Episode 248: Thingamajigger

Podcasting 2.0 for January 23rd, 2026, episode 248. Sing him a jigger. Hey everybody, welcome from a very cold, cold America. It is time for Podcasting 2.0. This is it, this is where we talk about podcasting. It says it right there on the label. This is the board meeting of Podcasting 2 .0. In fact, we are the only boardroom that doesn't host a party at Davos. I'm Adam Curry here in the Texas Hill Country and in Alabama, the man who coached with Claude and conquers with confidence.

Say hello to my friend on the other end, the one, the only, Mr. Dave Jones. I've had Robbie Dupree in my head for like days now. It's, it will not go away. Robbie Dupree, which song? Steal away, dun, dun, dun, dun, dun, dun. Why don't you steal away into the night? That's a, that's in hot rotation on Yacht Rock. Is that a, is that a Yacht Rock song? Oh yeah, that's a total Yacht Rock song. Oh yeah, absolutely. You ever listened to Yacht Rock on like, do you have a Sirius XM?

No, I don't have Sirius. Oh yeah, Sirius XM has a Yacht Rock channel. Did Robbie Dupree have any other songs that were his? No, of course not. No, no, that was it. No, it's like, it's Robbie, Robbie Dupree. It's a Rupert Holmes with the Pina Colada song. These are all, these are all Yacht Rock classics. Yeah, it's beautiful. Eddie Money has to be on there somewhere. Take my bone tonight. Sorry, sorry, sorry. Two tickets to paradise. I've got two tickets. Hold on a second.

Let's see, let's see what other, let's just, let's just go to the YouTubes. Let's see what comes up as Yacht Rock playlist. Is REO Speedwagon, is that Yacht Rock? Keep on loving you? Yeah, I would say. What is this? What do we have here? Ooh, how come I'm not hearing anything? Oh, here we go. Remember Keith Largo? Bertie Higgins? That's good. Yacht Rock is almost all one hit wonders. Oh yeah, well, here's a whole playlist. Let me see. Oh yeah, there's Kenny Loggins.

Absolutely. Yacht Rock all the way. Number two on the list. Number two on the list is, there it is. There's your Robbie Dupree. I cannot get it out of my head. I cannot get it out of my head. And right into. Yeah, baby. What a fool believes. Oh yeah. And then the classic. The classic. Boom, boom, boom. Remember this one? Was this Boz Skaggs? No, no, no, no, no, no. Come on. You know it. Oh, oh, oh, the, oh, oh. Oh, is this Daryl Hall & Oates? No, no, no, no. I need to hear his voice.

Where it's rocking it here for you on Yacht Rock. This is Player. Their hit, Baby, Come Back. Oh, yeah. Baby, come back. Wow, this is. Oh, oh, and then of course, we have the classic Seals & Crofts. Boom, boom, boom, boom, boom, boom, boom, boom, boom, boom. That's right. And then we have, oh yeah. Oh yes, classic. Come in Jamaica, winner. Yeah. Down in Jamaica, nigga. Remember On & On? One more, one more. On & On, On & On. This is also classic Yacht Rock, everybody.

Look in glass. Who could forget them? Brandy, you're a fine one! Brandy! There has to be Little River Band on there somewhere. Oh, no doubt. No doubt. Wow. I have not listened to these in a long time. I've been stuck in the Christian hits, man. Little River Band, yes! Nails it. Nails it. Reminiscing, is that the one? Oh, Reminiscing has got to be on there. Wait, wait, wait. But a classic from Dr. Hook and his medicine show.

Come on, baby. Remember Dr. Hook and he had that eye patch on with the cowboy hat? The problem with Dr. Hook era of music is that it was all on 8-track. And those things vanished. People have vinyl, but 8-tracks didn't survive. They melted from time or something. You can still find them in some classic cars. I looked at an El Camino the other day. Why did you look at an El Camino? Oh, I really have wanted an El Camino for a long time. No. Yeah, it's an SS, beautiful. Oh, yeah.

You could get those with a 454 in them. That's what it's packing, baby. The only problem is it is generally orange. Ooh. Yeah. It's one thing to buy a classic. That's a bold statement. Yeah. Then I have to repaint it and I'm like, I don't think so. Yeah. But it had an 8-track. Orange is a little much. It had an 8-track in it. 8-track player, at least. I think that I should have looked and seen what was in there. 8-track sucks so bad.

I mean, you had to switch and listen to a different song to rewind. Yeah. Oh, yeah. You had to switch tracks and you hear the heads go. Yeah. Yeah, everybody. We're talking about podcasting now. Yeah, that's right. Yes, we are. Podcasting all over the place. Bitcoin orange. Hey, congratulations to Sam Sethi and the kids over there at TrueFans. Closed their round of funding. It was the big news today. Oh, nice. On Power, yeah. I haven't heard Power yet. It was interesting.

He was talking about, I was talking to some other hosting folks earlier this week about bot traffic. And so the way, if I understand it correctly, the way TrueFans hosting will work is they will allow you to turn off bots.

Now, if you turn off the bots, that means that Apple, which appears to be one of the biggest bots these days, because it's sucking everybody down for transcripts and chapters and whatever else, the question is what will happen if you block the Apple bots, which truly are using quite a lot of bandwidth. So I've got this, I've got an interesting idea that I want to throw out to the boardroom here. All right, boardroom. Hello. Pay attention, everybody.

But I don't want to, I don't want to hog up the mic though. So if you, if you got, we can talk about it later, but I think it's related. No, why don't we, I mean, I think we both have some vibe coding to talk about. So we can do that now or you can. Why don't you talk about your vibe code? Dude. What's up, dude? Dude, I'm almost there. I have been wanting to get off of Windows for my podcasting ventures for years and years, years and years.

And so because our mutual friend Paul had recommended to me to take a look at Omarchi, which is March Linux with the tiling window manager. Oh, as they say, opinionated Linux. It's Linux with even more David Hennemeyer Hanson is what it is. Who's that? I don't know who that is. DHH, the guy, the inventor of Ruby on Rails. Oh, oh, okay. He's the guy that like, he started, he did, he's the mind behind Omarchi. Oh, hold on. We're not lit. I'm sorry. My mistake. Oh, thank you, board meeting.

Thank you, boardroom for reminding me of that. This stuff needs to be automatic. So I, so I loaded it up on, you know, I think I just need a test box. I still have that Surface Pro 6. Oh yeah. The one that was giving you trouble for a while. Well, it wasn't giving me trouble. Oh no, that was an older one. No, that was, that was a different machine. This one was running as my, what do you call it?

What's the, what's the open, you know, the one that had, like what you use for the, for your Bitcoin node, for your Albi. Oh, Umbral? Umbral, yeah. I was using it for an Umbral, but whatever reason, that just kept, you know, I had to upgrade to the new, it just, it became a mess. The Umbral over time just, I felt, becomes a mess. I got my Umbral fixed, by the way. Oh, you did? It's working, yeah.

Yeah, so I've been, and ever since I got the Start 9, that box is so rock solid, it's just, it's just running. I haven't looked, I mean, I pull up my Albi, my, I'm forgetting all kinds of, low T, ladies and gentlemen, my helipad, you know, that kind of stuff. It's all on that box. Works great, love it, don't have to think about it.

But, so I decided I'll install Omarchi on this, because if it can install on this, which are notoriously messy hardware systems, I think, the Surface Pro 6, just Surface in general, which I believe is now discontinued. Completely unreliable hardware, yeah.

