Friday, 12 June 2026

Episode 263: Chat is Dead

Silicon Valley’s pivot toward local AI models and agent-based super-apps signals the end of the chatbot era as hardware giants like Apple and NVIDIA seize control.

By Podcasting 2.0 | 1h 29m listen | 30 chapters
Episode 263: Chat is Dead cover

About this episode

OpenAI is reportedly pivoting away from the ChatGPT interface toward a super-app model as internal sources declare that chat is dead. Adam Curry and Dave Jones analyze this shift alongside Apple’s WWDC strategy, which effectively rug-pulled major LLM providers by integrating local, privacy-focused AI models directly into the iPhone operating system. This transition marks a critical turning point for the industry as the initial AI bubble faces a massive financial sustainability test.

NVIDIA CEO Jensen Huang recently introduced RTX Spark to power local agents on Windows PCs, signaling a move toward home AI supercomputers that operate without meter anxiety. Meanwhile, Dave Jones details his implementation of HTTP signing using Cloudflare’s web-bot-auth protocol to resolve 415 errors and bot-blocking issues affecting the Podcast Index. The discussion also covers Anthropic’s expensive token-based API billing, Google’s recent eighty-billion-dollar equity raise, and the success of the AMP group in shifting industry metrics from downloads to active plays.

Technical hurdles extend into the physical world as Dave Jones faces a sixteen-thousand-dollar HVAC replacement after a proprietary mini-split board failure. Adam Curry shares stories of Texas property security involving septic technicians and booby traps, while the duo checks in on the status of Cameron and the IPFS Podcasting node network. The episode concludes with a debate on James Cridland’s proposal to deprecate SRT transcript formats in favor of VTT.


CHAPTER 01 / 30 Discussion

Podcasting 2.0 Episode 263, Dave Jones and Adam Curry Introduction

Adam Curry and Dave Jones open episode 263 of Podcasting 2.0 from their respective locations in Texas and Alabama. Jones describes feeling overwhelmed by his current workload, noting he only has about ninety minutes of focused development time per day.

podcasting 2.0· adam curry· dave jones· texas hill country· alabama· podcast index

00:00 Podcasting 2.0 for June 12th, 2026 Episode 263! Chad is dead! Well hello and welcome my friends into the boardroom The Board Meeting of Podcasting 2.0 Just a bunch of teenagers here pay no attention to us Besides the fact that we know actually what the future of podcasting is going to be because We're creating it In fact, we are the only boardroom that you play for more than 30 seconds. I'm Adam Curry here in the heart of the Texas Hill Country and in Alabama! The man who's so far underwater he needs a snorkel Say hello to my friend on the other end The one, the only Mr. Daaaaaaaamnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnnn

00:56 You are. That was impressive! I like it Yeah, yeah It's Friday once again everybody Welcome to Podcasting 2.0 Dave how you doing? I saw notes on the podcast index dot social and whenever you say I'm so for underwater man I am like oh boy Oh boy I live in that state perpetually So i'm used to it but I just want to said sometimes I need to say Sometimes I need to set expectations about how fast code or something is going to happen because, you know, a lot going on. There's a lot going on...I always feel bad because people

01:38 email me or post something to me and want something, and I'm like you know. I really want to give that to you quickly but i just Right only get about an hour and a half a day max It's crazy to see how much output You have actually based upon that hour-hour and a half a day? I've been able to increase the output a little bit with AI coding. Ah, there it is! Increased productivity no one's replacing Dave he's just getting more productive and when you get more productive there's less time for your wife yeah that's really what it is I know The um...the...I think the AI coding has worked pretty well for my use cases because

CHAPTER 02 / 30 Discussion

AI-Assisted Coding and Repository Management Patterns

Dave Jones explains how he uses AI agents to increase his coding productivity within the existing 100,000-line Podcast Index repository. He details a strict process where the AI is prohibited from writing directly to Git, instead following a spec-driven development plan where Jones reviews all generated SQL and API endpoints.

ai coding· github· sql· api endpoints· repository· software development

02:33 There was already a large existing code base which it can draw patterns from. I mean, there's already 100 thousand lines of code in the podcast index repositories to in which to just sort of like route around, root around and delete stuff. Yeah I say I need an end point that does this on the API and it can just say well it can go look at the existing stuff and all the SQL statements because again thats not hard But wait when you say that do you let it build

03:19 I always go through the spec driven development process. Always have it create a plan. Well, yeah this is... Yeah it's a plan but also like what my typical pattern is I take... I don't let...I never let the AI agent touch the code base without a lot of, excuse me, touch the repository. So everything it is prohibited from writing anything to Git. Except for when it does it? Except for when it doesn't and then apologizes for it after the fact That's my favorite Oh man I'm sorry! I'll write this down so that will never happen again

04:11 And then it'll do it and say, you know what? You're right. I should have seen that. I have a note about that. Oh yeah! How many times I've been through that... It's just a lie unless you specifically say in your prompt and remember don't do this you've got a memory file about it. Eventually it will just do it. These things need a human to drive it. It needs a human when he, you know... When it comes back from if you clear out the context which I do a lot Comes back from contexts and then I'll have to remind it okay Remember we have memory about this go look for it before oh yeah that's right Yeah You're right? I don't actually know any of that Let me go check it out because it just doesn't know and it literally doesn't know it's not intelligent. It's a robot And

05:04 That's why I never let it touch Git. Of course, you have to have your stuff backed up. You must have backups of everything. No matter what. So I never let it touch Git and then... I think coders who are very successful with using AI agents to do coding are people who already think in small bite-sized chunks about projects? Small minded people, yes. People with small windows of thought processing, yes. Because you already... the process you already take is one where you take a large thing and boil it down into into small series of steps

06:02 And so you just, you naturally sort of think and talk that way already. And then you take that and apply it to a very... A code repository that's been around for years and is very mature. Mature in the sense of it has established patterns not that those patterns are necessarily optimized but it does have a set of patterns already. Yes Because then like, I mean adding a new endpoint to an API is typically not that difficult if all you're needing is data presented in a new way. I found... You know how skills work obviously? I've been able to make something called doctrine or what Claude calls doctrine So an example of doctrine is my example

CHAPTER 03 / 30 Discussion

AI Output Skepticism and Personal Productivity Tools

The hosts discuss their conflicting feelings toward AI technology, balancing significant productivity gains against a deep dislike for AI-generated text in emails and documents. One host describes using "doctrine" in Claude to handle repetitive tasks across different podcast projects while maintaining a human-driven workflow.

claude· ai generated content· productivity· email· automation

06:59 do the credits. So it's all uppercase, credits and then now depending on what show I'm working on that's how it adjusted so it's not just a skill specifically for no agenda for podcasting 2.0 and it understands doctrine so it kind of handles it as a skill and has the scripts already built It just knows the variable is where I'm at at that moment And that's worked out pretty well for me I think obviously AI is dominating everything in life, all of our lives are just so saturated with it. Yes. I don't think I've ever been so split-minded about a single piece of technology as I am with this. Wow!

