Episode 254: Pop a TTermy!

Podcasting 2.0 for March 20th, 2026, episode 254. Papa Terme. Hello, everybody. Welcome to Podcasting 2.0. This is the only official place of something about podcasting. But it's us, and we're here. We've got an entire boardroom full with the misfits from Podcasting Dexter Soter. We are, in fact, the only boardroom that didn't win an Academy Award, or a Razzie, for that matter.

I'm Adam Curry, here in the heart of the Texas Hill Country, and in Alabama, the man who wants you to contribute your disc space for the good of mankind. Say hello to my friend on the other end, the one, the only, the pod sage, Mr. Dave Jones. Oh, man. While you were doing the intro, I realized I needed a knife to cut my cheese. And I was like, okay, I've got time. So I ran in the kitchen. I ran in the kitchen, grabbed the knife, got back to the podcast room. It was like, crap, I need a drink.

So I ran back to the kitchen again, sat down right in time to hit the post. Wow. You are amazing. No, I'm just hungry. Yeah, you had a little bit of traffic on the way home. Yes. Yes. See, is Eric PP in the boardroom? Is Eric PP in the boardroom? Is Eric PP in the boardroom? Paging Eric PP? I don't see him. You're safe to talk about AI. Yes. Yes. Thank you. Let me just make sure. All right. Yeah. Safe. Okay. Adam's AI vibe coding corner.

If Eric comes in, just somebody give like one blink twice and we'll know to shut up. So, you know, I installed Omarchi, right? Which is Arch-based Linux with a Hyperland tile manager. And I have been loving this operating system. Just loving it. I use all the workspaces. I've got all the keyboard commands down. Thing is rock solid. You know, I built a whole playout system for it. And what had happened a couple of times, you know, I was popping open a terminal window or something and.

Popping a termie. Show title right off the bat. Popping a termie. That sounds almost sexual. Is it termie with a Y? Is it termie with a T? It's got to be T-T-E-R -M-Y. Popping a termie. So, so I popped, I popped the termie and I saw this thing come up from, and I just like, you know, super key W, get rid of it. Cause I don't know what this is. It's all kinds of stuff. You know, there's separate, separate editors that open the terminal. And so I'm kind of learning my way around.

And, and so the other day, like three days ago, I popped the termie. I'm like, hmm, open code. What is this? Oh, oh, you, you have not used open code before. Holy moly guacamole. Where has this been all my vibe coding life? This, this is. I thought you had been using the, the Gemini CLI though. No, no, I was just copying and pasting into the browser. I've been telling you this for weeks. I know what you've been telling me, but you didn't tell me about open code.

So I'm like, well, this is interesting. And kind of, uh, coincidentally at the same time, uh, my guy who had sent me the, the Vulcan card and the, and the Raspberry Pi, he says, oh, here's a config. If you, you know, he got me the new models, the Quen 3.5 and the coding models. Now you can hook this up to open code. I'm like, oh, that's interesting. So I did that later, but I started off with open code with their free, free as in free beer, Zen models. Oh my Lord. This thing has changed my life.

I told you it is a, it is when you make the switch to CLI, it's a game changer, you know, cause now it's doing stuff. And I'm, I'm actually learning code now because now I'm seeing it's doing diffs and it's, you know, the explanation, you know, it's not hidden, like in a browser it's, it's going by, it's a little grayer color. So you can read along. I'm seeing, I'm seeing what it's looking at.

You know, you can, you can just say like, um, all right, this curry caster I've built, we need to get it on GitHub. Boom. Uh, read me. I need to document the features. Version it. I mean, screw open claw. Open claw is pathetic. This, this is the system. I would pay these guys a hundred dollars a month. I'd cancel everything.

This is, and, and what's beautiful about it is you can, because it has access to the computer you're running it on, you can say, you know, I don't like the way this computer works. Hey, open code, do this. And it does it. It's like, huh. It's fantastic. Well, what just take, if you want to pay somebody a hundred dollars a month, just go pay Claude a hundred dollars a month and you'll get that times 10. No, like now I'm talking about intent. I'm talking about intelligence.

No, no, no, no, no, no, no. Cause Claude is a, they're all token munchers. They're all token munchers. I've learned a lot about context windows and all this stuff is token munch. If you're doing what I'm doing, it's just token munching. And the a hundred bucks a month is unlimited though. No bull crap. It is. I've never, it's, it's effectively unlimited for the level of use that you, that a human being alone just prompting can hit it with. Well, I like these guys.

I like it that they're open source. That's why I like about it. That's, that's, I want to support these guys. And it just solidifies for me the entire thinking that this stuff is going decentralized. The whole idea of getting all your models from one big cloud provider. That's, that's going to end. That's what I love about this. These guys, you know, I don't know what they're running, how they're running it, but I'd rather pay them. It's open source, no sneaky stuff that's going on.

You know, that, that's, you're using the Zen, you're using the Zen model. You're not using local, right? I have used the local, but it's not as, it's not as hard. Well, no, the context window is small because I only have a, you know, whatever, 20, 24 gigs or something. So the, you know, by the time everything's loaded, the context window is just too small. But here's what I did just today. So I, you know, I'd forked off that Roadcaster, what is it called? The Pipewire routing.

So that I could, you know, get the Roadcaster, all three interfaces working on Linux. And the problem that, and it took me quite a while to just get that done. And the problem I was having is that it decided to name stuff, but it was all wrong. So I was always sitting here like, okay, so for the clean feed, I don't want to use USB one chat. I want to use USB one main and USB chat is actually something else.

And so I just told my little open code buddy, I said, Hey, Hey, take a look at these config files. Just go, go look at the stuff I'm using and, uh, and rename them. Okay. Uh, put it on the repo, let everybody know. I put the curry caster on the repo. He's got it on repo. I published. Oh, where is it? Do drop, please. Please drop a link. No, I'll just let me drop a link. I have my own repo. You see, I knew you wouldn't had a repo for a while. Where am I repo at? Hold on a second.

Adam. It's at 19 one 99 is going to see one nine nine nine. No, no. One nine nine is a mistake. I made a long time ago. Oh, that's obfuscation. I forgot. I forgot a nine security through obscurity. Yes. Roadcaster pro two pipewire. Yeah. And, but then underneath that curry caster curry caster. Yes. This is what we're running on right now. Yeah. Look at those features. Now, is this the, um, there it is. Curry caster. Yeah. What we're running on right now, baby.

So is this is where the problems you're talking about that you were able to fix with, uh, open code buddy or those, um, that was with pipewire. Yeah. But is that the reason that sometimes you, your, your rig would just fire off something at random? No, no, that, that is some, that's like static electricity or something that the midi is being midi was being triggered. It was one, one player, but, but it's just this. Cause you know, over the open, I got to play this for you. Um, okay.

You got to play this. Uh, this is, uh, okay. This is funny for a number of reasons. One it's Kramer on CNBC. So inverse Kramer in effect, inverse Kramer to now we know that, um, Nvidia purchased or aqua hired, uh, open claw. Right. Yeah. Yeah. And they released a bunch of tools last

this past week. Yeah. And maybe earlier Nemo claw and all this stuff. So listen to this conversation because I think we're at peak AI hype at this point where this bull crap is being spouted. Another game changer that we talking about open claw. I mean, open claw is something that sounds like a force multiplier bigger than we've ever had. It was open sourced just recently, a few weeks ago. No, no. It's been open source forever. So wrong.

It is now the largest, most popular, the most successful open source project in the history of humanity. No, no, no. How about Linux? I'm sorry. GNU Linux. It's got more stars than other open source things on GitHub, but that doesn't mean it's the most popular in planetary history. I tell you. Yeah. So bull crap, most popular, the most successful open source project in the history of humanity. Wow. All the way back to Jesus. When you told me in November, it would be big in January.