Yeah, unreliable. Well, installed great, and... Really? Oh, yeah, right out of the box, nope. Well, no, that's not true. I had to do one thing for the Wi-Fi, which, of course, is very typical. But it was interesting, because I've been using Gemini now, I've dropped Grok, Grok is dropped. But I started using Gemini, mainly because I got a free, you know, Pro account with my phone. And I think that's the same thing that happened to Alex.

He started using it because he got a free plan, like some free time on it for a few, it's like three months or something like that. Yeah, but I mean, I'm all in, because for research purposes, it's dynamite. And I'm all, I'm in Pro mode, so nothing is fast. You know, a question or a research topic can take easily a minute or a couple minutes to come back. But because it has access to Google search APIs and YouTube transcripts and all this stuff, it really has, it does great for research.

And it gives me the sources, so I can check where it's coming from. There's some bias in there, but, you know, but it's been very useful to me. So I jump into Gemini and I say, okay, here's what's going on. And right away, it's super helpful. Oh, this is not a problem. And it says, and by the way, you know, while we're doing this, just grab the USB cable, plug in your Galaxy 7Z Flip, because it knows I have that, of course. And that'll tether, no problem. And it did.

And so I got the Wi-Fi fixed. I got the touch screen, everything's working. And I did that at the kitchen table in maybe an hour, you know, and most of that time was just the install taking place. And I'm like, okay, I am, and I say, listen, Gemini, I am ready to move my podcasting set up to, and I played for a day or two with Amarchi. I like the tiling window system. That is the way my brain works. It's like, oh. I thought you would like that. Oh, I really do. You know, and all keyboards.

I'm a keyboard guy. I love keeping my hands on the keyboard, moving windows around, you know, different, especially because it's a limited screen size. It's just, I'm blown away. You know, the super key space bar brings up the apps you want, type it in, launches. I mean, it just, it's kind of the way I've always wanted to work. Really nice. So I'm ready to move off. It's like, let's start with something that has been impossible in my life.

I want to take my Rodecaster Pro 2, and I want to be able to see all these devices. Wow. Wow. Pipewire and Helvum did the trick. Really? Yes. Finally. Yes. And so Helvum is really the, you know, that's really the magic. Pipewire, by the way, is great. I mean, I've been, the last time I looked at this, it was, you know, Jack, what was it? Ulsa Jack or something. The Jack system, you know, wonky, breaks down, falls apart. Pipewire, amazing.

And, of course, the only thing that's really different between Windows, when it comes to USB drivers and Linux, is that you can plug in this, you know, 16-channel USB device, which is basically what one of the devices, you know, because you have two different devices in the Rodecaster. One is the standard secondary out. It recognizes that. No problem. You plug the other one in. It's like it just shows you one. It doesn't show you the chat, like the mix minus device.

So with Helvum, you can permanently map those two channels for the system to recognize it as a device. And so now I can just pull up one of three devices exactly the way I'd want it on Windows. I'm like, okay, I've been waiting for this all my life. And Gemini walked me through it. Then I'm like, you know, okay, what am I missing that's on Windows that I don't have in this environment? Because in order to do my podcast, I have been working for years and years with M-Airlist.

And M-Airlist is a commercial product, licensed product, which is really for radio stations. But with it, I've been able to build this interface that I like. In essence, I have eight players at the top of my screen. And you can drag and drop a file into it, which I can do pretty quickly. And, you know, it's mapped to a MIDI controller. I have the Nano Control 2 from Korg. So I can control volume, start and stop, and I can dump the clip out. Those are basically the three things I do with it.

And then I can zoom in on the waveform if I want to go back. And I can, you know, just by watching it when it plays, okay, we're going to go back five seconds, click here, and I'll start there. That's what we do with No Agenda all the time. So I tell Gemini this is what I want. Surely you can run that in Wine. I'm not kidding. I mean, Wine is really good these days. Screw Wine. I mean, the whole point of getting away from Windows is not to run a Windows app on Wine. Okay, you're a nativist.

I'm going to go native. I'm going commando, baby. Fair enough. I do not want to be running crap in Wine. And by the way, audio stuff notoriously iffy when it comes to Wine. So I just didn't want to go that route. So I say, you know what, here's what I want to build. Within an hour, I replicated this whole system using, of course, Python, using the QT UI framework. QT, not GTK. Interesting choice. No, QT6. Very, very, very slick. It draws the waveforms. It does, I mean, everything.

Then add to that Mido for the MIDI connection. I mean, I literally said, okay, now I want to be able to map these buttons to these buttons on my MIDI controller. Boop, boop, boop, boop. The code spits out. I run the code. And I can right-click on a button, and then it can learn the MIDI control, which is easier, by the way, than the actual M-AIR list. It's easier to do it in the system that I built.

I mean, so basically it's using G-Streamer as the audio engine, so I can mix these different things together. The long story short is within a week or two, I will be switched over, and I guess I'll open it because I made a repo. Woo, look at me. Did you make a repo? Are you going to open source it? Not yet. Oh, yeah, of course, of course. I want everybody to enjoy it and for people to improve upon it.

But, I mean, it's even better than what I had because the way I work, and maybe not everybody is going to work this way, so I have a folder tree, a file tree open in this M-AIR list, and I categorize my clips by topic, and then there's JCD clips, and this is kind of the same for any podcast I do.

So now another thing that was not possible with this other system is I can just scroll down to a clip, or if I want to use the mouse, just click once, boom, it occupies the next available player, and I was using the Everything Search engine. I don't know if you're familiar with that, the Everything Search app. No, I don't know what that is. Yeah, it's an open source search app. It's really good. You can just say, okay, I only wanted to index audio files.

And so I've built into this file viewer a quick little shortcut. You type it in. You get all the – because I'm always looking for an old clip or something or a jingle or something like that, and it populates the list. Click, boom, it's loaded super fast. And so I'm doing all this stuff, and this is within an hour. I mean sliders are working. Faders are working. The mix is working. And then all of a sudden Gemini is like, oh, you ready to build the cart wall? I'm like, holy crap. A what?

A cart wall? Cart wall, yeah. This is where you load up a whole bunch of stuff like – Bombshell. You can just – you have your little – Oh, like a soundboard? Yeah, it's a cart wall. It came up with the name cart wall, which is correct.

And I have to say I was well impressed at the speed, at the context, window size, the cross-chat understanding of – because at a certain point it says, yeah, this will work great on your Surface Pro 6, which was a whole different window that I've been working on that with.

I'm pretty impressed to be able – now, I still don't – I mean, basically they're paying me extra money to do it because I don't see how my free account or even $20 a month can in any way compensate for the cycles that have been used making this. But holy mackerel, this is – it's been exciting. It really has been. And so to say AI has improved my life in this area, yes, yes, yes, yes. I need a screenshot of your new app. I want to see it.

Okay. I want to see what you've built, this in-air list clone type thingamajigger. Yeah, I want to see what it looks like. Thingamajigger. In fact, I'm going to call it the podcast Thingamajigger. What is that? I want that to be your repo name. The Thingamajigger. I got to write it down. The podcast Thingamajigger sounds great. Well, so my own dive in with – so I've made – I think this is the natural progression that people go through. You start with a chatbot.

Then you start in the chatbot world. Then you go to in your IDE, whether it's Cursor or Juni in JetBrains or whatever you're – whatever this is, Copilot, some sort of add-on that's living in your IDE programmer tool with you. Then at some point, you jump to just the command line. You just have an independent agent that's just going to – lives in the command prompt on the system because that way it just has full access to everything. It just understands how to do everything on the system.

