:55]>> Uh, wow. Designed by uh Sir Johnny IV.

[00:50:59]>> Yes. Yes.

[00:51:02]>> Okay, Johnny.

[00:51:05]>> Wow. Yeah, >> that's an ugly car.

[00:51:08]>> Yes.

[00:51:10]Um, and then what else? Oh, one other thing that I saw was really cool.

[00:51:16]>> Uh, you've got a browser. Check out audiommass.co.

[00:51:22]>> That's the right one. Audio mass.

[00:51:24]>> And there Massachusetts.

[00:51:26]>> No, >> I'm teasing it.

[00:51:29]>> And there it looks like a web based >> audio editor.

[00:51:33]>> Multitracker.

[00:51:34]>> Yeah.

[00:51:34]>> All on the web. And it look pretty cool.

[00:51:36]It works pretty well.

[00:51:39]It's It's pretty slick. I don't have much to say about it. I just saw it.

[00:51:42]It's like, "Oh, that's really neat."

[00:51:44]It's always neat to see, you know, what can you do in a brow? I mean, like in a browser and it's it's, you know, either load up the sample or, you know, upload your own audio. It's pretty neatly done.

[00:51:55]>> I'm squinting because it was released in 2020, July 2022, actually, 2022, >> and then this year had like massive changes from from January. Yeah, it's been around since July 2022 and did not change until January 2026.

[00:52:12]>> So yeah, this is like another possibly prei.

[00:52:17]>> Definitely very cool.

[00:52:19]>> It's uh 100 kilobytes kil bits probably kilobytes >> compressed >> and and has no backend or framework dependencies. Amazing.

[00:52:30]>> Amazing.

[00:52:32]So anyway, um, what else I have?

[00:52:35]>> Yeah, >> I >> that's the whole list.

[00:52:38]>> I guess we're we're kind of a Oh, well I So, you know, whenever you get a a a message, uh, like a a text message or an email said, and it you're not sure the sender and yet they include a link >> and like that's always kind of sketchy.

[00:52:56]I so I I mean I always tell my mother, you know, if or and her boyfriend um you know, if if uh um if you get an email, even if it says it's from Xfinity or from >> anybody you you know your bank um from Boeing Credit Union, whatever you you get that thing and it says click this link or whatever or dial this phone number. Don't find your most recent bill and look at the website on there and go to that website or look at the phone number there and go or it's on the back of your credit card, look at that phone number or any, you know, do not follow a link in whatever. So, I just got a text message um saying and the entire message is from an unknown number uh you have a document that needs to be notorized to continue. Follow this link.

[00:53:51]like there's no way I'm clicking on that link. But uh it said also sent to your inbox. And so in my inbox there is a message an email a legitimate email from Riven >> from Rivian um saying you know here's you it's time to sign the title for your new R1s that we get uh day after tomorrow.

[00:54:13]>> Um and so it it >> I think it is legitimate. I'm not going to claim that.

[00:54:20]I'm going I'm going to sign into riv and go that way rather than clicking any link because this is >> I mean now the >> I wonder if it's like a clearing house they're using or something like you would think Rivian's advanced enough to know that their market can click on this link.

[00:54:36]>> Yeah. Yeah. Yeah. Um, but generally, I mean, we're excited about getting our Rivian and uh, and we know it's, you know, the best car for us based on all our needs, >> but buying a Rivian has got to be the most difficult experience any I mean, it their sales process is horrible.

[00:54:56]>> Do do they have dealerships?

[00:54:59]>> Uh, so they have uh design centers.

[00:55:04]>> Okay. So, and they're not allowed to talk about although in the state of Washington they Riven is >> as of like June in a next month I think.

[00:55:14]Okay.

[00:55:15]>> They're allowed. consider like a car dealer >> because because um because Tesla is able to talk about it and but it's like an exception just for Tesla and Rivian sued and said well this is ridiculous and state of Washington said yeah okay Tesla and Rivian >> so >> fair >> um so anyway I thought it was interesting that this text message just came in it's like uh I'm not clicking that but this I'm pretty sure this is legitimate And Rivian should not be setting that stuff like that.