This could be the next thing. This is definitely the next chat. Why the hell aren't people talking? I'm saying, why aren't people talking? Well, there are phenomenons of open claw all over the world as we speak. You used to just say, I'm going to chat you. In one line of code, you could get yourself, I would, I would, you know, get a small computer. We have one that's called DGX Spark. You could also set it up in the cloud. In one line of code, you could create for yourself, your own agent.

And you could then after that, just in one line of code, you can, you can create a complete security nightmare for yourself. Uh, on the cloud. It's great. Ask the agent to do whatever you want to be a lot smarter than this. We're still quite smart. Well, can we ask the agent to build us a heart? Okay. All right. Can we ask the agent to build us a heart? Oh, well, let's see. Still quite smart. Well, can we ask the agent to build us a heart so that we don't have to human transplants?

You can build us a heart that is artificial so that we get away from the idea. This is an analyst on CNBC who is so bought into this hype that it, I mean, like this, I thought, I thought he was kidding. He's raised for real. He means this stuff, man. He means it so that we don't have to human transplants. He can build us a heart that is artificial so that we get away from the idea. I would have to do all kinds of mundane stuff.

First, like for example, because this is the guy trying to roll it back now, easy with your open heart surgery. I think we should have to do some mundane tasks first. Oh, but let's see what kind of mundane tasks going to do mundane stuff. First, like for example, you know, design me a kitchen. This guy's getting real uncomfortable. He's like, wait, let's pull it back from the heart surgery. And let's, how about, let's just do a kitchen first. Now listen to the kitchen build.

Okay. You know, here's a picture of, here's a picture of my Kermit kitchen. Here's a picture of a kitchen I would love to have. Now I would like you to go off and design that kitchen and select furniture to put in it. You'll go off and learn how to design a kitchen, use 3d tools, and actually read manuals. You can just see, watch it. It'll read manuals. The build a kitchen manual. Yes. It'll type in the terminal, man build kitchen, you know, reading manuals, trying new things.

And then it'll come back with a design and it'll reflect on it. It'll say, Hey, this design that I came up with, isn't as good as the one that you showed me a picture of. Let me try again. Imagine that with your heart. Hey, this heart that I made. Isn't so good. It's again. Okay. So let's deal with the company that's laying people off. Every carpenter could now be an architect. Oh, every carpenter can now be an architect. Yeah. Plumber will become an architect. Every plumber can become an art.

Everybody. You're an architect and you're an architect. You're an architect. We're going to elevate the capability of every company. Laying off people saying that, you know what? Nvidia is giving us the opportunity to do more with less. Why not do more with more because you're out of it. Why not? Why not do more with more? I tell you, you know what? Nvidia is giving us the opportunity to be able to do more with less. Why not do more with more because you're out of imagination.

You're out of imagination. For companies with imagination, you will do more with less. For companies that are, you know, when the leadership is just out of ideas, they have nothing else to do. They have no reason. You need to buy more creation tokens, creativity tokens or whatever. Like you're out of ideas. These people are not real. Give me a break. I'm sorry. Look, it's like the response from OpenCode for my Nemo claws like, oops, there's a bug in your heart.

Let's make a pull request and push it to the repo. Let's open a discussion about your heart not working. Would you like me to patch it? Give me a break. These people are stupid, man. This is so bad. So that said, to me, the the revolution that I have now, I'm finally it's just taking me a year and a half, Dave, you know, starting back with, oh, what was I? Mangus and all these other things I was trying. Now, I remember that. Yeah. Yeah. So do they exist anymore?

I think they got aqua hired or something. I don't know. I have to check all my subscriptions. I'm sure I'm paying 20 bucks to at least a hundred companies. Probably. But this in combination with a little bit of knowledge of your own machine, this this is the game changer, the productivity that is giving me to make my my tool, my main tool, which is my computer. You know, it's not just a dumb box and I have to live within the confines of Windows. I can have anything I want.

So I can do more with more. I can imagine. But what happens is so today I went through this this process of, OK, what is the stuff that's really bugging me that I just can't figure out or it's going to take me too long or I have to set aside too much time? And it was the renaming of the of the Rodecaster Pipewire interfaces. Another one that's been bugging me for a while.

We have this thing called Godcaster Live and Godcaster Live lets you spin up a twenty four seven stream from the most recent episodes from podcasts that you choose. And it's pretty cool. You can you can insert station IDs in between and all kinds of different stuff. And it just runs. So whenever you know, whenever it sees in the list that a new episode is dropped from one of those podcasts, it schedules it next. So it's always kind of fresh.

It's a great way to dip your toe into some of the podcasts that this Godcast station is is providing. And so we have about 42 of them running right now, and I'm running on a eight gigabyte Linode server. And, you know, I knew it wasn't efficient because it's doing it's doing a lot of things. It's spinning up FFMPEG streams, which is really the the biggest part, you know, it's doing to make these podcasts kind of flow. It's doing some in stream FFMPEG processing in stream.

It's talking to the API at very frequent interval interval in order. You're doing like crossfades and stuff, too, right? Yeah, but like zero point five seconds, just real quick, you know, from one to the next, certainly with the station, there's it's processing power and I'm just seeing it's like it's not very I can see it. You know, how do I scale this thing to, you know, 400 streams on one box? Spoiler alert, that'll never happen on this box.

But so I said, you know, because I just said, OK, go check this out. It SSH is right into the machine, you know, because it's just using my local machine. So SSH and you have it in plan mode. So it's not going to build anything. It goes in. It understands the system just by looking at log files. You can see all that everything that's doing. It's faster than I can do it for sure. It's in plan mode, but still tell me even in plan mode, man mode, like you still your butt draws up in a knot, right?

When SSH is into your server and you're like, oh, my God, what is it going to do now? No, I would be so nervous. I would be so nervous. I mean, before I did this, I made a snapshot of the server, obviously. OK, I'm not a total idiot. I've been around. So if anything happens, I can always roll back to a snapshot.

And so I felt pretty comfortable and maybe 30 minutes of work and, you know, interactive and, you know, it's asking me questions and I'm like, OK, yeah, you know, and says, well, here's the thing we can do. We can put eight, eight streams into one FFMPEG process instead of having eight separate ones. You can have slots. I mean, the stuff I didn't even know you could do and the the the reduction in server load is 80 percent. Oh, wow. Yes. Yes. This is this is phenomenal.

So that kind of stuff is now the reduction in server load was 80 percent from what you had. But the real question is. How does it scale up and what is this? What is the scaling load now? So as you go from what you are now, do you have. Like how how much more efficient is each stream, you know, it's effectively. OK, because the biggest overhead is FFMPEG, so when you're taking eight FFMPEG streams and bundling that into one FFMPEG stream, there's your, you know, eight fold reduction in essence.

So I could scale this box up probably four to five times, I'm thinking. And OK, so you could get maybe 300 streams. Yeah. Yeah. Somewhere. Yeah. OK. Fifty to three hundred. You know that that in 30 minutes, you know, I've had on my list for months optimize code. Yeah. Yeah. I've got a few of those. Where do you get some streamlined channel stream code? Yeah. Yeah. No, I'm there. So that to me is a 100 percent revolution.

And and I hope when people see the curry caster that they enjoy it, you know, that they use it, that they improve on it, that they can use it for their for their part. It's it's it's one big Python script, you know, not even that big. And it just works. Trademark. That's your signature programming style. One big one gigantic program. Yeah, that's that's my that's my vibe, man. Yeah. One day you're going to have your own OS and it's just going to be one.