07:51 I mean on a daily basis, I get tons of use out of it but i also just despise it down to my core. Of who? I hate it! Right down that your backbone... I hate yes my very cerebrospinal fluid quivers with disgust at AI Generated output. Well, I cannot stand it when somebody sends me Oh like a document or an email Yeah that starts off and you immediately within the first two no those AI yeah And it's in its like 12 paragraphs of thing. I'm like man. I can't read this well we had

CHAPTER 04 / 30 Discussion

Godcaster Live Stream Debugging and Support Bot Theory

Adam Curry recounts debugging a Godcaster live stream issue involving Live365 metadata and "off air" status messages. He proposes a future where personal AI support bots communicate directly with company help desk bots to resolve technical conflicts without human mediation, though Jones expresses skepticism regarding the complexity of system errors.

godcaster· live365· icecast· metadata· tech support· automation

08:40 This is interesting. I know you followed the Godcaster chat that we had this morning There was yeah, it was kind of fun how they came in like it's off-air bits broken and that was kind of the message That we got and so I go to investigate let me see what's going on It was a live stream And I didn't know if it was broken as you know for was not You know what off air meant And so after probably 25 minutes, I figured out that what was happening is a Godcaster live stream station. We have a product called Your Town Live Stream was being relayed by Live365 which is correct for licensing but it was actually Live365 who for whatever reason their system was just putting out metadata saying off air

09:31 without passing through the metadata which comes through, what is it called? CTY I think it is. You know that's the metadata that comes along with IceCast and so I told the robot write up a document for the help desk on that end if the user can't figure out how to restart their relay or whatever it is. And so, it wrote a paragraph I don't know if you saw that? If you had time to look at it But I was curious A, if he thought Oh wow this isn't AI generated thing So did you see it or didn't see it? I was split-minded with... Or split brain with work Okay Let me go back and see Yeah Well

10:18 Oh, is this the one? No I did not see this. Okay you said I've debugged everything. I haven't read any of this yet. Here's a template... Does it scream AI to you? The one that says okay I've debugged everything? Yeah what's underneath that? What they would have to give to the help desk at Live 365? Not necessarily. So as I was posting that, I thought to myself well what we need here is the world where my robot talks to the robot at the help desk at Live365 and they just resolve it without two other people sitting in the middle That would have been a beautiful thing

11:12 Because now Gordon has to talk to the station guy, the station guy has to go talk to Live365 if he can't figure it out. He has to email in this document I can see a world where that would have been very handy Man...I'm gonna have think about that one I mean you know Handy and Well less That would be more productive for everybody involved We could do other things with that time It de It depends on what you mean, like... Productivity has a lot of consequences. I am not going to make you feel good about this at all, I can tell.

12:05 So the, I'm with you on emails. I'm with you on songs. I'm with you on images. I'm with you on all of that. That's annoying. Well it's because I live my life in 50% of my job on a daily basis for the last 27 years has been tech support Right and now your getting just longer emails and uh...with superlatives that you don't need Um, that is more work slop. Work slope? Yeah and less help desks. On the help desk side of things the issue I think is that you... The way software...I think the reason I'm having a reaction to what you're saying about just let my support bot talk to that support bot and then let them resolve it Errors in systems have many

13:08 they look the same. You can have a large number of problems that manifest themselves in the same output. Sure, the same dialogue box. Exactly and so like if if the customer is not involved or the end user or whatever is not involved or if they're the ones that trigger it themselves. Yeah, I understand. It just all falls apart like it's kind of a false equivalency let me give you a podcasting example of productivity so

CHAPTER 05 / 30 Discussion

Automated Podcast Editing for Pastor Jimmy's Daily Grinds

Adam Curry describes automating the production of "Daily Grind" devotionals for Pastor Jimmy using a custom script. The AI-driven process handles transcription, scripture identification, flub removal, music bed placement, and ID3 tagging, reducing a three-hour manual editing task to thirty minutes of quality assurance.

pastor jimmy· podcast editing· ffmpeg· whisper· id3 tags· automation

13:53 Every month or so, Pastor Jimmy records 23 daily grinds for me. This is a one-minute devotional I have a script that takes the most... you know the... I dump all 23 of these into a bin and then the script as a cron job it runs it says okay let me get the next number in the sequence It actually does some fun things. It whispers it and then figures out what the scripture is about, and then it does the ID3 tags uploads to the server adds the episode to the RSS feed so that's already kind of cool The fact that its you know this less work because who wants to be titling

14:38 You know, all of this like 23 episodes. So when he sends it to me, he sends it to me as one big file and he's got flubs, he's got pickups and so that used to be about a two-and-a-half to three hour job I'm pretty fast at editing. So yeah, pickup...I can see on the waveform oh we picked it up okay I'll set out point there in point their click let me listen okay that edits good then I'd have to put music blocks on each of those then after cut each one out output it, and then it goes into the bin. So about 23 minutes. So I did an experiment! I said robot here's a file there's twenty three daily grinds on here their pickups their flubs their restarts... I want you to cut each one put the music under it make sure there are no flubs title it do the ID3 tags in them put it in the bin brother

15:29 It had, out of 23 there were two... because all I had to do was just audition them in real time listening to them. There were two that had a little audio flub at the beginning just a little glitch of you know little snippet of audio and one where it did an edit but it picked up a double breath otherwise flawless total time for Adam about 30 minutes most of it just spent listening to him then make sure the output was correct This was mind-blowing to me. Okay, so what you're describing is a process that you could have written code for yourself by hand? Yes it wrote code for it exactly Exactly right So... You could have written code that does this yourself

CHAPTER 06 / 30 Discussion

Linux-Based Podcast Studio Automation and Media Tools

A host details a new soundproof studio project utilizing a native Linux environment and custom-built software to automate podcast engineering for non-technical users. The workflow relies on FFmpeg for media processing, Whisper for transcripts, and Python scripts to manage the end-to-end recording and distribution pipeline.

linux· ffmpeg· whisper· python· together ai· podcast engineering

16:20 Any process that you can envision as a automated code-based process, the LLM can do it and just do it very very fast. Yeah exactly! And then it has this built in and now I'm able to reuse that code because we've made this into a doctrine so now it understands that code and so now I'm taking the next step Because I'm setting up a studio at the church, they've given me an office and it's already soundproofed and everything because people want to do podcasts. And like you know the last thing I want to be is the podcast engineer guy. You feel like this? Now the beautiful combination for me is Claude with Marci so its Linux, I'm running it native

17:08 I don't buy software, don't look for software. Just say build this software for me." So I said, I need a piece of software that is idiot proof you click on the button, you select your podcast, you hit record if you flub mess up take it do over etc people have to worry about that they just keep recording and then when they're done they hit stop and then you go detect all the pickups, anything necessary cut off the beginning cut off the end have them give an audition of any edits that were made and ultimately just ID3 tag it put the image in boom good to go. And then next step will be you know poop it right up to RSS.com I mean this is

17:56 The three tools I have which is FFmpeg Support those guys send those guys money. Ffmpeg is just one of the most beautiful tools I've ever seen in my life It does everything Whisper Which I do on, you know just to get transcripts of everything. I live in audio So that's that's the next tool for me and I use together dot AI for super fast You know can do an hour in like 45 seconds And Python and then the bot the robot and it's changing my life so I understand where you're coming from

CHAPTER 07 / 30 Discussion

AI Psychosis and the Erasure of Development Drudgery

Dave Jones analyzes the "AI psychosis" affecting the industry, distinguishing between genuine productivity gains and the mere automation of repetitive "copy-paste" tasks. He argues that while AI excels at writing boilerplate code that developers previously dreaded, it struggles with creating high-quality, customer-facing products.

ai psychosis· software development· automation· llm· code production

18:35 I also see how it's, because I can't write code like that. Yeah maybe I could kind of get somewhere but no not really But i understand what its doing and understand how to tell it to go reuse that code And it just blows my mind how productive I've become This is the issue with the current state of AI psychosis that has just gripped the entire known world is the conflation between the incredible gain in productivity that you just described and I would describe it as not even really a gain in productivity so much as an erasure of the things that you used to dread.