I made that plunge with Cloud Code. The first thing I had to do – there's this thing at my day job, and I'll describe this. There's an app. It's a production – there's two. There's two apps. This is going to be interesting for a number of reasons. There's two basically identical apps. They communicate with an in-house custom system – platform that we wrote. All it does is takes a file and uploads it with – it communicates with an API and uploads this file to the API.

But it does – for reasons that are largely unimportant, there's two different routes this thing can take. There has to be two separate apps. Trust me, there has to be two. There's two separate apps that are virtually identical except for a few different – there may be a hundred lines of code that are insignificant or they're different between these two apps. What's always bothered me – these apps were written maybe seven years ago. They're .NET C Sharp apps. They run in Windows.

I wrote them in Visual Studio 2019, I think. Yeah, a long time ago. So I wrote them in Visual Studio 2019, and the way they're monolithic. So they have the user interface for the app is basically just some configuration text boxes and a log window just so we can keep track of everything. But these things are monolithic, so they have to just stay running on an open desktop all the time. So what I wanted to do – I wrote them that way, and I just wasn't thinking about it.

I just needed to get this thing done and blah, blah, blah. You didn't realize you had to dedicate a machine to it full-time, basically? Yeah, and it has to stay open on a desktop all the time. So really the only thing that – And this is running some Fortune 500s accounting? Is that what you're telling me? Yeah, maybe. But the – so what I've been needing to do for, gosh, for years now, what I've been needing to do is spend the time to split this app into two parts.

Yeah, a backend that just runs and then a UI. Exactly, exactly. A Windows service and a UI that'll talk to the service. And it's just like – it's not – that's not a hard thing to do, but it's going to take me 10 hours to go through, create the new repos, copy-paste over the code, do the testing, then create an IPC between the two. What's an IPC? Inter-Process Communication Channel. So like a named pipe or something like that where the front end can talk to the backend. Just use XMLRPC, man.

It's the future. Soap. Soap. Soap. Yeah. Inside joke, everybody. Yes, right, yeah. So this is my – who can peel off 10 hours to recreate a thing that already is working? Right. And you're not going to get anything for this. All you're going to get is once every few months when this thing crashes for some reason, you have to go restart it. And that's the only thing this does. So as usual, I can't just commit half a week to doing this.

Right. So I was like, okay, perfect, perfect trial run for Cloud Code. Yes. So I grab a copy of the repo, stick it into a directory by itself so we're not working on the live thing. And I'm like, okay, I described to Claude this is a C-sharp.net app that is monolithic, and I want you to split it in two parts. I want to have a UI that can – excuse me, a backend Windows service that can be installable and then a front end that communicates with the service to show me the logs and blah, blah, blah.

So I just gave it a simple, maybe like a three-line prompt. Fifteen minutes, it was done. I copied it to the server, launched it, and it worked. I never touched a single line of code. Line of code, yeah. And the thing that got me – and it also said, okay, I split everything. I split it into three repos, a service repo, an IPC repo, and a UI repo. And I also created you two batch files, one to install the service. You did that without even asking? I never even asked for that.

And then I opened up the UI, and it took – it's the same UI I had before, but it stuck a new thing up there at the top showing me the status of the service and a start and stop button so I could restart it myself. That's great. I was blown away by this thing, but here's the kicker. So this was perfect. So I'm like, okay, this is great. I'm just going to do – I'm going to take this same prompt and do it on this other repo, which is basically an identical code base.

I'm going to do it on this other repo and just change the name and say, do it to this other repo. I did it. Completely different outcome. And this time it did not split it into three repos. It split it into two. They were differently named. It did not create the install service batch files. I tried to – so I created the install service batch file by hand, installed it, and installed. It gave me a failure, but it still installed and started. And then I tried to run the UI.

It would never connect to the backend. It was pretty much broken. This is the frustration. This is exactly the frustration that I've had with – and was it in the same context window or did you open up a new window for the – a new context, a new chat, or whatever you call it. I don't know how that works with Claude. Yeah, a new session. Yeah, a new session. Yeah, this is the frustration. The consistency is nowhere to be found.

I don't understand if it – because I'm telling you these code bases are virtually identical, and I don't know if there's some sort of temperature that's involved. No, temperature is – temperature, I believe, is exactly the issue. Because it shouldn't be. No. The temperature on that sort of prompt should be zero. I believe it's the actual temperature in the server farm. Okay. I believe that actually – yeah, yeah, I really think so. I really think that makes a difference.

You know, the cooling was half a degree off. You get different results. And how frustrating is it? And you can sit there and you can tell it 100 times. It will not create what you had in the previous session. So the way I got around it is, as I said – so Claude does this very helpful thing. I'm not sure if Gemini does this or not. But one thing Claude has that's very helpful is when it creates – it'll go into what it calls plan mode. Yes, yeah, the boardroom is talking about plan mode.

Okay, so plan mode, it'll take your – you can go back and forth manually, but a lot of times it'll understand it needs to go into plan mode automatically. So it'll go into plan mode, and it will essentially write a plan of what it's going to do. And it will stick to its own plan. It will, yes. And so the thing about this scenario is that Claude leaves those plan files behind. So what I did was I said, okay – Take these plan files.

I started over, and I said, look, I added an extra little bit of prompt at the end and said, I've already done this conversion on another repo. Here's the plan file from that previous conversion. I want you to follow the same exact plan file as closely as possible. Does it leave the plan files in the repo? Is that where it leaves it? It leaves them in the Claude – like in its working directory. Okay. Outside of the repo. And this time it did it, and it did it exactly right. Wow. Perfectly.

Wow. I feel like it's – so the sense I get from this is that it's sort of like the experience you have when you're doing the image generation stuff where when you get a prompt, and it gives you something really good, it's like you want to save the prompt. Change this and add this. No, no, we can't do that. No, it doesn't work. It doesn't work. If you ever get a good result from a prompt, you need to hang on to that prompt and keep it forever.

So that was my first out-of-the -gate experience with Claude, which I would say was highly successful. I mean, it saved me so many hours of work on a thing. Now, was it additive? Not really. Well, it gave you the service start and stop button. No, it did. It did. Yeah, no, you're right. You're right. Well, like in your case, you were already on Windows. Mm-hmm. You've mostly replicated what you had before. The agent helped you do a sideways move, essentially.

Well, no, no. It has actually improved because all of the interface issues that I had with mAirless, which I would have loved to have it work differently, I just said, okay, implement this this way. So, for instance, in mAirless, if you want to do a pre-fader listen, so you want to listen to something, for instance, end -of-show ISO.

So if I want to hear an end -of-show ISO, and so I'll get it from the ISO bot while you're talking, I'm listening to you in one ear, I'm pre -fader listening to this ISO that I'm trying to get, and I set an in-point and an out-point because I don't have time to actually edit the file. So that's a whole different window that pops up. There's a whole different subsystem of mAirless that I'm working in. It also prohibits me from doing some other things with the live play-out system.

So now it's baked into it. So when I go to pre-fader mode, it just turns that one player into a different color so I know that it's in PFL mode, and then I can click in the timeline, set my in and out points, and I'm done. There's no extra windows popping up, none of this stuff. And I've also made it so I can select which interface I want the PFL to go to.

And all these little improvements, and then also the clip search system, that was two different applications, so I've baked that into one now. So it's definitely additive for me. I think Leo Laporte is right in his characterization of this stuff as the real thing that is happening right now is an unlocking of the ability to have hyper-personalized software. Yes. And it is enabling individuals like myself who have a concept of systems.