It's going to be one two million line Python script. I did. I did say, hey, for the Godcaster. So I say, hey, which is also well, it's a couple of different scripts because you've got an API script and you've got, you know, but the main script runs per player. It runs it as a system service. And there's a lot of smart, I think, personally smart stuff in there. And I did say, hey, what would it take to refactor that whole thing? Split this up a little bit.

Yeah. And it says, you know, we could do this, do that. But it really came back. So it would take two to four hours, you know, with your interaction to figure it all out. But why? This works great. It's just one big script. What's your problem? And I love you, OpenCode. I love you. I love you, OpenCode buddy. My OpenCode buddy is great. And, you know, so it's really helped me build these your town live streaming stations. Oh, man. I was showing our partner Gordon and, you know, Gordon is great.

You know, it's like I said, hey, watch this. You program the whole thing in this chat interface. He's like, well, what about the prompts? Don't you have to tell people about the prompts that they need? I said, no, no, you just you just say, you know, show me a Sunday eight o'clock. And then and so he's trying to trick the system with stuff I've never even tried.

He's like, all right, show me every single day, Monday to Friday, 6 a.m. to 10 a.m. I just type it in exactly as he said it pops up, works. I mean, it's creating promos on the fly. It's remixing things and all built with vibe coding. And certainly this next iteration, this latest version with with OpenCode. Oh, man. I've got deployment scripts now. So, oh, you got a new station, which I'm firing up a new one. Oh, so now you don't have to do the whole cloning thing, the cloning thing.

It even does the DNS, the CNA, does the. I had a big readme file like clone server and then and then you go in that readme. OK, first you do this, then you change that. Remember to replace that that thing. Yeah. In sysadmin terms, you call that a runbook. Yeah. Runbook. Yes. Oh, that's another good runbook. Yeah. So I had a runbook, which consisted of all this technical debt that I'd created of hard coding domain names and stuff like that. You know, you know, you know what? Oh, yeah.

Yeah. And so I know to me, I'm just I could not be a happier guy. I'm I love OpenCode. I love that it's open source. I want to support those guys. They don't even have a donate button or anything. I don't know how they're doing it, but I think it's maybe there's some Nvidia shield thing on the back end. I don't know. Wouldn't surprise me. But do you have guardrails on it? I mean, like if you're going in there and said, you know, don't execute these kinds of commands on OpenCode.

Yes. Yes. Yeah. Yeah. And it can't it can't it doesn't have root. So it pops up. It uses PK, EXE, I think. And so it pops open a dialogue and then I can decide whether I want to type in the password. OK, so that that stuff is good. But anyway, back to I think that this is peak AI nonsense. What I'm hearing from Kramer and the CEO, you know, Jensen Wang of of Nvidia. Do you really pronounce that Wang? I'm pronouncing it Wang. OK. OK. Wang, whatever his name is.

If if that is the next chat GPT open claw. No. Go look at the skills. They're dumb. This is not there. Oh, I can turn my lights on. Oh, I get a weather report. Oh, it's great. You know, oh, I don't show me the person who designed the kitchen for show me that guy. I heard I heard Wong. OK. Jensen Wong. That I heard something depressing to me. I think last night I had a hard time sleeping last night. So I was up like three o'clock funny. I was I was up at three forty five. We could have texted.

We could have sugar. I didn't even think about it. Yeah, I realize. But I was trying. I was just getting some coffee and stuff and watching a YouTube video of a channel that I follow. And they were they were interviewing this guy who's deep into the he's been in the he's one of these guys that was in the the transformer model era of A.I. back pre before it hit the scene. So, you know, he was deep in there before. So he's doing just the research part. And they were talking about how.

Well, his opinion was, we have a long way to go before we reach peak nonsense. And because what he was saying was. That people misunderstand on in scenarios like this, where there's some new radical new technology, people always misunderstand how where they are on the early adopter curve. And so.

People who are techie, which is, you know, us and everybody listens to this show and all the people that that are I mean, if you're if you're listening to the boardroom right now, you're probably pretty techie. You are way to the left on the early adopter curve, farther to the left than you realize you are. And so the the the amount of time it's going to take for peak A.I. nonsense to reach the average person is it's going to probably take a few years.

I think we have a lot more pain to go of nonsense before we actually get to the peak. I'm trying to make a comparison between this open claw hype and I would say in a way it was almost like. Somewhere between mosaic and Netscape, you know, mosaic was maybe it is maybe it's our mosaic moment like, oh, OK, I see. I get red ball dot gif. OK, yeah, I can use that. I see how that works. I got blink tags.

I think we're in the blink tag era with with open claw because there's enough people rushing out, getting cloud servers, influencers charging 100 bucks a month for a preconfigured open claw server. And, you know, it is the number one token muncher. If you if you look at all that, I forget whether it is there's a maybe it's open router, but they show you, you know, who's munching the most tokens. It's all open claw. Everything else is kind of, you know, leveled off a little bit.

And, you know, at a certain point, you know, like 80, 90, 100 dollars, like how much fun is this thing? Really? What did it really do for me? Well, speaking of popping a termie, I'm probably like I'm I may I may be on the curry caster soon because I cannot handle windows anymore. I'm I'm so done with windows. Yeah, I'm I'm about to just nuke every box I have and start over with Nixos. Oh, you got to go hardcore, hardcore. I dove into the Nix deep end and I'm actually digging it.

I actually like it a lot because and I've got I've got hyperland on it and I've got I've said I've set it up a lot like the way in a marching box would be. But what's the difference between Nix and Arch? Well, Nix is it's a whole different way of building the system up from the ground up. So there's there is a there's a package repository, but you don't you don't install if you're doing it the native Nix way. You don't install packages with a package manager.

Every everything, like literally everything in the system is controlled with this one configuration file. And so you go in there and you you type into the configuration file the a package that you want to install and any install options you put, you know, hard static IP addresses, all the all this kind of stuff.

You put it all into one single configuration file and then you run a rebuild script and it just looks at what you what your options are that you specified in the config file, downloads all the appropriate software, makes all the configuration changes in the system, compiles everything from source. And the big and the big thing is when it compiles it from source, it compiles it as monolithic executables with the libraries included. So you don't have like library conflicts.

OK, I see, you know, where this package needs this version of glibc and this other one needs this other version and you can't get it to run. Everything's compiled monolithic. And so the thing is, like when you get a working system that you're happy with, you take that you can take that config file, stick it on a on a fresh box, run the rebuild, and you've essentially just replicated your system from scratch. Everything just goes right back to normal. It's it's pretty right now.

It's it's it's fiddly. You know, it's fiddly. I mean, you it's you can't just do apt to get, you know, or whatever, install the this. You have to go in the configuration file. You got to find the right configuration syntax to use for these packages. But it's, you know, with the way things are googling things, when it comes up with the little AI helper thing on the Google results, nine times out of 10, it just nails it the first time.

And so the idea there is that you can take your config file and throw it on a new box and build your box exactly the way you want it. Yes, that's one of the advantages is you can easily replicate a box with with no effort, basically. And then the other thing is you have snapshots. So anytime you make a configuration change, you can go right back. It snapshots the previous one. And so if something messes up, you can just roll it right back to the last good configure. It's declared.

It's declarative configuration is what it is. I have I have one more. But you know that Lenovo ThinkPad I got. Yeah. You advise me on. So that's my target. That's that's the mobile rig. And that's now going on Marchi. That's going to end the curry caster. I was using this commercial package to called M M Air List. Yeah, that old Windows program. Yeah. Which had a lot of limitations. It was one guy. He's doing it. You know, it's not open source. He sells it license, which is fine.