19:34 Like, so you get to a point in like in a code base where when somebody describes where you begin to build out a feature and what you see as the developer is just a whole lot of copying pasting. You're like, this feature shares so many things in common with this other feature and I'm just going to have to go in there and I'm gonna have to build a different version of the same class. I'm gonna have to import these same methods. Right? I'm gonna have to build the same unit tests... I'm gonna- you're just doing the same thing

20:11 with slight tweaks over and over and over again, which is exactly the process you're describing. The going out and getting clips that getting, you know, the automatic edits these are just things that it's not... And so but then that process which is completely valid that an agent can take a process like that Write all the code that does, that automates it for you. Which is code that you would have never taken the time to write because it would have taken a hundred hours No, you used to write it for me until you got busy That's how this is I'm using the exact same process as you and I have used for 16 years You're just talking to the agent instead of me. Yeah But but I call it Dave just so you know Hello Dave

21:11 So that, but the problem when the problem comes in is that gets conflated with actual customer facing products that are supposed to be good. Yeah like you described customer support everybody despises AI agent customer support. Sure! It drives you, it drives you mad which I was good at it Which I wasn't suggesting Of course oh no I know what you're saying but I'm saying like that's where there's a... You have completely streamlinable automatable code production on one end that is getting rid of rote

CHAPTER 08 / 30 Discussion

Human Friction in Customer Support and Call Center AI

The discussion shifts to the role of human interaction in customer service, with Adam Curry praising high-end concierge services like American Express. They contrast this with the trend of US-based call centers firing staff in favor of AI agents, noting that bots often fail to understand the nuances and "lingo" of non-technical users.

customer support· call centers· american express· tech support· automation

21:59 copy-paste work for developers and content producers. Also, when it comes to helpdesk people hate helpdesks and help desks hate people searching for help I feel we're just removing human-to-human friction yeah yeah I mean I book travel...I call American Express And I talk to a human being and I joke with them, and have a good time. And always tell him please put me through your robot so that I can give you a good review. For those of you listening to this recording give Jacob a raise! This man deserves a raise. I'm always doing that it's very enjoyable for me. When people don't know... technically oriented people do not understand

22:48 that non-technically oriented people don't know how to describe problems. No kidding? They don't know and what you get is, hey this thing doesn't work. Well no... And thus starts the question and answer session Yeah back and forth back and forth back and forth It's like it takes a human to understand the flow of the conversation that happens during a support encounter. Yeah and sort of glean out because every person that you talk to is also different, like this person describes things in this way this person could describe the same problem in a completely different way and it takes a human to understand this persons lingo and pick up on oh I see

23:43 When they say this, what they really mean is this other thing. And there's some people you talk to on a support encounter where you know that they are lying to you. You just know it, the first thing you say is did you reboot your machine? Yeah yeah I did and your like no no you didn't you're lying to me. You go an check their logs, pull up their machine in real time says its it's been up for 12 days without a reboot. And so the bot will never know this, but you know it because you have an instinct about the way this person is talking to me." But okay now I'll just say from a business buddy of mine sells AI based systems to call centers. These are US-based call centers but they're basically sweatshops and those guys are firing 65%

24:43 Once they have the agent or whatever figures out the corpus of the company, and they're very happy with it. I personally depending on what the company is, there's levels where you want a human being. It actually took me longer to get through an operator, human being. The bot kept cock blocking me But I see from a business perspective where that's like the one instance where I say, okay depending on what your company is you know for travel American Express travel. I think I pay fees and want to talk to a person and feel good about that. A concierge, I'm sorry, a travel concierge they bill it that way And I don't mind paying the fee for that but all this other stuff there are so many different levels

CHAPTER 09 / 30 Discussion

Freedom Controller Longevity and Vibe Coding

The hosts reflect on the Freedom Controller, a tool built years ago that has maintained nearly 2,000 days of uptime. They compare traditional development to "vibe coding," where concepts are turned into products in thirty minutes, using a recent Zoom video processing script as a primary example.

freedom controller· vibe coding· zoom· ffmpeg· software engineering

25:33 that I just find super enjoyable. My creative mind is, look we built the Freedom Controller it took us years to do and there was newness for you because you know...I don't think he'd ever really built something like this and it was a lot of trial-and-error and me testing stuff and breaking stuff That process for me I now have You know, the freedom controller could be vibe coded in an hour and a half. Today? Yeah today complete I mean but that's the point you know with higher quality with higher code quality possibly although that thing is rock solid man it has been running for eight years like never reboot uptime on its like 1900 days or something like that so anyway i mean i'm just saying for what

26:23 I mean like, and just the other example. I did this...I think you saw that We did a Godcaster U it's a Zoom call. Zoom by the way teaches me a lot about changing features and removing features Whoa! Those guys change stuff all the time And then it throws you for a loop man It's like ah now I can't share how do I share my... So go through these horrible loops of things But it's at the end of spits out a video and then we want to put that into our Godcaster you RSS feed And I did the same thing. I said hey this video starts at 38 minutes It ends at 1 hour in 59 minute or 59 minutes, so whatever it was 127 I Want you to cut off the beginning but cut off the end put a god caster you logo on the top left-hand corner? I want captions

27:14 in the lower half or the lower third and output it for web format. And it did it! That was amazing, that's all SFM Pig 2. Yeah of course it did because it's a known issue with it thats clearly scriptable. But do you see how me as an as a frustrated non-coder just enjoys that so much? Oh, 100% yes. Yes absolutely I want you to find some joy like that Dave You have no... Can i give you some of my joy The... the... I don't know that people who've been writing code for a long time really find any joy in the AI coding experience Like I think that

28:08 because it's not some, what you're doing. Because what you've done is you've replaced copy and pasting with just a chat session where are saying yes? Yes. It gives me multiple choice questions. No no yes and then you say write this spec document to an HTML file so I can print it. Yeah now I understand but for me its like Concept and idea and brain fart to product within half an hour. I mean that's mind-boggling And it's just for me, I'm not gonna sell this. I'm not gonna put it on the App Store You know that's that's a whole different level. I don't think that would work you know The idea of we'll have one person billion dollar companies yeah, I'm not so sure about that I

CHAPTER 10 / 30 Discussion

OpenAI Strategy Shift and the Death of Chat

OpenAI reportedly plans to overhaul ChatGPT, moving away from the chatbot interface toward a "super app" model that emphasizes coding tools and AI agents. An internal source claims "chat is dead," as the company seeks higher-value products to generate more revenue from its billion-user base.