So I understand Linux, I understand command line, I understand Bash, I understand some Python, I understand the basics of the computer. So the networking, the audio, I understand all these core principles and I've had interactions with them, albeit very poor interactions throughout my life, usually with destructive results, to be able to prompt the right things. So if someone just sits down and said, okay, I want to have a podcast playout system, I doubt it would come up with anything like this.

It doesn't really exist. Even the way I customize mAirlist is not the way anyone else has customized it. And so we may see very small companies that provide software products or services that they can spin up really quickly. So, I mean, like us, like Godcaster in a way, I mean, I've created two or three products that would not be here because we only really have one, one and a half, one developer on the backend, that's you.

And so I've been able to spin up these different products and optimize them over time just by knowing what the customer wants, knowing what I want to build for the customer and having enough understanding of the systems to be able to coax the AI into coding what I want. And the actual coding part is just, it's like a layer on top of it. Yeah. Yeah. I think, like, I'm afraid to let it touch existing code bases that are important. Like, I've not used any AI on the podcast index API backend.

I'm talking about the API server itself, you know. I have not let it touch anything in that. I haven't let it touch anything in Godcaster. There's code bases I'm really afraid to let it mess with. Sure. And that's understandable. I'm just freewheeling my ass off here, brother. I'm like, woohoo! Sell it! Yeah. People will buy this. Yeah! I mean, like, the thing, but it's like, I think I'll get there one day when I sort of figure out, okay, I can bring it in to help me do this. Well, okay.

For instance, the other day I needed, this was for something I was doing in Godcaster. I needed a PHP function that would just take any string as input and just return nothing but an, and strip out any characters that were not alphanumeric. Super simple. I've written this function a hundred times. It's not hard. It's like one line of code. If you, you know, if you do it right. And so I was like, well, but like, it was faster for me to just ask Juni to write the function. It did.

It was like, boom, done. And I copied and pasted it in. Yeah. Like, I can see it. Like, it's kind of putting out fires. It was easy. It was quicker for me to have the thing write it than to go hunt down another repo, copy and paste the thing I wrote out of it. Yep. That was faster. So I think the other, the other thing that I observed with Claude in this, I'm going to take this back to Silicon Valley, just grossness. This is just the Silicon Valley.

Anthropic, I think, are they Silicon Valley or are they? Oh yeah. Oh yeah. Okay. Okay. This, this is their way. So I did the $20 a month plan. They got your money. We got one. We got one. And it would. So the first, that first repo that I asked to do the conversion on that windows app, it used up like 80 something percent of my, of my usage window. Oh, okay. Not of your credits, of your money.

So the way this, the way this thing works is you have, you have time slices where you can, you have an overall credit amount that lasts for like a week. And then you have, excuse me, you have a credit amount for the month. Then you also have individual time slices. There's one that's a week long and then there's one that's like hours long.

And if you, if you bump up to your usage window, uh, uses limit for that window, it'll come back and say, you know, Hey, you've used almost everything for, uh, for this session. You're going to have to wait until one o'clock for things to reset. And so the, the, the effect this has, they give you two options. Now here's your options in the, in the, in the, in the app is you can either stop and wait for your, for your time window to reset or upgrade or upgrade exactly.

Or you can upgrade and the next plan is $200. It goes from 20 to 200. But see, this is what I've been saying. And I, I don't think that's a Silicon Valley ickiness. It's, they've been giving this away. They've not, they've been losing 90% on, you know, on, on their, their cashflow because it, you know, for every dollar you spend, they're giving you $10 worth of compute.

So they have, they, and 11 labs has figured this out because we use 11 labs for our, uh, for our, the, we call it your town live. So, um, you can do a couple of different things. You can load up the podcast. You can have the most recent episode from a, from a podcast play in use in this live 24 hour stream. It does music. It does a station IDs, promos. Now most churches don't have somebody to voice promos or do a local news report.

So I built a little quote unquote studio and, um, you know, and so you just copy paste some texts in there and it's actually pretty good. I mean, the, the voices that I've, that I've selected for this, they can use it. Customers can use any voice they want, but works pretty well. And so just from running a few different, uh, uh, servers like yours, Gordon, uh, hello, Fred, uh, hello, Franklin is still on it.

Uh, although paying customers now use their own API key for 11 labs, you know, it's like all of a sudden, Oh, it's 140 bucks a month all of a sudden. But, but it's fair because you are using those compute cycles. Now, what I'm personally now thinking is, well, in two years, maybe less when all of these Nvidia cards have to be replaced by the hottest new Nvidia thing, I want to go buy a couple and use them at home. Yes. That's, that's, that's the ticket.

I was talking to Alex last night and I made the same point. I think the future of the, of all of this stuff is clear. And this goes back, this is kind of what I was getting out with the Silicon Valley ickiness. The future of all this stuff is it's going to be local. You're, you're going like the, everything is running to my hardware right now is going, is moving from discrete GPUs and then a system on a chip. Excuse me, a system on a chip with a discrete GPU connected by PCIe.

It's moving to unified memory with an, with an APU or NPU built on the package. Like Apple stuff, you mean? Yes, exactly. It's going to, so the new Ryzen hardware is that way. The Nvidia hardware, like in the DGX Spark and that it's also that way. And you'll, the future of hardware in general is going to be, you know, 128 gigs of RAM on an SOC and it's just unified memory that's shared by the GPU and the, and the CPU.

And when we get to that point, like, so one thing, one thing I've been doing is I've been testing a whole bunch. We have for a project at work, my day job, we have, I've set up a Olama server and we're testing some, some different models for some things. And so the, the model that I've been testing with right now is GL, GLM 4 .7 flash. That model is fantastic. It is amazing at generating code.

Like if, if you, if you had 128 gig of RAM, unified memory, local machine running that GLM 4.7 model, oh, you could cook man with, with open code, which is like an open source version of cloud code. It supports plugins just like cloud code does. Like you, I think within two, two to three years to local LLMs that rival what Claude and them are doing are going to be completely accessible. And isn't this, we were actually talking about this last night in our group chat.

Isn't this where, I mean, the whole world is always centralized, decentralized, centralized, decentralized. With this most recent 365 outage where the whole corporate world had no email. Yeah. I mean, come on. It's like, it's like a no brainer, but we've, the whole internet was built for decentralized use. And they were like, Oh, the cloud, put it in the cloud, put it in cloud, cloud, cloud, cloud, because they didn't want to pay Dave Jones to maintain the servers.

No, it'd be more, so much smarter. Look, I got the, I got the pitch from Amazon. But then when it goes out, like the whole world falls apart. Like the lie, the lie of the cloud has been, and it's been the lie, the lie of, of big tech from the beginning. Do you remember, do you remember when the cloud push first started? The big talking point was in this, as an IT manager or sys admin or anything like that, you heard this ad nauseum to your blue in the face.

You heard the phrase from sales guys, tech sales guys. You can, this will get moving to the cloud will allow you to convert from CapEx to OpEx. Yeah. Yes. Except your OpEx is X10. Yeah, exactly. But, but that was a lie because the, you, you converted to OpEx, which only just means a month, monthly billing. You converted to OpEx and then in the last seven to eight years, all the cloud providers have switched back to, you have to buy a year in advance. Yes. So it's not back to CapEx again.

They lie. They're liars that we saw this with, with Intuit and QuickBooks Online. They were giving that away, that product away, giving it away. I mean, you could get that thing in the early days when they first launched it. You could get that thing for nothing. It came free. It came free with your, with your tax, with your TurboTax. Yeah. Yeah. Buy a pixel and you got a free, you know, it's like free three months. Yeah. You could get that thing. Now it is through the roof expensive.