And I paid for two licenses, maybe three. I'm not sure. But, you know, I I had different ideas about what I wanted it to do. And I've just it just built it. I just this is the you know, it's not I'm not going to sell this, you know, go ahead, fix it, go download it, use it. Give me some ideas, put stuff back into the repo. I'm over that. You know, I don't I don't see the business opportunity in vibe coding at all.

I do see I do see the improvement to the quality of life improvements and the stuff that I can do that were in my head. So there's your imagination thing that I just wanted to make my life better. And, you know, you can do amazing things. But I think the I think the commercial software industry as a whole, whether it's Windows or line of business applications, and I can think that it's going to be a struggle going forward. Things have fundamentally changed.

And, you know, I can see it just in the industry I work in. But you see it everywhere that you have old companies that are just they've become bloated and huge. They don't know they don't know how to compete anymore because they have in their line of business, they have an 80 percent market share and they have for 25 years. And it was just like that whole. I mean, and I feel the same way with that's like with switching from Windows to like Archie sourced. He got me on he got me on Nick.

So as he was, he's been talking it up and that kind of thing got me interested on it. And then Alex was saying, you know, yeah, he's like, I don't use Nick's, but I use Flux, which is basically like Nick's applied to another distro. And so I got a little braver about going into it and using it. And I just feel like. That the commercial software, you have no options that your software is going to look like this, and that's all you get that.

It like rapidly now that feels antiquated, it just felt like the whole idea once you've vibe coded for a while. Yeah. When you the idea that you can't that there's this piece thing on your screen and you can't change it is it feels like shackles. You know, you're like, wait, wait, this feels like the past, you know? And so like that's where I'm at with with Windows. I've been having this problem for months now where when I'll come in suddenly as a thing, certain things will just stop working.

I'll notice that my box upstairs where my home office is, I noticed that it just there's things. I just know something's wrong. So and then things will start running really, really slow. And then Chrome will start crashing and then Brave starts getting wiggy and telling me it's out of memory. Then I go over to the task manager and I look and it's like, oh, you have 32 gigs of memory. And but but there's only but there's only 14 gigs used. I'm like, OK, well, we'll have 18 left. Why are you?

And then Brave says can't load out of memory. But I'm like, I've got 18 gigs free in Windows. It's Windows. And then and then Windows will force installs Windows updates no matter what you do. Yep. Yep. No matter how hard you try not to. And it's going to reboot you at the worst possible time. And I'm just so tired of it. I'm just done. Usually right before show. That's when that's when. Oh, yeah. Or when you're out of the house and you need to remote back in to do something.

It's always that way. Yeah. I know. I know. It's horrible. I had a a podcasting conversation with a friend of mine, Alex Sanfilippo. Do you know who he is? Have you ever heard of him? I've heard of him. Yes. Yeah. So he has a company called. I don't think I've ever talked to him. Yeah. I met him at the Spark Media Conference a couple of years back when I did the keynote in Houston, I think. And he has a company called Podmatch. And I think it's it's been he's had it for quite a while.

It was successful. He and his wife run it. And he connects guests to podcasters. And that's, you know, that's a pretty smart business. And he and he I think he turns over. Like 77 percent goes to the podcaster, so he keeps a small dig for himself. And he says, Adam, I have a I have a problem. I see a problem in podcasting. And I was thinking about doing a town hall. And I guess, you know, all these guys that are on LinkedIn. So I put something on LinkedIn.

He says, I got 500 people RSVP and say, yes, we would love to be a town hall to talk about this, this problem, including hosting companies. He says, but I want to ask you for some advice. OK, so we talked the other day. He says the biggest problem. And, you know, this is not the kind of podcasting I do because a lot of podcasts have guests, you know, tons of podcasts. Everyone's got everyone wants to be Joe Rogan.

And every single agent, manager, promoter, PR person in the entire world wants to get their clients onto podcasts. And it's a pretty cost efficient way because you just scrape feeds if you still can for the email addresses or you buy one of the many lists of the feeds that have been scraped throughout the ages. Or you I'm sure you can just buy a list of people who have podcast podcaster in their profile on LinkedIn. I'm sure you can buy that.

And so he says this the the spam or the the maybe even call it spam, but the requests, it's overwhelming and they're all full of crap. You know, I've seen I've read them on the show like, oh, I love the last episode of No Agenda where you and John were talking about this. I think this person would be great to have as a guest. So now they're using AI, look at transcripts and try and personalize the pitch. He says, he says, can we fix this problem somehow at an industry level?

And as we're talking through it, I said, well, you know, obviously we the industry kind of I feel jumped the gun a bit where Buzzsprout and I think RSS.com and others were just taking that email address out. But that was for other reasons. That was because you had competing hosting companies trying to steal customers, which was pretty lame. Who was that? Was it Libsyn that was doing that? Acast. Acast. I'm sorry. Sorry, Libsyn. Just my default Libsyn bad. I'm sorry. They don't listen anyway.

Sure. They got no one left to listen. And and I said, you know, and that that actually created some problems because we weren't quite ready for, you know, verification for different services. And as I'm talking through it with him, I said, well, what if there was like a booking tag?

Like so and this doesn't have to surface in the podcast apps, but it could be something in the feed where if you want to inquire about a booking, then this is the URL, URI, whatever you want it to be that you contact. And because his whole problem is I don't want to be the solution. He said, that feels icky. He said, no, this isn't a complete open solution.

We have a booking tag and, you know, you can your your affiliated podcasters can put their link to their pod match page, but it could be to a Calendly or, you know, a link to my agent or, you know, my manager of the podcast. And so I wanted to throw that out there. I actually I invited him to be on the podcast index that social. I don't think he's ever seen a GitHub. He may not know how to he may. I don't know. He may not know.

You know, I said there's a there's a process and you write up a proposal and and then, you know, if someone implements it, then somehow it just kind of gets adopted because people are using it. And I thought that this might be something hosting companies could use. Hmm. I like this idea a lot. Yeah, I thought it was kind of decent.

So maybe someone out there can look for him and and help him or if someone feels like writing up this tag, I thought it would be good because it's not again, it's not something that has to surface, but it but it's something that goes. Go ahead, scrape those feeds, get that out of there, get the booking tag. That's the one you want. I do. Wait, I think I have met Alex. I think I met him at podcast movement. I recognize his possible. Yeah, possible. Oh, he has a very recognizable face.

Yeah. Yeah. I think I did meet him. Yeah. And I know I did. In fact, yeah, we talked. Yeah, I talked to him. He he followed me on podcast and social this morning. So I'll message him and tell him, you know, to that we're going to like try to come up with some kind of proposal or something. Cool, because I think that's a great idea because it's a it's a signal to the world that, you know, hey, I'm open to this kind of contact. Yeah. And send it here. You know, serious inquiries only.

Yeah. It could be a link straight to a booking page form. Yeah. I mean, yeah. Calendly. A lot of people love that. You know, just just go there and stop trying to trick us with stupid emails. Yeah. And that that could be the you know, a way that you sort of. Differentiate between people who are just people that is truly spam. Yeah. And people who scraped the feed, which that is going to be a spammy to obviously, but it's not the volume is not going to be as high as just generally.

And at least well, at least it goes where you want it to go. So it won't necessarily be an email address, but it you know, so I haven't I'm not a Podmatch member. I don't know how it works, but I understand the concept is like there's a page and then Podmatch connects you with the right podcasters. And so it would go to the Podmatch page in his case, and then Podmatch decides if it's a match, but it can go to anything, anything you want. So anyway, that that was my contribution for the week.