[00:55:51]>> Uh I don't have I got my my one of my two other things are actually like pics.

[00:55:57]>> Do you have anything else you wanted to share to talk about?

[00:56:00]>> Matt Matt BCO that that's mine. You got to check his Yep. Check out the other Matt on YouTube. He's he's pretty good.

[00:56:07]This Matt >> The other Matt There are a few Matts.

[00:56:09]>> There are a few Matts.

[00:56:11]>> Yeah. Um >> there are a couple. So, I uh day after day before yesterday, uh our daughter wanted to watch something and and we've been watching a lot of Little House on the Prairie. We're we're up to season 9, which is the last season. Um and uh but seasons 7, 8, and nine, they got into some really dark topics. There are >> Oh, really?

[00:56:35]>> There's um there's a a suicide in there.

[00:56:40]There's uh murder. There's uh a really >> Little House in the Prairie.

[00:56:45]>> Little House in the Prairie. There's some bad things in there. Um and so we like look at the description. It's like, okay, she's not watching that. I mean, she's seven years old. We're not watching that one. Skip, skip, skip, skip. And we go through like that's how we're able to go through, you know, like a season in basically two nights. Uh because we're only watching two episodes of >> avoid all the family discussions.

[00:57:05]>> Avoiding all the really hard topics. Um and uh and then more recently we've switched on to um All Creatures Great and Small, the newer newer version.

[00:57:15]>> Uh newer meaning last I don't know 5 10 years ago that it came out.

[00:57:19]>> It's funny. Yeah.

[00:57:20]>> Um which is a great series. We love that.

[00:57:23]>> Um but uh and so we were going to watch that, but then I saw a recommendation for a movie. I thought, well, let's let's try it out. And there are almost no scary scenes in it. There's one that talks about the war, but uh at World War I, I think u but it's a movie called Hugo >> and it is so magical and lovely and wonderful and such a great I mean I our daughter loved it and I loved it. I thought it was amazing. Um and uh uh and yeah, it was just this amazing find.

[00:57:59]It's got um Ben Kingsley and uh Jude Law and Emily Mortimer and it's just this amazing little movie about this uh kid who um his father dies off pretty well before the story starts. But um his father has left this automat takes place in the early 1900s has left uh the child this automaton that's amazing. Um, and uh, and he's trying to repair it, but also solve the puzzle of who made it.

[00:58:32]And it might be uh, the person who uh, I've forgotten the guy's name, but he did that original movie of where you see the the guy landing on the moon and it goes into the guy's the moon's eye. Um, so it's that that is the uh filmmaker who apparently made it made this automaton and a lot of the the underlying stories are actually true and apparently that guy was really into these automatons at the time.

[00:59:01]>> Um, and so it, you know, took the basic story and and twisted it and made it really interesting.

[00:59:07]>> Um, and and so we loved that movie. Um, and we had just found it as say on uh Sunday, I think. Um, and a book that we also found that we're we're slowly reading and going, you know, as as Sully goes to bed. This wonderful story. Oh, and I don't know if it counts as a I don't know if it's necessarily a child's book book, but it's it's all uh totally fine to read for a child, but it's also I mean this it's so beautifully written and so wonderful as a as a this book called the evolution of Kalpernia Tate.

[00:59:48]>> Evolution of Capernia.

[00:59:51]>> It is a lovely book. It is there's nothing bad happens in it and it's just perfect.

[00:59:59]>> It's such a great story and and it it's it's a great story. Um so uh so watch that well for that one. Um and then the the so I wasn't I didn't have that on my list but I mentioned it because that was a recent thing. But here's something that is on my list. Portable >> magic.

[01:00:16]>> Oh no.

[01:00:17]>> This is such an amazing book. It's a history of books and their readers. So portable meaning they're they're the these are magical items.

[01:00:26]>> You're getting into magic. That's why I did the face. I'm sorry.