Yeah. Through the roof and it goes up on a, and the price increases on a predictable schedule and you're, you're now you're in the treadmill. So that whole thing and, and the, the way it was being sold in the early days is, you know, Hey, switch to the cloud version of, of your accounting softwares. Cause it's so much cheaper. Big tech will lie to you all the time.

And I think that the story of, I think the story of AI is going to be that it is going to empower people, coders, because I don't know that AI is going to be super useful for much beyond code, you know, and tech oriented things, general knowledge stuff. I mean, it's still just kind of lame, but like outside of, you know, that special use case, it's going to allow technologists and coders and those kinds of people to hyper personalize and bring their stuff back out of these big tech stacks.

Yeah. I mean, this is going to be a big deal for that. N8n, you familiar with N8n? Oh, why does that sound familiar? It's a, it's a AI workflow builder. Okay. So, so you can connect a model to, you know, to a, to a cloud model, something local to some other service, you know, it's you basically draw a little flow chart and of course you can buy their service so we can get the open source product.

But I just see in general across all types of services, we have to move away from these gigantic services. I mean, CloudFlare will be the last one to go. And that's a complicated one just because of the nature of, of web traffic and, you know, just caching. It's, that's complicated. But why in the world would you rely, I mean, this latest 365 outage being because we saw some bounces coming back from Marriott. Okay. Marriott. Oh, it was everybody. Yeah. Yeah. It was everybody.

I mean, this is when I, you know, when I tell Tina, it's like, blah, blah, blah, blah. Like the, the, the 365 took out corporate emails. Well, like Marriott. And then she's like, she realized, Oh crap. Well, that's a problem. Yeah. Yeah. It's a huge problem. This, this stops business dead in its tracks. You know, it's just, it's, it's, I just don't think it's useful anymore. And of course ultimately we have to eventually, I hope go back to allowing people to run a small service from their house.

You know, this, this will be problematic because of the asynchronous nature of cable, internet access, et cetera, et cetera, et cetera. And legalities and all this stuff. But we have to decentralize it eventually because these big outages will only get worse and worse and worse. And life will just come to a grinding halt every single time. I mean, I could, you know, the good thing is like most, most people have available to them at their home resources they don't even know exist.

Yeah. Like a, a corporate, you, you could probably call your call AT&T if you have AT&T fiber or some kind of cable fiber or whatever, you can probably call them right now and get like a quote unquote corporate corporate plan and get a static IP and all this kind of stuff. And you could just start serving things from your house. Well, even, even with Tor, because that's what Start9 uses and Umbral as well. I am perfectly happy accessing my Start9 because I run a, a NextCloud on it.

It's fine. You know, I start up or bought or bought onion thing, whatever that is on my phone. And it's like, Oh, okay, this working just fine. You know, I can give somebody that address and then they can access it. You know, it's not like complicated, uh, nat traversing and all this stuff. And of course we have other ways to do that as well. But for simpletons or noobs or whatever, plebs, whatever you want to call them, you know, that type of system that you can run yourself for your family.

Now, of course, email is complicated because of SMTP blockage. And there's all kinds of issues. But man, I love void zero who has been running my email server for over a decade. Then it is, I just love the fact that my email works, you know, and, and we can stop certain things. We can hamper things.

We can do all kinds of stuff to make it customized for me, for the way I want it to run and what I want it to allow and not allow, you know, and, and I don't know, it's just, it's, it seems like we're, we just went down this path and we all went nuts. And so AI, yeah. I mean, you're absolutely right. It's still too expensive, but it'll, it'll start becoming available. And I'll just put something here at the house and I'll just use that.

Well, I started, you know, I, I tried to make this point a while back, and I don't know if it's true or not, but it feels true to me that, that I feel like, you know, a lot of the expense of the local AI components right now, it's not just an accident of the market being in a bubble. I think it's also somewhat intentional because like Ram is incredibly expensive right now. You know, the Ram shortage is nuts. Yeah, it is bad. Who does that help? Big tech. Yeah, of course.

Because they're the only ones that can afford it. I think it's beyond just a coincidence that this, that these things are very, these components are very expensive for home users to get their hands on because if they weren't the demand for big, for the big hyperscaler AI providers, I think would vanish. I think that's part of the moat is the, is the bubble induced expense. I think it's just poor planning. I don't know. It's sees more like supply chain, like people just buy.

I don't know if that's on purpose. It's just the way the beast works. I think no one was thinking about it. I just, I don't know. I have a gut, I have a feeling that the, that part of the whole site, you know, incestuous financial cycle between the big AI providers, the cyclical finance is to push the price up where it's out of reach of anybody that's not them. I think maybe, maybe. If you had, if you had, if you had a DGX spark level equipment that was 1200 bucks.

Yeah. Oh, I mean people would be buying them like hotcakes, but because they're $4,000 most people won't. Yeah, but it's, but their whole business, listen, Silicon Valley's business model online is advertising. That's so they're in a whole new, the whole new world that, you know, like, Oh, we got to actually charge people for stuff. And yeah, people will pay 20 bucks maybe for the chat bot, you know, but that gets old. It'll be the first thing to go.

I think the world of software developers is quite small on a, on a, on a scale. So they're always going to try and find things that will, you know, the parlor tricks, you know, like art generation and song generation and stupid stuff that, that, you know, that will, that it just gets old where they're, where they're still trying to figure out what it's a, it's human level intelligence. Okay. Artificial intelligence to me is like artificial flavoring. Tastes good. We'll kill you.

It will kill you at the end. So anyway, how does this fit into podcasting? Yeah. I don't know that that part does. Well, other than the fact, well, well, uh, CC Lawton created a podcast index MCP server. Yes. Which, uh, yeah, we, we didn't have a show last week, so we didn't talk about it. Have you, uh, fired this up? No. I'm going to put the, uh, I'm going to put the GitHub issue talking about it into the boardroom. Um, I have not had time to fire it up.

Um, but he came back today and said he wrote a sample app. Oh, sample Android app that uses the, that uses this. Oh, wow. Did he email us both about this? I haven't seen. No, he just posted in the issue. Oh, okay. In that issue as a, as a reply. So it would be good to have people jump in and give this a whirl.

If, if anybody has time to do it, uh, it's still on my to do list, but how, how does it, my, my question is, since I'm not familiar enough with MCP to know what the backend of the MCP looks like is, I'm not sure how it is translating for those who don't know what MCP is, it's model context protocol. It's basically a, it acts as a proxy between an AI agent and some API based service somewhere that you want to get information from.

So it, it tell it, it allows the, if you ask a question to an agent about podcast index data, having an MCP interface to the AP to the podcast index API would allow the agent to call and ask for that API information in a way, in a reasonable, sane way. It can, you, it can intelligently talk to the API in the way that the API wants it to. The question I have though, is that I don't know how those translations play out in practice.

So if you, are there scenarios where you, where the agent would just blast the API with sort of like trying to grip it? Could it just flood the API with, you know, a thousand calls a second? I don't know. So that's kind of what I need to kind of get my head around is what, given a certain set of inputs from the agent on the front end, and how does that translate into API calls on the back end? What, what does that look like?

Speaking of that, Lacey, see losses love sending a boost in the split to podcast index and get albie.com aired. Well, the, the, the Albie hub is up. That's not, that's online. No, but I'm not seeing anything. In fact, nothing is coming except I, I just got a 22 sats from the, uh, from IPFS podcast thing. The podcast index Albie hub is up. I'm, I'm, I mean, it's up, it's running right now. But just cause it's running doesn't mean it's receiving. You don't know. Evidently. You don't know.