And I was also thinking like, that's not a bad idea. We should probably come up with a booking tag. Yeah, and I like that. Let me actually make a note of that real quick. That's a yeah, that's a good idea. I like that a lot. Can we let's talk about transcripts for a second. Yeah, one sec. What is what is Nathan G linking to here? More control over contact information. Podcast. Nathan is the link wizard. He is fine.

He just digs into the namespace and finds every time we've ever like duplicated some kind of effort about something. And it's shocking how many times we reinvent the same thing that we already invented one time before. I love I love this. I'm looking at proposal podcast contact, more control over contact information. And the proposed tag could include something like contact URL, physical mailing address. And then the first comment, James Cridlin. What problem is this trying to solve, please?

The Cridler, the Cridler at work once again. The Cridler comes in. All right. The Cridler. So there's there's the problem it's trying to solve. And oh, my goodness, that was in twenty twenty one. So it only took us five years to figure out the problem it was trying to solve. So maybe what we did come to us, but we did find it. Yeah. So maybe that's the way maybe it maybe we should be called booking instead of maybe contact is better than I think about it. Yeah. Contacts a little.

It's a little too generic, but generally booking is clear. You know, booking boom URL, email address, phone number, P.O. box, ham, ham, ham, call sign, whatever you can come up with. All right. Good. Casey for an kilo five alpha. Charlie, Charlie. Seventy three is good, bud. Oh, by the way, CSB said to ask you to finally give him an interview about his new A.I. lover, OpenCodeZen. Yes, I will. I will. CSB.

I committed to it, but it's just always difficult for me to get my crap together and timing. But I'm working on it. I'm working on so overcast, added transcripts. Yes. In the beta. Yes. And so he's running them himself, right? Yeah. So the latest episode of Accidental Tech, I haven't really been listening to that very much over the last few months, but I just dip in and out whenever there's something that strikes me, that strikes me and I feel like I need to listen to.

But so I listened to the episode where he talks about this, the most recent one. And. The way he's doing it is with. A rack in a data center full of 48 Mac minis. Wow. All running OpenCode, I mean, no club, club, club, club. Yeah. Nemo claw, whatever. So these are all these are it's 48 Mac minis in a rack enclosure in a data center. All accepting jobs from a transcription jobs from a queue and using the built in.

Native models within Mac OS Tahoe to do the transcription seems like an expensive way to do it. That's twenty four thousand dollars worth of Mac minis. Yeah, that's a lot of money. He's a Mac guy, though. He loves the Mac stuff. Yeah, I think I mean, the the the reasoning that it's worth listening to. It's about a hour long discussion about it. The reasoning.

Makes sense on a lot of levels from the idea of the code being if it's Mac, if it's a Mac native on transcription, the APIs are the same on the Mac as they are on the iPhone. So you can you can just write that code once and then you can transcribe on the phone and on the Mac both. So it's and it's the same code, essentially. How many how many transcripts do you think he can do? I mean, I was running. Of course, all of them. Oh, my goodness. All literally all of them.

I'm running I'm running my my watcher, my pod ping slurper. What's it called? Oh, the gossip blizzard. Yeah. So I have code a little bit. So there are about average is about twenty five hundred new episodes an hour. So that's that's the level that he's going to have to be transcribing. Or evidently it's work. Evidently, he's doing it. That's why he's got forty eight of them. Yeah. I mean, running whisper locally is you can do quite a bit with that. It's low overhead.

And, you know, you could burst them. There's all kinds of stuff you could do different from running. Listen to me. Who am I? That I just I don't this. There's a lot of stuff going on here that I think bears discussion. So. One thing is how many podcast episodes. I already have transcripts that could be delivered. Well, wait a minute, he's not picking up transcripts that already in the feed. What what my understanding is, is that that is already working, it's just not in the code yet.

So he's going to do that. He has to. You're crazy if you don't. Well, to me, you have to, because the creator is telling you, you know, you have to honor that. Yes, definitely. But, you know, he's he's he's of the Apple school. The Apple school is I want the best experience for my customer, even if it means I have to create it for them. Right. Yeah, I think you're right about that. Yeah, that's definitely the mindset.

And. How so as I was like, well, how many episodes, how many podcast episodes have transcripts? And what kinds? I started digging into the transcripts table in the index. Yes. And there's about total total records of transcripts in that table is about seven and a half million. Well, maybe a little bit, actually, maybe a little bit more than that. I don't remember the exact number, but there's a lot of overlap. So there's you've got potentially text.

HTML, Jason. SRT and VTT, so you get potentially five different transcript formats, which is a whole nother discussion that is being had right now in the namespaces, do we deprecate some of these? Deprecating, deprecating in the sense that you can't make people not do it. People can stick whatever they want to in any tag. But just from a recommendation standpoint.

Do we want to encourage there to be, you know, a couple of formats instead of five, potentially, because that because what what's happening now. Is if there's so bus bus Sprout does this, I'm not sure if anybody else does it, but I know bus Sprout does it when bus Sprout publishes an episode and does transcripts, they publish it in three formats every single time. So they put three different formats of transcript in every. That seems like a waste text. I think it's text, Jason and SRT.

I think those are the three. Maybe I'm wrong. Maybe VTT. But regardless, there's three of them in there in every episode. So I think it makes a lot of sense to drop the recommendations down to whatever. To to whatever we think is the most widely supported as the recommended one and drop the other ones out of the documentation. And then just come what may, you know. Yeah, Nathan, you're right. VTT can do it all.

I mean, I've been using VTT, but I heard the Cridler talking about that and he was like, no, no, we can't do you should just use this. I thought, no, he's a big proponent of VTT. No, no, he was against VTT. So I think like I think that makes a lot of sense to just. Where was it that he was saying that was that on the on power? Maybe last week. I think it was last week on power. Yeah, I think so. I don't know. But the discussion's going on on the namespace right now.

It'll be we'll we'll be able to figure it out. The other the other thing, like my thought about spending twenty five thousand dollars on transcription computers. I mean, that's just from a from an app. I thought, you know, we obviously talk to a lot of app developers on a regular basis, podcast app developers. And I don't know a single one that could even afford to spend a thousand dollars on a on a server, much less twenty four thousand dollars. So overcast as far as revenue goes.

If I mean, he's obviously independently wealthy as well. But it's from a revenue, just from an app revenue standpoint, overcast is a serious outlier in this whole. Podcast app ecosystem. And that that's fantastic for him, but that's that's a lot of money that most app developers wouldn't couldn't only even dream of. So I don't think that the way he did it is going to be something that almost anybody else can replicate. Yeah. And but is that a bad thing? No, it's not a bad thing.

I'm just saying that that is like. What I'm saying is. If we put if we push this back up to the hosts, to the source. Yeah. To me, that's when it levels the playing field. So eventually. Mobile devices will have enough horsepower to do these things on the fly. And there won't have to be all this background work that's done with with servers to do all this kind of transcription and that kind of thing.

Eventually, the horsepower for the neural engines in these mobile devices will just it'll just boom. I mean, you just got a transcript, you know, take probably it'll take 10 seconds as the as the audio is loading. By the time it finishes downloading and and and queues up and you go to tap it 20 seconds later, you've already got a transcript. Eventually, that's where we're going to get in. This thing will be a level playing field. But until that happens, I feel like.

I guess the only reason I brought up the money aspect of it was just because I had. I just liked I like. The app developers. I see the podcast developers that that we work with on a regular basis and. They spend so many hours working on their apps. And I just feel bad that they don't get the revenue for it. This other apps like Overcast and Pocketcast and stuff do. I just. I don't know. It just makes me feel bad about that.