openai· chatgpt· super app· elon musk· ai agents

29:02 Yeah, I think AirPP. I agree with you 100%. It does take the fun away out of coding And it's not that there was a ton of fun in most coding already. Most coding was just a lot of rehashing of the same patterns over and over, but you would have moments of fun where... Where I did yeah Yeah, where you'd run up against a real problem… And then walk around the block for an hour like what am I doing wrong? You come back and I know! I've been there too And now that's gone. With AppleScript? Now it's just an answer, now it's just a yes. Yeah I know the challenge is a bit gone, I understand But at this...I guess the bigger issue though is that I think I said like a couple of years ago in the boardroom that podcast companies that have baked API calls to LLM providers into their process

30:09 At some point, that is going to become... You better have a plan B. And well yeah sure again the customer facing product there's all kinds of risk there Well no you did you see what the open AI said this past week they said chat is dead Did you say that? No! That's their main thing Let's see, I should have pulled that article. Yeah chat is dead open AI preps an overhaul of chat GPT so basically they're saying that

30:51 They're just the chatbot aspect. Let's see, OpenAI is preparing the biggest overall chat GPT since its launch kicked off blah blah blah The company intends to transform the chat bot into a quote super app unquote. Ah ha ha they using the Elon Musk super app moniker oh very fun yeah That combines coding tools and AI agents adding products that executives believe will generate more revenue More tokenage Yeah, so chat is dead said one senior open AI employee. Open AI executives increasingly view chat GPT which has attracted nearly a billion users as a gateway to introduce users to higher value products. Chat is dead long live fireside!

CHAPTER 11 / 30 Discussion

Apple WWDC and the Local AI Model Rug Pull

Dave Jones argues that Apple's recent WWDC announcements effectively "rug pulled" major LLM providers by integrating local AI models directly into the iPhone operating system. By offering privacy-focused, on-device processing and MLX support for developers, Apple may eliminate the need for users to pay for third-party services like ChatGPT or Claude.

apple· wwdc· siri· llm· mlx· privacy

31:39 The issue that I see after the WWDC stuff this week... Which seemed pretty underwhelming. It didn't even, when i was reading about it, I didn't even want to look at the keynote or what used to be a keynote which is now like still a COVID era based video take away from that, from WWDC was totally different than yours. Okay well I didn't see it so this is just the report of what i got. I didn't see it. I look at what Spurlock said and like oh that doesn't sound very exciting Well a lot of people were saying that it was boring but... The boringness of it may have been true but the impact of it I think is going to reverberate It's..I think we just What? I think we've just witnessed

32:32 with the, with what Apple is doing with AI. Can I tell you what I hoped for and then you tell me what happened? What I was hoping for, was for them to say look at our machines. These are the best machines with unified memory you'll run our MLX LMNOP agent it will create anything you need software it'll be your all-in-one box that's what I was hoping to hear and I didn't get any takeaway from the postings or articles There's a lot of that, what you just described. A lot of that is in the individual developer sessions which you can watch on YouTube. Okay. There are a lot of dedicated sessions that show you how to get the most out of MLX for local models even going so far... there are even sessions on chaining Mac minis together to get parallel processing like they've went pretty in-depth on that on local model support

33:32 Um, so I don't think that that part was a disappointment at all. Okay? I think the takeaway that I had overall is that they just rug pulled the big LLM model providers. Ooh, well that's good because what did they just say? They are hosting their own models in Google's cloud and with privacy, they say Google cannot even access the data. And so if what they demoed actually works and works well nobody will ever use a chat GPT chat bot again why would you? Wow can I play some clips about this when you're done with your analysis

34:33 Yeah. So if the hundreds of millions of iPhone users suddenly can just pull down their Siri button and say, you know blah blah blah blah blah... The thing they would say to a chat GPT chat bot in the past or Anthropic or whatever which is still the predominant way that most people interact with these AI model providers like Anthropic and OpenAI If they can do that Why would they ever pay for chat GPT or even think about it? It's good. It's it's over It's a is just over I will because because that because Android will then do the same thing And it's done and it's done you you have no more reason for open AI and Claude in Anthropic to even exist so Jensen Wang who for purposes of this discussion, I'll call Jensen jobs is

CHAPTER 12 / 30 Discussion

NVIDIA RTX Spark and the Reinvention of the PC

NVIDIA CEO Jensen Huang introduced RTX Spark, a new line of AI-integrated Windows PCs designed to run local agents 24/7 without "meter anxiety." The vision includes a personal AI agent that manages home systems, travel booking, and security, utilizing evolving models like Nemotron.

nvidia· jensen huang· rtx spark· cuda· ai agents· dgx station

35:30 Because man, he wears the leather jacket and he tries to do one more thing. And so he says but there's more So we did this big thing in Korea I think it was like it's a big computing conference and he went through 87 minutes of them this chip that chip and this partnership and is great and But then the last three minutes He does this. But that's not all, that's not all One more thing! RTX Spark is a reinvention of laptop but in fact- Laptop it's a reinvention of laptop Not of a laptop or the- It's like iPhone Yeah exactly Laptop Of laptop

36:16 But in fact, Microsoft NVIDIA is reinventing all of PC. And today we're announcing a whole new line three revolutionary Windows machines covering desktop laptop and workstations all 100% windows compatible 100% CUDA 100% Nvidia AI Tensor Core Everything that runs, that you see that runs on NVIDIA and all these different platforms around the world runs here. This is the first completely re-engineered reinvented line of PCs That has happened in 40 years Okay so this had my attention although I'm like okay right You've got a partnership with Windows blah blah whatever but then... This agent could run 24 7 meter free And you could download your agent You can raise your lobster in here Meter free This is your claw

37:09 It's running all the time, no meter anxiety and it's sitting here connected to your whole house. Connected to your laptop." Now so this I'm like wow okay that's exactly what i've been waiting for! No meter anxiety is running locally But then the guy, he gets nutty. Connected to your display all the cameras your dryer... What? Water cooler... You're trying! My water cooler?! No what?! Everything whatever you want Your security system all connected to this Who gives a crap about that And this becomes your personal AI agent

37:49 It gets smarter and smarter and smarter over time because today we have Nemo tron 3 ultra tomorrow We have a new Motron for the motor on 5 6 And we just keep getting us smarter and smarter smarter, and meanwhile this is sitting at home helping you do things if you want to book a travel no problem book of travel and If you if you want an incredible system. This is a DGX station that for windows. Or it just doesn't see an act because he told I see a future. I see your future and here's the so I See your future like yeah, yeah, I want something that just runs locally no media anxiety big programs It does stuff for me creates applications Oh, I can think of a million things to make my life better with that but here's his vision something is happening here Yes remember

CHAPTER 13 / 30 Discussion

The Future of Home AI Supercomputers

Adam Curry and Dave Jones discuss the concept of every household eventually owning an AI supercomputer, similar to how homes currently have theaters or game consoles. They mock the long-standing promise of smart refrigerators while acknowledging the industry's pivot toward local, hardware-based AI processing.

home theater· supercomputer· ai assistants· rss.com· smart home

38:51 15-20 years ago, we used to have an idea called a phone. Today we have an idea called a PC today when you think about your phone The one thing you don't do with it is make phone calls You do just about everything else. And so that phone means something very different to you than a phone of the past It's kind of insulting to technical people, but okay I am certain what's gonna happen here is that the PC 10 years from now and the PC that you think about today? A tool where they you launch applications click and type and this PC is going to be completely different