Evidently. Um, but anyway, that, so that's, that's an interesting thing that I think that we, you know, I think would be a nice thing to put out there into the world, that MCP server. Well, I think, you know, to expand upon that because we kind of have some unique opportunities with the podcast index within an MPC server like that. If, if somehow you can have another service that can be queried that accesses transcripts based upon the availability of a transcript, et cetera.

Um, then you can start to get some pretty interesting context searching for discovery. And I can see some apps wanting to possibly, uh, experiment with that. So this brings up, this brings up the, uh, the idea I had. That's a perfect segue. I'm the segue king, baby. Yeah. Um, but by the way, on the get hub, on the namespace repo of the get hub, I, I, I've turned my email notifications back home.

I turned them off a few months ago cause I was just getting flooded with stuff that like just people griping about things. I just could not, what eating up all my mental people just griping about stuff. Oh no. So, but, but all that's calmed down. The, the namespace repo has calmed down. So I, I've turned email notifications back on and I'm going to get back more engaged with, with that. So that's what that was about.

Um, the, uh, James had mentioned that it was kind of a ghost town over there and I was like, and I'm like, yeah, cause it's cause I'm not getting any emails from it. Um, but that's back on. But, but anyway, the, the idea I had was, um, we need a standardized way to attach our own metadata to a podcast that does not supply it. And so the, the reason this came up for me was that, um, I subscribe, I think I've mentioned to you before. I'm, I've subscribed to two shows from Twitter.

These are, uh, it's windows weekly and security now. And the reason I, and I listened to these shows religiously because, uh, they apply both to my day job. This is, um, when I listened to these podcasts, it's work, not pleasure. When I listened to them, it's work, not pleasure, but that's a different reason. It's for a different reason.

Um, and so the, the, the problem that I've had mostly with security now, not with windows weekly, but the problem I've had with security now for quite a few years now is that, uh, there's a, so the nature of that show is that Steve Gibson reads, he reads like entire articles on the show. And sometimes it's about a topic that I'd really just do not care about. It doesn't apply to me. It doesn't, it doesn't apply to my job. And I really, really would like to skip over it. The chapters.

Well, twit doesn't has never published chapters and evidently because it's podcasting 2.0 and those guys are racist. I think even before 2.0 chapters, he was like, I don't want people skipping ads. Oh, that's a point. Yeah. So, but then what happened was, you know, I subscribed because I want to see those two shows supported. I want them to continue. So I subscribed to the private feeds. So those are ad free for a subscription per month amount.

And I was like, Oh, okay, well check, uh, maybe they'll put transcripts and chapters in these. No. So the private feeds do not have transcripts or chapters either. And I think it's because they're using simple cast, which does not support those. Okay. I think. So it could be a technical reason, but it doesn't matter what the reason, the long and short of it is that they do not support it.

And so what happened was Apple released their automated transcripts and then recently released Apple podcasts, released automated chapters based on the transcript. This was awesome for like three weeks because it was awesome and also unfortunate and weird at the same time. Because what happened was, even though I'm a paying customer that subscribes to the ad free version, suddenly the non ad free, the, the one with ads in it automatically had chapters.

Hmm. So I was using Apple podcasts for a few weeks to get chapters in security now so that I could flip past all the stuff that I didn't care about. And it was so nice. Wonderful. So I'm like, well, I guess, even though I'm paying for it and the private feed doesn't support the things that I want, at least I'm getting what I want now. You feel, you feel good about yourself. Yeah, yeah.

Well, of course, this was not going to last because Apple podcasts allows you to turn that off in your Apple podcast connect. And they clearly got in there and did that because suddenly now the chapters are gone. So because of, because of the ads, because of the ads that I don't hear anyway because I already paid for it. Yes. All those for all those seasons. Yeah. All those reasons. So, so this got me think this has been annoying me to no end for so long. It got me to thinking.

I already like where you're going with this brother. I can feel you, but I feel it in my water. He feel it in your water. It's a Dutchism. Okay. Okay. Fool dumb of Arthur. Yeah. Uh, so the obvious, the obvious jumps into everybody's mind way to solve this. And I don't think it's the right way, but it is the one that everybody's going to think of first is create a second feed, create a second feed that references the enclosures of the original feed and put in your own chapters and transcripts.

Mm. You're basically creating a proxy feed, almost like feed burner. Right. Um, of course this is a, for so many reasons, this is a bad solution because you're, that feed is going to get out in the wild. Now there's going to be another feed that people are subscribing to. It's just going to be a mess and you don't want to get into that. A bootleg feed. Yes. Cotton gin. So I just don't want to do that on principle. Right.

It seems like though, instead we have, we would have the op, we have an opportunity here to create a standardized interface that podcast apps could accept. So a podcast app could allow you to put in a, a, a, a, a, the URL of a API that you run in some sort of way, right? Not even has to be an API. It could be something else. I don't know yet. A thingamajig. But it could be a thing, a podcast thingamajigger.

And the app would simply call the thingamajigger whenever it got a new episode and ask the thingamajigger for amendments, metadata amendments to that episode. So let's say Castamatic downloads an episode of security.

Now it takes some, some, some universally identifiable part of that episode, most likely the episode of the enclosure URL and then passes it to the URL of the API that I gave to Castamatic and asks Castamatic for, and asks my API, is there anything that goes along with this episode or anything you want to do? Like, is there anything I need to know about this episode? And, and that API of course will be run by John Spurlock as all, as all extra bits are. Yeah. This is a, yeah, this is a JS.

JS project. Yeah. Johnny player special. So this keeps everything private. There's no public bootleg of a feed that's out there in the wild and it's something I'm paying. So heads, heads up, Daniel J. Lewis, this is aimed squarely at you also because podchapters pod.

If I pay, for instance, the way this could work using Daniel J. Lewis's podcast chat podchapters service is I could ask, I could put it, I could have an account, a podchapters account, and then get an end point specific to me, add it into your app, add it into my Castamatic. Then now every time Castamatic downloads a new episode, it takes the enclosure URL and asks podchapters, my private podchapters end point asks it, is there anything, is there anything else you want to add to this?

And it could be just, you could return just XML. It could just be extra XML that comes back in and gets sort of virtually merged into the feed item with trans. It could, so you, you're going to get back from the podchapters private API end point. You're going to get back a chapters tag and a transcript tag and all these kinds of things that you paid for.

I would be paying for these things to be generated, but the app needs to needs to know a standardized way to communicate with the service in order to call it and get a response. So here, here's the obvious issues with a great idea. I am all for, because I had to re realign my heart with people playing my podcast at 2.0 speed with people using silent skippers, because I feel that that's my art and this is, this is how I want you to consume it.

But of course I have no, I have no say over how someone wants to consume my art. They may burn it to a CD. You know, there's all these different things you can do. The minute you introduce a paid service, because this is no different than the the, the app I think that James was talking about on pod news today or the past two days that skips ads, that reroutes around tracking URLs, including OP3. So to have that personally and to use that personally completely valid.

The minute this is a paid service, that paid service has a huge target on its back because the publisher I feel has right to say, Hey, you're taking my content, adding on some value to it that diminishes my value. Do you understand what I'm saying? Because it will, if the minute you have this, one of those extra things in the API will be skip the ads. It's, it's, it's, it's clear as day. So if you have that yourself, or let's say there's an open source app that does that, nothing can be done.