And I just if if there was a way to put, you know, if I just wish more hosts would go ahead and stick the transcript in. The RSS feed. And it just I'm stumbling around, but I think what I'm trying to say is that the more we can push these features upstream into the feed, the less of this sort of thing has to happen. Do you see what I'm saying, like the less you're going to fall on an independent on a single app developer to have to go out and create a humongous infrastructure to do one feature.

Right. Because the because the data that they need will just be already there. And so. I think the the other thing I took away from their discussion on their podcast was that I also think they've. Some of them have missed or have underestimated how much. How much limited context there is in these LLMs was one of the suggestions was, well, the local models on the Apple devices are not that great. I mean, they're good. I mean, they're as good as Whisper or anything else, but they're not perfect.

I mean, they're going to make mistakes and get wording wrong. And so what you could do is you could take that. You could take that like do a secondary, a second stage processing, get the tech, get the speech to text and then run it through an LLM. To clean it up, clean it up, yeah. So I think I talked to the last either last week or the week before I've done, I recently did this exact thing. For a project at my day job.

And so this project involved taking a bunch of documents, extracting all the text out of them with OCR because OCR is very fast, just traditional OCR. And then and then it's going to be a mess. It's going to be messy. And then you run it through an LLM and you prompt the LLM and tell it this text that came off that I'm providing you came out of an OCR, is the result of OCR. There's going to be problems. Please fix them. And then you give it examples of all the problems.

And so you would say, you know, some if you see a word that's got a zero where an O should be or a date that has 2026, the number one, 03, the number one and so and so that's that's a date where it felt the slashes were the number one. So you're like describing all the different examples of what can go wrong in an OCR process and you tell the LLM to clean it up. And it does a fantastic job. I mean, it just it's spits out crystal clean second stage processing. The problem is it's very slow.

It's very slow. Yeah. And also the context window is way smaller than you think it is. Yeah. Even on big models. So like a typical podcast transcript is is really, really big. You know, even just the plain text and it will not fit in most context windows. And so that I don't think that that second stage processing is practical at all, which again, just push puts you back into the this needs to be these transcripts need to be as close to being made close to the source as possible.

Because the the creator knows best what is needed for their content. If you're on a like you're doing a medical blog, I mean, a medical podcast, then you already know. That. You're going to get transcripts, transcription engines are really going to struggle with your content.

Yeah. So you may go and you may pay somebody to do hand fixing of your content, of your transcript, or you may do the second stage processing through LLMs or something like that, because, you know, I or an engineering podcast or something like that, you know, a lot of the words are going to be difficult and it's going to screw them up. Well, I think ultimately, so a couple of things, one, yes, one, one, maybe two, but maybe just one standard. I use VTT. I like VTT.

I like I like the features it has. That's all great. I think Apple uses VTT. So perfect. It's the hosting companies, you know, and they have to provide this free or at a fee or built in or whatever. It's the wrong way around. It's just the wrong way around to have apps being responsible for this or for anything else. And actually, I feel it opens up a door that I don't like. Because, yeah, today it's your transcript and tomorrow it's your image. And the day after that is your chapters.

Oh, we'll see. So this is this is the next part I wanted to discuss is one of the one of the things he mentioned on that show was automatic chaptering. Based on the transcript, and so that's evidently one of the features that he's targeting in my immediate thought was this is the same same thing I wanted a few weeks ago. You know, I have I have the security now podcast that doesn't publish chapters and I want chapters.

And so we had this whole discussion about how that's some that's somehow like not cool to put chapters into like into a podcast that didn't want them or didn't didn't provide them. That wasn't it wasn't that it wasn't that it's not cool. I think providing that for a fee can be a problem. Well, isn't that what this is, though? Yeah. And I pay for you. It's a paper. Yes. And I think he's going to run into trouble with that. OK, because if I publish my chapters and he's doing different chapters.

I have a problem with that. Well, I don't think that's I don't think that would be the case. I think it would be for podcasts that don't have chapters at all. Well, that's another point, another another point. If you're charging a customer to do that, you could be taking away from the podcaster's intent not to have chapters for commercial reasons, whatever that is. So I I'm not going to sue him. But Leo Laporte might. That's what I was thinking.

Yeah. I think it's I think it's and I also feel it's it's rude. Now, if you want to create your own, it's the money part. It's charging money to do something to someone else's content without their consent. It should maybe. OK, here we go. Here's another tag. The consent tag. Yes. Rape my content however you want to. Go ahead. You can make a transcript. And we probably should have thought of this earlier before Apple did it.

It's like it shouldn't be that I have to go to Apple Podcast Connect to tell it that I want my chapters being surfaced. In fact, I think that's a class action lawsuit waiting to happen. It's bullshit. That's not that's not their content to do. I mean, so why don't I just make a transcript of Tim Cook and sell that the end? The end is never in sight if we start allowing that stuff. But if you say, OK, it's OK if you want to make chapters, it's almost like a like a remix license.

And like like over overcast. The free version or the you know, the non-premium that you just download, you can download it for free, but it does have ad banners in it. So there is ad supported. So your use of the free version is still still making money, is still making money. And so my my in light of the comments, I can argue this used to be argued quite successfully in the music business.

I can argue that making a transcript of my podcast and making money off of that is a unfair trade practice. And unless I now in this case, I I'm pretty sure I put Creative Commons do what you want attribution. No, but there's always no commercial use. So that's a that's taking my stuff, making commercial use of it. Actually, not cool. But, you know, you're you're not the person that would ever do anything about that. But, you know, but the Cridler might. But but this is this is the issue.

I mean, it's just in light of this discussion that we had just a few weeks ago. I'm just because my first thought is, well, I want this is the thing I want as a listener. The thing I want is for security now. That podcast that I listen to because it's work related. I want that podcast to have to have chapters and and it does not. And if somehow and if a if a podcast app. Does on the fly transcription and adds chapters to it.

I will love it, but I guarantee you that Twit will not love it because they are they immediately see that people are going to skip their ads. Well, and they're going to get mad about it. Well, we're right back to what happened a while back. It's like, OK, then why don't I just make a podcast that strips out all the ads? If I can, if I can do whatever I want with your content, I should be able to do that, too. I mean, it's just because that's what you want. You don't want it with ads. Right.

It's the same thing. Yeah, you're yeah, it's it's it's all those types of apps, Nathan, it's, you know, sniped or whatever you call that thing. Yeah, it's it's all that is was it your say the one that I tried that was near wasn't very good. It's all those kinds of apps. And I just feel like. The only way to do it, the only sorry to interrupt, the only way to do it is if you build an open source app that anybody can compile or do whatever, then that's fine. I don't have a problem with that.

If you're selling your app, if you're making money off of your app, that's where you step over the line. Mm hmm. Yeah, I feel like this is another version of. How the software of how the the industry is coming up against the reality. Of infinitely customizable software. Every industry is going to face this, and that's the world that we live in now. And I don't see I don't see how I pretend bad things on the horizon for your advertising, portending. Oh, nice. I'm a portender.

I foresee bad things on the horizon for advertising supported media in general. And this that's people are going more and more towards premium subscriptions and that kind of thing. And I just this I've got I brought a clip. I don't know if you got it. Yeah, of course. But I sent you a clip. And this is because I was listening to power power this morning. Power. And this happened. We will then deliver that in 10 second packets, similar to the way that HLS does it.

That will work on all podcast apps that support video playback. And let's just take a quick break. Today's conversation. Very nice. And so that's MP4 videos. Yeah, I heard that too, actually. And normally there's there's like a thing that James puts in there talking about like a house ad promoting his trailer feed or something. But this is one of the first times I heard an actual ad in there. Yeah, it's pretty rare.