39:29 Here's my theory. I can totally imagine just as every house today has a home theater Do you have a home theater Dave? Yeah, oh yeah, yeah, I guess Do you have like seats that recline with the cupholders and everything? Yeah, I've got a podcast room with my food is currently sitting on the floor I have a big TV Oh that count where many houses of home theaters big TVs lawnmowers Dishwashers dishwasher I could totally imagine that someday there's actually an AI supercomputer in your house, and it's running all of your agents It's running all of your assistants What's it doing? And they're doing all kinds of things for you All the time and you have to have it in your house Just like you have a home theater in your house. You have stereos in your house

40:14 You have game consoles in your house? ASU computer in your house brought to you by RSS.com The only thing he's missing is saying it will know when your milk is out, in the refrigerator It will order the milk and it will automatically get replenished which has been the promise since 1997 This is dumb. It's dumb, but he is showing us where they're going They're already thinking hey man We're not gonna be relying on these local models on these large language models and what they call them model labs Foundation Labs You know why they did all this? I think they are pivoting because the bubbles already popped Apple popped it

CHAPTER 14 / 30 Discussion

The AI Bubble Burst and Financial Sustainability

Citing analysis by Ed Zitron, the hosts discuss the financial instability of OpenAI and Anthropic, noting they require hundreds of billions in annual revenue to remain sustainable. They posit that Apple's entry into the market has popped the AI bubble, forcing companies like NVIDIA to pivot toward hardware-based inference in laptops and appliances.

ed zitron· anthropic· openai· revenue· ai bubble

40:59 Apple popped the bubble. They knew it was coming, the open AI and chat. So in order to be as Zitron has done this work. Zitron is awesome. He's hilarious so he's done this work in the financials of OpenAI and Anthropic they need to have 200 billion a year in revenue by the end of 2027 to be sustainable as companies. They have 1.1 trillion dollars in commute and compute commits, and that is money that has already been paid. And so

41:47 What Nvidia sees is that model of being a large inference provider is no longer sustainable. That is not a valid business model that will proceed beyond the next two years, and so they are pivoting to PCs. They're pivoting to cram AI into laptops. Yes! What do we need? Laptops That's what we need in order to keep this thing going because Nvidia Nvidia is the one that comes out smelling like a rose every single time because they make the chip Yeah, of course and they can just stick it anywhere. They can stick it in laptop They can step it in refrigerator. They can stick it in washer dryer in white hand dryer Yeah, they can they can stay here. I think you're sticking in bicycle What?

42:43 They can put chips everywhere. I love it! They don't care what it is We just gotta keep talking like this... They can stick in an app App? They can stick it... Yeah, they could stick it in keyboard Mouse. Stick it in mouse! AI mouse. Reinventing mouse and that's all they need, that's all they need is a place to stick a chip and they'll do it In head we put a chip We're gonna reinvent head Which is what the 80s did, but that's another story. I don't... I just think- I make the jokes around here. Oh God yeah. But the bubble has popped and Apple popped it. Bold statement. Bold statement. And Apple... Mark my words when they release the AI enabled Siri

43:50 That's the end of this whole experiment. It just is because then the only people that'll be left who are paying for inference are gonna be API users... Right, idiots! Which it was not a sustainable business it's not enough money No, it's just not. I cannot wait for the bubble to, for this whole thing to unravel. Because we, the sooner we can just get rid of all this nonsense and all the marketing

CHAPTER 15 / 30 Discussion

Anthropic Strategy and Token-Based API Billing

The hosts critique Anthropic's business strategy, suggesting the company is pushing users toward expensive token-based API billing for models like "Fable 5." They warn podcast developers to have a "Plan B" and use flexible API routers, as current inference provider business models may not be sustainable long-term.

anthropic· claude· fable 5· api· token billing

44:28 crap and go and just get down to brass tacks. So are you currently doing your any coding that you're using AI for? You're doing that with laptop, you're doing that with RSS DGX computer? Yeah, yes yeah with desktop. I'm doing it's about...I'm using the DGX for inference for everything podcast index related. Wow! That's code as code... That's QIN 3-6, 20 billion? QIN 3-6 35B A3b mixed MOE model What agent? Pi. Pi right and you've customized that Right Everything else I'm still using just Claude because i'v already got stuff built up in those memory files and all kinds of stuff

45:26 But, and honestly it's get while the getting is good. Yeah because this I think the Anthropic strategy is clearly go public put money in pocket. Nvidia put chip and laptop yeah Anthropic put money in pockets and get the hell out of Dodge as fast as possible but They're model, bad use of terms but their strategy is to me is clear. They introduce more and more big scary models that we can't let you use for very long

46:12 And because they're so powerful, it'll like kill children. That was the stupidest marketing ploy ever! Well they did it again! They did it with Fable 5. Well Fable 5 is the artist formerly known as Mythos Yeah, they're going to dump the Fable. So they released Fable 5 and they're like well this look this thing is so powerful you gotta pay extra for it You gotta switch to let you get a switch to API based billing Oh is that? Is that the only way to access it? You have to use it via API you get two weeks free and then it goes token based billing. Wow. Fable five. Wow, yeah well they're going to push everybody took a base billing but they can't come out and say it because they're going to lose customers so what they have to do is say if you want our best model

47:04 you got to go to token-based billing. And really their quote, best model is just sort of like a rehash at the same... Of the existing model except it uses more tokens. But cyber? Yeah okay yeah now this whole thing's a joke and I'm so ready for it to be over but but the warning stands if you are If you have built a podcast product on inference providers using APIs you better be using some sort of flexible API router where you can switch over to different providers or find a plan B because if your back end relies on chat GPT, or flawed it's not... You better have a plan B. Right. Because those are not sustainable business models they're just not Okay I mean

CHAPTER 16 / 30 Discussion

Tech Industry Financials and Google's Equity Raise

A brief discussion covers the underlying financial motives for AI adoption in big tech, specifically the desire to reduce headcount. They highlight a recent $80 billion equity raise by Google, noting that a significant portion was reportedly allocated to payroll, signaling instability in the sector.

google· equity raise· payroll· tech layoffs· financials

48:08 I think we've exhausted the topic. I'm tired now. I mean, I had some clips on it but i think im just going to jettison those clips because its the same crap that everybody already knows. You know like me and you talked about the real issue here is not... The real issue here is that these large companies these large tech companies they need this thing to work for themselves You know, because they have financial their their financial life is wrapped up in being able to fire people. Yeah that's what they do yeah they have to

48:53 Because that Google $80 billion equity raise, nobody talked about the fact that 40 billion of that was for payroll. Well the No Agenda show talked about it because you told me about it and I talked about it so not nobody or anybody but we're nobody 50% of the 80 billion was for payroll yeah does that make any sense from for a company Literally Prince money money. Yeah, they're not all is not well in the tech in the tech financial world It's just not You want to talk about? 30 seconds versus 60 seconds. No, this is the amp thing Well Spotify has adopted it that changes everything that you know, I'm this super secret whatsapp group is hilarious