And it would be up to the individual like you do like, well, I'm paying for it over here, so I feel good about it. But that will be 1% of the people. So in theory, it's a great idea, but I just don't see how anyone can implement that in a commercial way such as an app, even if you're taking donations or, you know, you have an extra, uh, full paid version, et cetera. I think that's just asking for trouble. Yeah. The, I think the concept works.

Yes. But the, the podcast, the pod chapters example of a paid service may be the wrong example. It could, it would need to be something like it would just need to be something that I, that you provide on your own privacy, which of course, no, which almost guarantees this won't happen. No, you will, you will use it and just use it. Yeah, this is going to be, but there's tons of examples out there of podcasts that just for whatever reason, they're just not giving you something that you want.

And there is just an annoyance. And if you had a way to sort of amend the feed in a private way, because you know, if you know how to do it, you're going to do it anyway. Because like what, what I will do, I will absolutely do this.

If, if, if, if Twit does not work out some way with Simplecast or whoever their, their private feed provider is, if they don't work out some way in the near future to get some kind of chapters or something into these shows, I will just write my own podcast app and just do this all myself. Yeah. And I think that's, that's what it comes down to. Because, and it's just sad.

It's sad that we are in this, we're in this world, the podcast industrial complex, you know, it's like the, and it's also a mindset, you know, it's like, Oh, I have to have ads. It's the only way I can make money. Whereas I'm sure, well, no, I, it's proof positive that you pay, you will gladly pay for windows weekly, you know, you'll gladly pay for it. But this whole, well, I mean, windows weekly should just add these things to their ad free feed. That's the answer right there.

Well, that's the answer. That's the answer. And they can use a commercial service like pod chapters for that to do it. And they should. Or may, you know, or maybe this particular feature is only available for private feeds. That would be another way around it.

Like somehow it only, you know, if, if you're going to have like some, something like, you know, something like a Daniel Day-Lewis's service or whatever, I don't, you know, I don't want to not be interested in any of this, but something like that service could say, okay, we're, we, we will provide a, you know, amended, amended XML information based on your own, like, you know, your own backend processing or whatever you want to add to it.

But it's only going to be for paid like subscription podcasts. Like you got to prove that you're paying for it first. Yeah. You know, so it'd be like simple cast or whatever. I don't know. I just feel like I hate, you know, I hate solving problems for me in a one-off way and not like it. I always want it to be like a bigger open public solution to the problem that other people can do themselves. It's kind of annoying to have just some private thing you've rigged up that nobody else can enjoy.

Cause if, cause if I'm having a problem, I know that other people are having the same problem with other feeds and that kind of thing. But I don't know, maybe, maybe you're right. Maybe there's no way to do it in a way that makes, Leo should just think about his customers. That's all. He doesn't care about his customers. That's the bottom line. If you're offering an ad free feed, add in the bits people want. Yeah. So maybe, maybe it's not that. Yeah. Maybe it won't work.

Just before I, as I remember, I heard James talking about he's going to write up a new issue or proposal for podcast comments because I guess at a podfest, everyone discovered that, wow, people really like the comments on YouTube. So we have to have comments on podcasts. So be on the lookout for it is what I'm saying. And we already have comments that work. Oh yes. No, it's just going to be a Jason comments. Yeah. Well, okay.

I guess I'll wait and see what he does, but I think Daniel, Daniel J Lewis had a pretty comprehensive Jason comments proposal already, you know? Yes. Yeah. I think he already wrote back in software and everything for it. I'm just in general, just not an agreement that that people desperately want comments on podcasts. I'm kind of over that. Chat is fun. Comments that live forever. Not so much. The YouTube, the YouTube comments, it's part of the entire video experience.

You know, you're interactive with the screen. I find podcasts are just passive. Comments on YouTube are, are typically real time comments is really what they are. Live and live streams. Yeah. And then they suck and they suck. Well, what happens is on these live streams, people get in there and they just treat it like a chat room. They're talking to themselves. They're not even really interacting with, yeah, it's like, it's like an ephemeral chat room.

Yeah. So, but anyway, I just want to make sure everyone was on the lookout for it because it's coming. But yes, I think there's many, many comment proposals slash et cetera, chat proposals that have been put out. So, well, the other, the other thing that's continuing and I'm actively working on this is the, you know, AI slop and quality control stuff. And, there's a big thread on the namespace repo. We really need a better, we need a better place for these discussions to happen.

The namespace repo is there's all kinds of stuff in there that has nothing to do with the namespace. And that's one of the reasons I turned off the emails about it is because it was getting like, people are talking in there about things that are like index related and stuff. So anyway, it's a big mess.

But the, um, but the, the other thread on there is just about the David Marzal brought up broadcast quality control and, you know, how to in the AI slop and all that in the junk feeds and everything. So I'm continuing to work on that. And I've almost got, I've got everything working. I just have to put the, uh, the output side of it on. So where people can pull out the, the mark, the mark slop. Cause I've been marking it. I do about, probably do about 10 minutes a day total.

I just have it running and what I'm really, I have it running in a window. I'm like, Oh, Kanji characters. Let me check. That's kind of, that's kind of my, my main, uh, like, Oh, Kanji characters. Let me see. Okay. Slop, slop, slop, slop. Oh, I'm seeing so much duplication so much. It's amazing how much is just spam or slop. It's an amazing amount. It really truly is. I'm just, it's abusive. It's abusive. It's like, it's so sad people do that.

I think, I think the general sentiment about we, you know, and you, and you had said this too is we should, you know, we should not be blocking stuff. Um, from, uh, from being put on the server as of, as of now, we should not be necessarily blocking stuff. Uh, but the, there should be some sort of like scoring system to help identify these things. And I, and I agree with that for rent for now. It's okay. It's running fine.

You know, there, there does, it does cross the threshold though at some point where it becomes sort of a, of, it becomes where you're having to play self-defense because if, if we begin to get truly swamped and overloaded with, with junk content, we will have to begin to make, we'll have to play self-defense to keep from having it cost us real money. And the, you know, store storage, quote unquote is storage is cheap, but not block store, not block storage.

Block storage is disk storage is expensive. And these are, remember, you got to remember these are these, these podcasts, these junk podcasts are going into the database. Yeah. They're not just living, they're not living on object storage somewhere for 0.0001 cents. They're going into a database and that is real, you know, that's like a, you know, that's CPU cycles that's in database storage never goes down. It just expands.

Yeah. So, well, I mean the, the, the, the thing, I mean, really I've only been marking two different kinds of things. It's a slop and spam and the spam is worse than the slop. The spam is, is outrageous. And it would be nice to find kind of easier ways to identify it because it's like three or four clicks, you know? So you really have to look at the XML and then you got to say, okay, do you have an enclosure?

And you know, sometimes they go through the effort of putting an enclosure in that's an actual audio or video file, but usually they don't. And you know, so it's just, there's too much, we could shorten the spam identification cycle, I guess.

Okay. And I think, I think the, the way this will work is when the feedback loop gets completed on this next round that I'm finishing is that it'll, the, the identified, the human curate, the human identified problem feeds will feed back into the LLM, which will generate description files of what, of what the signs of spam feeds and low quality feeds are. Then that will feed back into marking.

It will not automatedly market, but it will, it will feed back into like a score so that we can then try, it'll be easier to identify. We'll work through it. I do like it. I like it. I, I built it on my, on my surface so I could actually use the sound card, you know, hitting the P button to play. That's, it's cool. That is definitely cool. The interface is still a little, it's like, it's kind of slow. Not that slow. Good for me.