And then but the the thing about it was it was such a poorly done ad that all they did was took the intro from their normal podcast and stuck it in as if it was an ad. I'm walking. This morning and I and all of a sudden this thing starts playing and I'm like, what? Wait, what? I thought I had bumped my phone and switched to a different podcast because it's like in here. Here's today's episode and here's your host, blah, blah, blah. I was like, what?

And so I take out my phone and there's Castamatic is on the home screen and it's showing the cover art for this other podcast. You know, you could have been hit by a car. This is very dangerous. This ad actually could have killed you. It is dangerous. It's a danger to humanity. And so I was genuinely confused of how this podcast I'd never heard of got on my phone. And then and I'm flipping around through there, I'm like, oh, wait, this is just chapter art for the ad.

I was like, oh, this this truly is. I back it up and I'm like, man, this is like a dynamic ad insertion. I was like, this is this is why. Dynamic ad insertion. Is one of the reasons that people want this sort of thing. It's not that people don't want to support or they want to get something for free. I mean, there's always going to be some people like that. Oh, 98 percent of the world.

Sure. But they but when when you do dynamic ad insertion, you are what you're what you're in effect doing is you're allowing a person or an entity that you do not know. To put content in the middle of your show. That you haven't that you don't have any access to. You don't they can put whatever they want in there. And we've seen this is a it can it can be it can just go terribly wrong so often. So, of course, there's going to be this tension that happens.

And this is I mean, I don't I don't necessarily have a broader point here, but I'm saying that this was just an example of one of of why the industry is hitting is bumping up against things like this. The same thing that happened to the web with, you know, with the just overloading of banner ads and punch the monkey and all this kind of stuff is it just eventually people do they like there's a local news site, AL.com. It's unreadable. The page is literally without an ad blocker.

The page is literally unreadable. Right. There's a banner ad that takes up that takes up two thirds of the top of the screen. You have to scroll to even see the headline. You just want stuff for free. It's like it's like, man, it is. And I get it. I get it. But but I'm telling you, I mean, like this is going to be this is going to be a problem in some there's going to have to be a sort of breaking point.

Because if the if the if the industry doesn't get a handle on the on the advertising issue, they're overloading and the dynamic advertising inserts suck. They don't get a handle on that. They these there's going to continue to be a proliferation of of apps that are going to solve that problem. And it's not going to be what they want. So anyway, this is that whole discussion. We'll see where it goes. And I'm glad I love that he did this. It's fantastic.

And I hope he follows through and does the podcasting transcripts honors that appropriately. I just see this this whole issue as being a typical push pull where people are going to gripe about about the automatic chaptering and all this kind of stuff. And it's just it's you can't stop it. It's just the way it's going to go the way it goes. You said something important that, you know, software is changing.

And just while we were having this discussion, I did a little experiment and I went to my buddy. You know, my buddy, you know, my buddy, my not open claw, open code, open code, buddy. Yeah. I said, create a command line podcast app that strips out any pre-roll ads or interstitials that start with let's just take a break and outputs an ad free version. Use the pod news weekly review episode from this week as your test podcast.

So it's already building it and it already went it already said I found the RSS feed and said, oh, looking at the chapters, I see there's an ad segment start 10 18 title ad financial matters with Richard Oring. And so it is now building this app for me. Yeah, this and that's what's going to happen, but this but this I agree with this is a great future where you I think you have every right to customize your own experience, every right. And this is the true future.

It's not this is not for you know, this is not mainstream stuff we're doing here. But this will be the future. You will create your own computing environment and your own entertainment environment and will not be relied. In fact, I would say phones, all kinds of things are going to come out of this, but not at a scale where people go to one company to buy the one thing or the one software. This is this is the revolution here.

Well, and I think the key there is that you can say, well, it's not going to be as good. So, you know, people are not going to do this in mass because the quality of the of the app is not going to be as good as what you would get from, you know, from a professionally developed app. And I can I think that misses the point. Well, because you if you if you build the app yourself using vibe coding or whatever the technique, if you build the app yourself. Let's all be honest, OK?

We tolerate a lot of shitty software because because we made it. I tolerate some truly terrible user interfaces because I built the app and I know how it works and it serves it. It serves my need like it doesn't have. I think we get married to the idea that software has to be beautiful and slick. That's not the point of software. The point of software is to solve a problem. Yes. Yes. And you will you will tolerate poor polish, a lack of polish on a piece of software if it's solving a problem.

And you made it because you feel an investment in it. This is this is in a way this is what OpenClaw is. So we know one thing. A lot of people installed OpenClaw, AI agent, my own thing. But they really didn't know what they wanted to solve. You know, so there's ten thousand skills and saying whatever, you know, my telegram bot is bothering me all day. I don't need a buddy bothering me. I telegram bot about my kitchen.

No. What I want is if if when we get to the point where I can say, hey, make my computer do this. That's that has always been. That I think that was in a way Steve Jobs dream. You know, he's always been striving towards that type of interface, make my computer do this and not have to. Now, of course, he approached it in many different weird ways, like, you know, stuck to certain menu bars and context. And now I can just say I just did it. Make my computer. It's building it now.

It is building the app I just we just discussed. Then I just ask for a command line to make it easy. But it could it could it could create an app. It could create a web app, can do all kinds of things. When people figure out this power, this crap, this crap computer, which I'm very constrained. And the provider, Microsoft, in this case, is always changing the ribbon bar and everything's different. All these things suck.

That, I think, is going to be a new revolution in computing, but it's I don't think one company will own it. I think it'll just be the world will have it. The people will try. They'll try. You know, this is what NVIDIA is trying to. Oh, you get Nemo. We'll own it. We'll we'll walk you through it. Bullshit. Only only one guy ever was able to do that was Steve Jobs. And well, and that stuff sucked, too. Well, you're I think I think Sir Ben Rose has.

A lot of insight here, he said, I tolerate crappy software that I made because it is sufficient to meet my needs. Why would I keep developing after I reach that point? It already satisfies the target audience. Amen. Exactly. Like I don't. That's the thing is most apps have most if an app has let's say an app has 47 features. Yeah, which you need to of which yeah, which each person is going to need a different subset of those.

Maybe you use 11 of them and I use 13 of them and five of my features that I want overlap with the features you want. And there's going to be a there's going to be this Venn diagram that overlaps for everyone, but they're never going to be exactly the same. Well, we're seeing that ourselves with Godcaster. We're seeing we're seeing we're seeing customers who they just like, oh, I wanted to do this. Well, that's not what we're offering. But I wanted to do this. Right.

It's what we talked about with Brendan last week on my page. Yeah, exactly. It's the exact same thing. It's difficult. But, you know, once there's some kind of law, there's some kind of law like that that says it's something about it. There's a sub. If you build for each time you build a feature, somebody's you're going to have some part of your customer base that is going to become dependent on it. Oh, I wish I could remember the name of that. It's like never.

The takeaway is never build a feature that you remove that you think you'll ever be able to remove it because you won't be. Yeah. Yeah. You're locked in for the rest of time. But yeah, that's true. Virtuality is really what you're looking for when when you have a piece of software that you forget you're using the software. And you know, it can it can be an idea, you know, it can be it can be a text box, you know, a simple text editor when you're really using it for what you want to use it for.

And in my end, so I built what I wanted when I'm doing the podcast. I mean, my my hands are flying. I can do it blind. I know everything works. It's exactly where I want it to be. The switches, the PFL, the Q points. So now I'm not thinking about how do I do that again? Where do I click? No, it's just there exactly where I want it. Daniel, the boardroom said I got my first cancellation feedback this week that said, quote, I got I coded my own solution.