CHAPTER 17 / 30 Discussion

WhatsApp Groups and the Podcast Data Collective

Adam Curry comments on the lack of productivity in industry WhatsApp groups and mentions the Podcast Data Collective. Dave Jones expresses mixed feelings about the collective, citing the difficulty of "herding cats" when trying to get developers to participate in new industry-wide endeavors.

whatsapp· podcast data collective· sam sethi· james cridland· industry groups

49:47 Everyone keeps posting. Why? Because there's no actual conversation I mean, there's still an ongoing thread about some doofus who says I'm the podfather You know and I haven't even said anything in this thread everyone was like this guy is an idiot This is no good we already got one We don't need another one you know it's like all of these groups And I say it with love and respect they go nowhere It's just like, you know opinion. Opinion. Opinion. Opinion. Maybe I should turn it into a podcast and see what that sounds like. Turn that WhatsApp thread into a podcast but they're not really discussed... Well, I gotta add this guy. Hey someone from Amazon is here! Okay well that's gonna fix the problem. You know? The Podcast Data Collective which was interesting to hear Sam and James' thoughts on that

50:43 I don't know if you listen to power, probably don't have time for power. I listened a little bit of it... ummm....I don't know man! I've got mixed feelings on it On what? On the whole idea? On the collective yeah Well if your not going to do it won't happen so that's fine Well no no No if you don't do it will never happen That just and I'm okay with that I don't have ads I don't care But I also understand why you'd be like ehhh I don't feel like getting into doing or doing it I get it I feel like it's a worthwhile thing to do and i'm open to it. I keep having these thoughts, like... You know is..I don't know just- I'm afraid participation is just going to be so hard. I mean we have developers that's like hurting cats you know? And I'm afraid its just gonna be like a huge endeavor to try and get people on board. Like I think im sort of like

CHAPTER 18 / 30 Discussion

HTTP Signing Controversy and Developer Pushback

Dave Jones discusses the negative reaction from prominent developers like Marco Arment regarding a proposal for HTTP signing in podcasting. Critics fear the technology could be used to block specific hosts or apps, leading to a lack of consensus on the GitHub proposal.

http signing· github· marco arment· podcast index· open source

51:51 A little bit down on life. No, no, no. Uh, I think I'm a little bit down on new like pushing new endeavors because of the reaction to the HTTP signing thing. Oh, I, I like everything that was negative. What was the negativity? I like the whole idea. What is the actual negative pushback? Why am i in the minority, what is the issue? Well nobody wants... Nobody wants signing? Not nobody but a majority of people

52:38 Because they're afraid it will be used to block certain hosts and or apps. Is that the bottom line? Yeah, so well I mean one big thing i did this week was I put signing into our aggregators Now signing is something that Cloudflare already uses correct? yeah because say okay so what I did this week What Dave did this week Well, Eric I mean what you're saying. Eric said was it negative? There is criticism of the idea but don't think it was negative. It's hard to not take it negative... What I mean by negative is like when the proposal came out it was like I will die before I will put this in my app

53:29 Who said that? Marco and then other people shared, other people sort of shared the... I will die before i put this in my head. Yeah and other people share the sentiment And they are welcome to their opinion That is not a problem Was this a GitHub thing Oh yeah So see im a happy guy because I don't go on Github But I think the issue though was that it was just, there was such a like... There was no discussion really. It was like not doing it see ya! Right And so I mean there were there was lively beneficial debate back and forth and just like we talked about with Spurlock last week but but there was this strong sentiment from you know John and Marco and others that were like this can't No We're not doing this This is just not a good idea

CHAPTER 19 / 30 Discussion

Debugging 415 Errors and Cloudflare Bot Blocking

Jones details his struggle with "415 Unsupported Media Type" errors affecting the Podcast Index aggregator. After testing various user agents and IP addresses from Linode, he suspects Cloudflare is flagging the aggregator as a malicious bot, leading him to apply for their verified bots program.

cloudflare· 415 error· linode· user agent· bot detection

54:31 And so I was, so then we've been fighting this problem with the aggregator. Yeah people don't understand your particular pain they don't have your pain and i don't have theirs to be fair you know? I don't have their pain either You mean they're actually making money and you're not that's painful for them well Spurlock is not making a dime he's losing money just like us yeah but um The problem we have with the aggregator is that we get blocked through CloudFlare sometimes. Now, I don't know if that's what this latest issue I've been fighting is because part of it makes no sense. That's a 415 error? Yes and invalid media type which would be a very strange error to give back

55:29 If it thought that you were a bot, like why 415? There's like 403s, 401s. What does 415 stand for? Invalid media type I want to ask the book of knowledge Book of Knowledge what is the HCTP error code 415 stand for just I just like listening to The Book Of Knowledge and see what The Book Of Knowledge says According to the book of knowledge, HTTP error code 415 means unsupported media type. The server cannot process the media format of the requested data such as when a client sends JSON but the server only accepts XML. Thus it has been written." Well see and in a 415 usually shows up with a POST request but I'm getting 415s on GET requests

56:24 And so there's not even a media type that's included. There is no body, there is no body at all. So it's just a URL. So the whole thing is very strange and And John Spurlock was like, well maybe ACF. Maybe Cloudflare considers you to be a bot? You're no good! You're just the bot man. A malicious bot. I'm like well...I mean maybe but 415 that's that's a weird one so as I will okay why don't we do this if there's a chance so i did some checking and did some curling

57:05 And, uh... What user agent? PHP library? Various. Various user agents impersonating all kinds of things. So I did the same set of curls from my local machine and from one of the aggregator boxes at Linode. If I set my user agent to be a browser from a local machine, I got it 200 and not a 415. And this is to one specific... This happened on two different feeds. One of them was Castapod and one of them was a WordPress install so they're different platforms. So this is what led John to think well maybe this is Cloudflare giving you... The finger. Yeah, giving you static. And then I did the same set of tests

58:09 The Linode host and everything produced a 415 no matter what I did. And so, I was like well okay maybe it is Cloudflare? So I submitted the podcast index aggregator to their verified bots program heard nothing back for a couple of days and then found through some Google searches that people routinely submit to the verified bots program and don't hear anything back for quote weeks or months. Oh brother! So I was like well okay maybe let's try a different approach you can have their verified bots program has two different routes you can take, You can submit your user agent plus your known IP addresses as a combo

CHAPTER 20 / 30 Discussion

Implementing Web-Bot-Auth and Key Signing

To resolve blocking issues, Dave Jones implemented HTTP signing using Cloudflare's "web-bot-auth" protocol. He repurposed existing Rust code from the ActivityPub bridge to publish public keys on podcastindex.org, though he notes that Cloudflare has yet to officially verify the bot despite the implementation.

http signing· cloudflare· rust· public keys· activity pub

59:01 and saying, you know here's my bot. Here is what the user address string is going to be? Here are all IP addresses that the traffic is going to come from this is how you know it's me and in order to do that You have to have public descriptions of your bot Public IP lists all this kind so I did all that documentation got all that stuff submitted well that didn't work So The other option Is to do signing Mm-hmmm You can do a verified bot through What they call their webbotauth Program Auth Auth, yes. Got it. AUTH. Yeah. Which is just HTTP signing which is what we brought up in the GitHub. Yes exactly that! Yes got it. So I thought okay well let's see how this works so I will eat my own dog food since I support the idea of HTTP signing...I will harm myself for the benefit of the calls? Yes for the group