[01:00:28]>> Um uh and so the first chapter talks about um >> uh Gutenberg and the Gutenberg's Bible and how you know parts of that history that >> I never knew as you know because I grew up on in the West and and therefore I didn't learn about what really h when about movable types history. Um, and so apparently Gutenberg's Bible is what bankrupted him. Um, he had a number of other books that were published before uh using movable type, but he didn't actually come with I mean he came up with it on his own, >> but somebody else came up with movable type before him. And there are examples of >> um Chinese uh printed books or printed I don't know sure if they were books or schools or what the format was but uh using metal movable type >> um from 100 years before Gutenberg was born. Um, and but if you count movable type as not just metal but clay as well, then we're looking back in the, you know, 500 years before Gutenberg.

[01:01:41]>> Really? I didn't know that either.

[01:01:42]>> Like this history I I didn't know about.

[01:01:45]So it's a fascinating um next chapter goes into the u American service editions of books. So uh the US Army or US military produced these books. They were all a standard size that would fit in the standard in the solders's pocket.

[01:02:03]Right.

[01:02:03]>> Yeah.

[01:02:04]>> And so there were copies of The Great Gatsby and that before before the American Service Editions printed out The Great Gatsby, >> um, uh, Fitzgerald made like $13 on royalties in 1939 or something like that. Um but uh and had such a small uh run of books that you it was it was not super easy to get. Um but uh so I think it was saying the year before they started um there were 100,000 paperbacks in the US like >> total but within um but by the first year or second year that this program was in uh place um 40 million books were being produced.

[01:02:54]>> Wow. Um and uh and it's just you know how it changed the the acceptance of paperbacks as a as a thing and um uh the it used the printing press for Readers Digest and Reader Digest couldn't get paper and the subscriptions were way down during the war. Um but they printed out these books for the American Service Edition and then distributed widely and it was just there's all sorts of amazing little things in this book. It's I don't know how I picked it up. It was just like in the um there's a section in our library that is uh just a quick uh you know where you can the bookshelf right by the doors you like pick something interesting and just grab it and go or grab it check it out and go. Um and so this was I think in that section and it's so amazing amazing little things in there.

[01:03:44]>> So that was it.

[01:03:47]>> Yeah. Like is the Great Gatsby even a good book? Like >> I don't know >> of all the books. I mean, I had to I had to read it, but there were military before, you know, >> there were uh two there were a thousand different titles that were produced by this American Service Editions.

[01:04:03]>> Okay. Um, and uh the one that it starts out talking about um uh a biography of Queen Victoria um that this American was reading while stuck in a um uh in a trench somewhere, you know, and and just like trying to avoid getting shot. um and he was reading about Queen Queen Victoria and you know there's so many other people that were uh you know just you know they they would >> right so disconnected >> and they would um uh show up at a base and you know find the you know would they just grab a you know a few books and take them with them and um and read whenever they had a chance of downtime they would just read and it helped um get more people I think I think I was saying that uh before that you know reading I mean people could read but it wasn't something that you did for enjoyment and so many people left the army after the war enjoying reading because >> that was their escape for you know in these battles this was their way to escape at whatever time and they came back as >> big readers so >> that's really cool >> it's an interesting story interesting book so portable Cool magic.

[01:05:27]>> Love it.

[01:05:28]>> That's all I got.

[01:05:29]>> All right, that's all you guys. All right, folks. Thank you so much for all the comments.

[01:05:33]>> I did everything. And there's my ink stains, too.

[01:05:36]>> Yeah, I see that. They're still on your finger, too. And you killed a spider.

[01:05:39]>> And I killed a spider. Oh my god, that was terrifying. I'm going to have to watch the video just to see.

[01:05:44]>> This episode was just rife with character development. Well, you can see me kind of see it and I'm like, I don't know.

[01:05:51]>> Absolutely terrifying.

[01:05:54]>> All right. Thank you, Sean. Thank you, Juan. And thank you, >> I think we also had Jay Topham as well.

[01:05:59]Thank you all folks for the comments.

[01:06:01]>> Appreciate it.

[01:06:03]>> This was fun.

[01:06:04]>> It was fun. Let me find the finish button.

[01:06:08]I'm still my I think my heart is still racing from that spider.

[01:06:12]>> We got to work on your recovery. Yeah.

[01:06:14]>> Oh my god.

[01:06:15]>> Sorry. My hair is still