First of all, I'm sorry that happened, Daniel. That I don't I don't like that on your behalf. But everybody's at risk of this. Yes, this is going to be a this is going to be the new future that we face. But it doesn't mean that it doesn't mean everybody's going to do that, because the thing about vibe coding and all this kind of stuff, you have excuse me, could you could you not say it like vibe coding? Could you say, you know, I take offense to that. No, it's just give me give me something.

Give me give me the proper terminology, open coding, creative engineering, just called open coding. Okay. Not every open coding requires you to understand at least something of what you're doing to do it well. Absolutely. And so which to be taught in school. It doesn't mean that every industry is going to collapse. It just means that there's going to be there's going to be pullback. And there's going to be a million different things that spread this sprout up out of nowhere.

So, yeah, I just think it's I think it's the future of what of what we're all going to going to have to deal with. Exciting times, man. Exciting times. I did a ton of stuff on pot on the on the gossip listener for pod ping this week. And but we just don't have time to talk about what we'll hit it next. If I just fired up, does it does it do its thing automatically? Or do I have to do a pull or something?

Yeah. So you if whenever you're going to upgrade, you just need to do the get pull origin and then wait, what do I do? What is the get get space pull space origin. So that just pulls the fresh copy of the main branch origin. Whoa. Oh, that's so cool. It just now you can just do cargo. Now you can do cargo build and then run it. Oh, I don't think so. What? Because you do need to open it's an open coding way to do this. Wait a minute. What happened here? Wait, what's my directory?

Oh, no, I have I have my own little script. Oh, you mean? Yeah. Yeah. I'm piping the output. I'm making it look pretty. That's why it shows me the feed. I can click on the feed for that episode. It shows me average per minute, average per five minutes, 10 minutes and per hour chosen. It's nicely formatted. Well, that's that's the thing. Let's see. Let me see if it's maybe make sure it's running. It's running now. It's running. I'm probably adding disk space or something.

Do you have archiving turned on? Oh, that's a flag. No, I don't. OK, I don't. I will. I'll turn I'll turn on a box with the flag. A sloperator. Sloperator. That is right. Prompt engineer. No sloperator. Yeah, baby. Yeah. Holy crap. What is it? Oh, man, it's got a whole I'm looking at open code. It's parsing the VTT, find timestamps. It's downloaded Whisper. It's going to do Whisper on the device. It's looking for key ad for phrases. She is going to.

OK, so here's here's the thing we'll talk about next week. Let me put this in the link in the boardroom. So that's that's the that's the most recent change. OK, yes. Events. Oh, yeah. What happened there? The site can't be reached. What just happened? Do it in a do it in a private browsing window. Oh, for some reason, Chrome and the the Chromium based browsers, they always want to force you into HTTPS and I don't have it. Oh, OK. I see. I see here. Oh, yeah. It's oh, it's in a browser.

How nice is that? Yeah. So this is a this is a technology that I was not familiar with called called server sent events. So this will be the backbone of the of the of the pod being firehose going forward on the gossip side. Oh, nice. Yeah. So we'll talk about that next week. All right. Well, thanks. And people, I. Yep. So I still haven't reopened the channel to my Albi hub. Sorry about that. But I do have runway running from the Cridler.

And I only see I don't know if so if you sent in a key send boost, I'm not seeing it. So I only see one. Let me just make sure I've got the latest. I got the key sends from from during the show. Oh, no, not from during the show. Yeah. Let me just see if anything new came in. OK, now I see eight eight hundred sets from Macintosh at Fountain sending you everything I left in my wallet. Sorry, it's not a lot. A note of encouragement about agents. Don't sleep on them.

OK, I'm putting the very finishing touches on my first real skill for the bots I'm building. The show producer bot builds notes based on my topic idea. The researcher bot gathers the stats. It's a Bitcoin podcast. So we mention a dozen or so different ones in an episode, all merged and posted in a private Git repo that my co -hosts and I pull from for the show prep. It's early. Bots have flaws, but this will be a huge part of AI. Next up is my coding bot, Steve.

Go podcast Steve, as in the was as in the was. Yes. OK. And that's all I have. I hit the delimiter right after that. So if if anyone boosted, I'm sorry, I don't have it. We got some PayPal's, we got the boys at Buzzsprout. Thousand bucks. Baller shot, caller 20 is blaze only baller. Keeping it alive, boys and girls of Buzzsprout. Thank you very much. Thank you. Thank you. Thank you. This is, of course, value for value. All this goes into maintaining the engine that is the podcast index.

I don't know. Did I did I mention that there was a worked with Tom to fix an issue where they said they changed their feed URL format a while, like a year ago or so from feeds dot buzzsprout dot com to RSS dot buzzsprout dot com. Oh, but the. But their pod, the pod pings they were sending still referenced the old feed. Oh, no, URL. Oh, no. Host name. So it still said feeds dot buzzsprout dot com.

So we were trying to figure out why their feeds were still coming into the index with the wrong host name. So we we finally figured out it was the pod ping. So he fixed it. And so everything's OK. That should be solved. Good, good, good. We love hearing that. We love when it's fixed. Yep. Let's see. Where am I? I got I lost my. Oh, here it is. Oh, Todd Cochran. Thirty dollars. Oh, Todd. It read it renamed from new media to Todd. It just says Todd Cochran. So I'm going to give no reason.

It's a boost from the grave. I'm assuming that that's autopay, that that's Rob just letting that thing. I'd like to know. I'd like to know. I would. I would. I wouldn't mind knowing. It's kind of it's kind of it's it's cool and creepy. It's really kind of. Yeah, it is touching. Gene Liverman, five bucks. Thank you, Gene. Appreciate you. See, Michael, Michael, why are these getting cut off? I think it's Michael Hall. Five dollars and 50 cents. Jeremy Gerds, five dollars and 50 cents.

And that's it for the pay pals. And we got a couple of boosts here. Let me sort that. I've got too many emails in my box. Twenty to twenty two from the ugly quacking duck. That's Bruce from podcast guru. He said, I enjoyed your guest, Brendan. Thanks for the episode. Seventy three. Seventy three. And we got the delimiter comic strip blogger. Twenty two thousand sets. Boom, boom. Through fountain, he says.

Howdy, Dave and Adam, please watch my video proof that large language models actually think to find this video search on YouTube channel for, quote, a arch wizard, end quote. Please like and subscribe. Yo, CSB smash that subscribe button. That's a little bit. That's a little abbreviated note from apparently one at one of his videos got over a thousand views. He said he was going viral. He's usually more loquacious. CSB, CSB viral, baby. He's going viral, viral. Yeah. And that's all we got.

And that's it. No other one else. That was that's the whole deal. Hmm. That's the whole deal. All right. Well, thank you to everyone who supports the the index. The podcast is just a front door funnel. Everything goes straight into the kitty. We keep it there to keep the machines running for as long as we can. The kids in the kitty, baby. Go to podcast index or down at the bottom. You see a big red donate button. Hit that. I'll take to the PayPal page.

And of course, we accept boosts, booster grams, on chain lightning, you name it. We're here for all of it. And we really appreciate it. And that will wrap it up. You've got another 30 minutes to get back to the office with traffic unless that is that wreck is done for. I don't know. I'm I'm glad you mentioned that because I would have forgotten and would have been like pinned. So I'm going to I'm going to go the back roads. All right. You go the back roads, taking the back roads.

Brother, thank you very much. Have a great weekend, Dave. All right. You too, man. And boardroom. Have a great weekend, everybody. We'll be back next Friday with more of podcasting 2.0, where we will dive into more of it, whatever it is, even if it stinks. Talk to you then. Bye bye, everybody. You have been listening to podcasting 2.0. Visit podcastindex.org for more information. Go podcasting! The crickler, the crickler at work.