1:00:03 For the group. I will take one for the team, yes and i will do the signing stuff because i've already written signing code before that's in the podcast index copy paste it's just a copy paste another copy paste in your life Yes In the activity pub podcast index activity pub bridge since mastodon requires HTTP signing I already had all that code. So, I'm like okay this is pretty much the same thing but with a little bit of... with some wrinkles so i got the coding agent to go out and look at all the code for the podcast that I'd previously written because I'm like if this was the aggregator's written in Rust that other codes are written in rust should be just a boop you know right there, stick it in there. Right in there

1:01:00 It sucked it in, did the thing. Looked at the Cloudflare documentation and then made all the necessary changes so that we could publish the public keys in the correct format that then Cloudflare could verify against. So it did all of that work I had to go and stick the public keys into a .wellknown URL location on podcastindex.org website that is functioning now for key verification. So, now the aggregator is signing all of its feed download requests submitted it again to CloudFlare this time choosing the webbot-auth keysigning and since this all follows their proposal This is what they want people to do

1:02:01 and it's all public, and it should be immediately automatable. I thought this must go pretty good. This is probably going to go pretty quick. It's been about a week have not heard of Peep. I was worried about that! We also we still do not show up on the verified bots list for Cloudflare. Overcast is on there other podcast feeds are on... Other podcast aggregators are on there. Oh man Not us You're not in the app store? So, we... I don't know. But I do know that we are now signing all of our HTTP requests so all that code is net and of course the aggregator is public. Our website code is public anybody can go and grab it and start signing your traffic because now all that code has been published and you can do it

CHAPTER 21 / 30 Discussion

Browser-Based Signing and Future Abuse Prevention

Dave Jones plans to extend HTTP signing to media downloads on the Podcast Index website using service workers to overcome browser header limitations. He views this as a proactive measure to provide working code for the industry should abusive traffic become a critical issue in the future.

service worker· ajax· media downloads· open source· headers

1:03:04 The next thing I'm going to do in order to continue eating my own dog food is place HTTP signing for media downloads into the podcast index website. So if you play an enclosure from our website, that request will now be signed. Okay. The reason I'm doing this is because the most difficult Um, the most difficult way to do this is in a web browser because you have such limited options when it comes to signing. Excuse me? When it comes to adding headers too, to an Ajax request, the browser does not want to let you have access to add custom headers to those things. It's very limited but you can according to the research I've done, there's a couple of ways around it

1:04:12 One of them is through a service worker. And so that is my next project, to attempt to add support for HTTP signing for media enclosure requests to website hours in order to just prove that it works. I'll reiterate what I said last week. I'm not going to die on this hill. It's a thing that people seem to not want but if three years from now, abuse... abusive traffic becomes such a problem that this topic gets revisited. At that point there will be working code that's out there that we can all just- Oh I thought you were going to do an, I told you so! No no, I mean I may but that'll be a game time decision

CHAPTER 22 / 30 Discussion

AMP Group Success and Apple Play Metrics

The hosts discuss the success of the AMP group's mission to shift industry focus from 60-second downloads to 30-second plays. They cite Podnews data showing Apple "plays" significantly outnumbering "downloads," which benefits advertisers and publishers by increasing reported engagement metrics.

amp· apple podcasts· downloads· plays· james cridland

1:05:13 Well, I think in general the AMP group got what they wanted. They want more money so they found the perfect solution Let's take a play from 60 seconds to 30 seconds So you'll have more plays butter being everybody got more money overnight It's perfect it succeeded at its initial mission. I Think James had I think James was right that the play, I think he's right in his instinct that what this really was about was not about getting accurate download numbers. No. Or accurate listen numbers is to try to somehow judge the size of a show so you know whether advertisers are going to want to buy it or not Yeah So if you can get a play count... Yeah Oh well But listen, May numbers for Pod News

1:06:06 2,093 Apple downloads versus 3200 Apple plays. Boom! Money in the pocket. Number go up? A number go up that's all you need yeah That's not spoofable Yeah let them do whatever they want to do is fine Really Let them do whatever they want Call a podcast or whatever you want to call it You know thats fine its all good I tried I tried. I mean, I still think it makes sense to do something...I'm just not sure how to... No but it's... I'm depressed! Don't be depressed. It doesn't really matter. It's not going to affect anything what we're doing you know? Getting offers to sell the indexes did you know that?

CHAPTER 23 / 30 Discussion

Private Equity Interest in the Podcast Index

Dave Jones reveals he has received inquiries from private equity firms and family offices interested in buying the Podcast Index. He expresses confusion over why anyone would want to purchase an open-source project that generates no revenue, jokingly suggesting they should pay in Bitcoin.

private equity· podcast index· family office· investment· bitcoin

1:06:58 Why? I don't know. Who would want it? Yes, there's people who say they want to buy it but like this is not for sale. For what?! I don't know! I'm like do you need Dave with it because uh... without Dave its kind of nothing That's the weirdest thing I've heard in a long time. Why would somebody want to buy an open source project that doesn't make any money? Let me tell you, I don't think they exactly know what it is. I hope not. Let me see if I can get one of these emails. I think I can find one somewhere. Yeah, I hope they don't know what it is because otherwise they have bad decisions. What it is... It's private equity

1:07:45 That's even better. Isn't that great? Let me see, let me see if I can find this real quick. I may not be able to find it but you know private equity and like a home office type private equity and we really want...we really want to put some money into podcasting and so we're looking at everything that's out there it's like someone trying to do some shopping as some Spotify buying I guess Chad says sell it and launch Podcasting 3.0 We'll probably have a non-compete Why would anybody want this dog? Is what I wanna know

1:08:30 It's not a dog. It looks very... Hey, look at what we can go public! I mean come on. We can IPO. Just podcastindex.ai and we're billionaires that's all you just need to change the... What an oversight? I know let me see here it is. We have a family office is actively investing in companies in podcasting are you available to discuss Edith, VP of Growth Equity. I'm like yeah sure send me we want everything in Bitcoin please? Yeah but not right now wait till it keeps coming. Not at this very moment no actually we want it now

1:09:22 Now, all this equity there's going to be liquidity everywhere pretty soon. Once these once people get out of the open AI and anthropic you know the insiders once they get rid of it in my experience The buyer will give it back to you in about five years Well we had to have Nathan representing us because yeah You're an agent He knows how to do this stuff man Yeah he probably already knows that people We're no good Nathan knows how to do it Exactly, exactly. If y'all were boosting I'm sorry but i hadn't brought up my helipad and it looks like my start 9 has been down for a couple of days so I wonder if I even have channels open? Let me see... Yeah I need some kind of alert for this let me see though I still have channels oh good I've channels

CHAPTER 24 / 30 Discussion

IPFS Podcasting Node Status and Cameron's Wellness Check

Adam Curry notices his Start9 node has been down and realizes he hasn't received his usual alerts from IPFS Podcasting. The hosts joke about needing a "wellness check" for Cameron, who typically manages the notification system for node outages.