Cycling through with the, with the up and down arrows kind of, there's a little lag. You're right. Just a little bit. Yeah. I think it's like the, the, the UI updater loop is a little slow or whatever. It's, it's, it's vibe code, man. What do you say? Yeah, baby. Vibe code and vibe code and our way to heaven. Let's thank some people, which will be real short on my end. We got, we did get some route tests. Now, of course this is, I'm looking at the keys and receiving side.

So it doesn't mean we actually got these sats or we got 1% of them. Salty crayon, one, two, three route test. The, that was from Podverse. We got Chad F from fountain with a hundred sats. We have Eric PP. He used the boost CLI for 3333. Here's some sats salty crayon with three 33 from podcast guru. Test boost over key send. Yes, of course. Key send works. So I, yeah, it does. And, and then I hit the delimiter. So the sats have been pretty low lately. Although I did see people streaming sats.

Let me see. Let me just check the streaming sats here. Yeah. Yeah. Curio caster, Eric PP streaming away. Let me see. That seems to be, seems to be the big one. Yeah. Yup. That's what we got. Okay. What do you have on your list? Well, we got, we got buzzsprout a thousand bucks. Yeah. Hello. Shot Carla, 20 inch blades on the Impala. Buzzsprout keeping the vibe alive. Thank you very much. Boys and girls from buzzsprout. Appreciate that. This is value for value. This whole project.

We're just sitting here trying to keep it alive. That's a shot. Kevin and Tom, a note to see if they want to be on the show. It's been like a year. It's been a long time. I mean, it's been ages since we've had the boys on. I know. Let's see if they, let's see if they come back. Okay. Well, look at here. RSS.com, Ben and Alberto, that 1,001 let's say $1,010 and 10 cents. That's one Oh one. The machines keep running for another month. Thank you as well.

Dave and Adam, this is from Ben, a binary boost corresponding to the number 42. Yes. The answer as we look forward to 2026 and back at all the years before we recognized that the podcast index and 2.0 has been the answer to so many things. Thank you team at RSS.com. Oh, thank you. RSS.com. Thank you team. It's a pleasure to serve you. Thank you guys. We'll see. We got some boosts. Let's look back. I got instructions from a commissioner blogger. So I go, Oh, you have, Oh, you have been straight.

You have stuff to do. He said, do not read last. He posted on the social not to read last week's, not to read last week's boost because it could be interpreted as racist. So I'm going to skip it. Oh, I'm looking at it now, but I did read it. Yeah, we read it. Yeah, it's funny. Okay. We've got a true fan, a true fans, anonymous 4950. Thank you very much. That was on, that was for episode two 42nd is two 47 Aggie five. A tone record through fountain 2265.

And he said, Oh, this is like a, this is like a haiku. He says, middle of the night underneath desk rewiring. This episode plays. That is a haiku. It's a loco. It's a loco. Yeah. More anonymous 4953 true fans. Thank you. Thank you. Appreciate that. And here's the delimiter. Commissioner blogger. That was a short list. Very short. Commissioner blogger, 17,000 sets. Uh, let me see. Let me make sure I'm following all of comic strip bloggers. Uh, instructions. Yes. Uh, let's see what he said.

Today's Bitcoin don't, uh, don't read the one from last week. Uh, I screwed up the URL in the Bitcoin gram. It should be. Okay. I got it. So, uh, he says, uh, see 17,000 sets. He says, howdy fellow AI users, Adam and Dave. Today I want to recommend company 11 labs .io. Not only because it was found by two poles born in Poland, but also because it has apart from text to speech, incredible AI voice changer that offers quote full control over the performance.

Capture whispers, laughs, accents, and subtle emotional cues. Unquote. So podcasting anonymously is finally possible. Wait a minute. CSBA. Hold on a second. Is this real time? I don't know. I don't think it's, it can't be a real time voice changer. Interesting. Control over the performance. I don't know. Well, I'm an 11, 11 labs user and I power you. You're a power. I am a power user. I use the API and I think that's power. Yeah. Yeah. I, I think they are very, very good.

Um, they, I think they're doing good things over there. It's, it has changed a radio and it has changed podcasting, but radio, you know, for live streams, man, I use that all the time for promos and all kinds of stuff. You can create music beds. It's great. And it's, it's very useful. Very, very useful. Daniel said CSB should provide a rules.md file with his booster grams. I think that's a good idea. Thanks. A commissioner blogger, the AI arch wizard. We appreciate it. Yeah. Thank you, brother.

Uh, we've got some monthlies. I got Christopher Reamer, $10, James Sullivan, $10, Cohen Glotzbach, $5, uh, Oisteen Barrett, $5. Thank you. He's coming on right after this. He is. Yeah. They always mutton me to music. Timothy voice, $10, Gene Liverman. Hey, Gene, $5, Michael Hall, $5 and 50 cents. Thank you, Michael. New media productions. That's Todd and Rob, $30. Uh, Jeremy Gerds, $5, Derek J. Viscar, the best name in podcasting, $21, Paul Saltzman, $22 and 22 cents.

Nice. Always. Paul, uh, Damon Casajak, $15. Thank you, Damon. Uh, silicone florist, whoever you are, $10, Chris Cowan, $5 and Terry Keller, $5. Thank you guys. Yeah. Thank you all very much. And I can tell just by the numbers, I hear some people have upped their numbers. Thank you very much. Uh, that probably is because we were talking about keeping the podcast index running and it can't just be running on only two hosting companies. Cause that's pretty much what it is these days.

Uh, so all these smaller amounts, they are very much appreciated. So if you get any value out of podcastindex.org it's API, the work that is done, a lot of the work that Dave does, please consider supporting us. You can go to podcastindex.org at the bottom. There's a big red donate button support us there with your Fiat fund coupons. I actually, I think now, uh, because PayPal, PayPal, they're going to start expanding.

You'll be able to use other forms of, uh, digital money to, uh, to send, uh, to send stuff to us. And we just appreciate it. Boosts of course, are always appreciated through any of the modern podcast apps. Uh, and let me see. I got travel coming. I'll be here next week. We're here next week. We're here next week. Yes. I think we're here next week. Yes, that's right. Okay. Well, board, boardroom. Thank you very much for attending. Stand by for Oystein Berger with his mutton, meat and music.

Brother Dave, what you doing this weekend? Besides keeping warm for the big storm that we have coming, uh, you are going to get it worse than I am evidently, because it looks like the ice storm was going to be moved North of us. So it's North. Hmm. Yeah. It's like we were like Birmingham was supposed to get like, God, like two inches of ice. We were like freaking out. And then now they've said, well, you're probably not going to get any. Well, you know about the, uh, the exploding trees.

Uh, Oh, is that where they, where they, uh, swell up with water, with water and blow up extremely cold temperatures tonight. Please watch out for exploding trees. Folks. If you hear loud cracks, stay indoors. Yeah. Yeah. I think it's, uh, they just started some hype about exploding trees. I guess it can happen if it gets cold enough. Sure. Yeah. So no, they already predicted it would start today and it hasn't started today and maybe tomorrow. So I'm not too, uh, not too worried, but we'll see.

Austin's not, you're not in those crosshairs, are you? Yeah. But Fredericksburg is right on the cusp of it, but they were, they were all freaking us out too. So we'll see how it goes. Watch out for your trees. If we survive, we'll be back next week for the board meeting. Thank you all for being here for the one and only podcasting board meeting. We'll be back next week, right here on podcasting 2.0. You have been listening to podcasting 2.0. Visit podcastindex.org for more information.