ipfs· start9· node· cameron· alerts

1:10:18 You didn't get an alert email from IPFS Podcasting? That's interesting you say that. No, no... Well because my IPF... Wow that's interesting Is Cameron alive? Has anybody checked the cabin in the woods? That's a good point Normally that was my canary in the coal mine when I got an ipfspodcasting.net email Hey your node has been down for 24 hours like oh crap! I gotta go reboot that thing Somebody do a wellness check on Cameron. He may not be okay Yeah, that's a good idea actually it's a good idea Let me see what else did I have on my list? I had The user agent war Okay podcast standards project where the teenagers they're the boring adults we got that Apple WWDC okay whoo video on everywhere okay

1:11:17 Yeah, everything I think we've kind of covered all the things you know People need to understand this is just a moment for Dave and Adam to catch up We used to do this just on the phone. We just decided to record it yeah That's right Yeah That that's how some of the funnest things have have developed did you have anything else? You wanted to bring up? I got a whole bunch of stuff in my in my notes here, but I'm really struggling to make any sense out of it. It looks like I just brain dumped onto the page. You need a system man! I don't know what I did. You need a system. The uh...I did...let's see...try...yeah oh yeah James talked about the other day about deprecating

CHAPTER 25 / 30 Discussion

Transcript Formats and the Robustness Principle

The hosts debate James Cridland's desire to deprecate non-VTT transcript formats like SRT. Dave Jones argues for the "Robustness Principle" (Postel's Law), suggesting that open-source systems should be liberal in what they accept and conservative in what they send to maintain network flexibility.

vtt· srt· james cridland· open source· robustness principle· john postel

1:12:14 Non VTT formats in transcript tag. Yes, he keeps one No, yeah anything but VTT yes He wants to kill it I'm still you doing SRT by the way Yeah And this is which which I could switch it But it's kind of like me versus the curry versus the criddler at this point I'm like I'm not gonna switch that go ahead Criddler try and ban me publishing my SRT files That's what I've been doing for a long time I don't think we can do that. No, deprecating is gay! Can't do that We can't deprecate formats because... Because James doesn't like it? Well no- I'm sure he has very valid points actually He's always got a good logical flow of his reasoning

1:13:20 But I think sometimes he misses just the fact that this, that open source just really can't dictate what people do. Because at Apple says we only take VTT but other apps take other things and other people create different things. Oh, Apple doesn't recognize my SRT? No. Oh! Well, I didn't know that I have so many pipelines written for SRT, it's kind of... It's an easy swap. Yeah, I guess. Your bot can do it, I'm sure. Yeah, yeah. But this—I've talked about maybe— I think I've talked about this before but like one of the problems with namespaces in the past and that it annoys him when we have very open-ended tags that don't narrowly define a format. It is a form of OCD

1:14:17 Yeah, well it's like you should have one format that you support. But I am... maybe the namespace is this way because it's an expression of a belief pattern within open source This it's it goes back to that. I was at Bill Joy This said be be liberal and what you accept and conservative on what you send That's the either that or Dave Weiner, I don't know one of the two no when Dave Weiner It wasn't it wasn't a wine now if it predates him But it was I think it was Bill joy but that that ethos of Be very be very open with the formats that you accept

1:15:14 but very strict with the formats that you publish, allows for a loose coupling that is really important. So like the internet itself works... it's built on this foundation because if you publish the robustness principle okay yeah so John Postle, that's right is John Postle, the guy that invented SMTP If you, so the internet itself is built on this idea that you like when you're sending packets and they travel across a router. Routers are expected if they cannot deliver it in a quick timely manner they are expected to drop packets. It's not an error. Dropping a packet is not an error

1:16:19 It is the way that the network is designed. You drop if you, under a certain set of circumstances it is the correct thing to do to drop a packet because then the sending entity knows they don't get an acknowledgement so they just resend it. It's not a problem. That is sort of the ethos of open source in general that when you are very accepting of all kinds of things being sent to you, and you just do the best you can to handle it. Because when you try to dictate strictness to a sending party there is... It only goes so far. You get to a certain percentage and you'll never get beyond it. So then the onus gets flipped back on you as the consumer of that format or that data

CHAPTER 26 / 30 Discussion

WordPress Enclosures and US-Centric Podcasting

Jones uses a WordPress blog example to illustrate why podcasting should remain open to various transcript formats, including simple HTML links. They also briefly address James Cridland's criticism of podcasting being too US-centric, listing famous Australian actors as a humorous rebuttal.

wordpress· rss· html· transcripts· australia

1:17:19 to then make adjustments. And this is the way RSS works too, I mean... Yeah absolutely! Yeah aggregators make all kinds of hoops to deal with non-standard XML stuff We're used to it that's just the way this world works so otherwise it wouldn't work you are absolutely right thats a perfect example Yeah, like what about somebody who publishes a blog? Let's say they got a WordPress site. And their WordPress site only... They're reading their blog and they're gonna turn that into a podcast so they're just gonna record their voice reading their blog post

1:18:07 and make that a podcast. They upload it as an MP3, they're using WordPress built-in enclosure support, and then they download a plugin that will modify their RSS feed to also stick... Enclosure? A link to their HTML page in as the transcript because all they're doing is reading the page Right! So the transcript is just a link back to their HTML page. Well, this person's not an expert in podcasting formats? No. Do we want to say well you have to publish that in VTT? You're banished! You're banished from podcasting! Your HTML is no good here I'm Adam Curry the Podfather. You're no good go away

1:19:02 So that's... and I think the answer to that, from my perspective is no. We need to be accepting of other formats And other children in other cultures from other lands Yeah don't be so Australian-centric Be open minded Oh you just hit a hot button there He hates that podcasting is so US centric He doesn't like that. Which it is, of course! It is... Now you guys got Mel Gibson be happy. Olivia Newton-John. Hugh Jackman. They've got Hugh Jackman too. You've got Kidman, Nicole Kidman. Chris...

CHAPTER 27 / 30 Discussion

HVAC Failures and Mini Split System Costs

Dave Jones describes a major air conditioning failure at his home involving a melted main board in a seven-year-old mini split system. Because components are often proprietary and married to specific models, he faces a potential $16,000 replacement cost for a new central system.

hvac· mini split· capacitor· air conditioning· repair costs

1:19:52 Chris, the guy that played Thor. You got Thor! What's your problem? That guy, you got Thor. Hemsworth, Chris Hemsworth. All kinds of people. They've got all kinds of people. Did you say Olivia Newton-John? I did. I did. I GOT THREE OF THEM ON THE PLANE! Yep. Xanadu. Yeah baby yeah yep well I wish i could, hope i put you in a little bit of better mood. I don't like it when your down. Not really. No? Didn't help? No. You gotta go camping again man! Go back into the woods. It's good for you Well our air conditioner has been broke for about two months right now Oh... And this is so you dont have air conditioner on breaker guys We have

1:20:44 We had two sort of mini split systems, like toward two zones down here on the bottom floor of our house. The back one is broke. The one at the front of the house works. It's the capacitor I tell you it's always the capacitor I walked out one day and then, and I could smell a horrible smell. That's the capacitor burning! No it was the main board. Ooh no that's not good. It melted itself. Yeah that's an automatic $3200 repair So it was going to be more than that because for people who may be not familiar with mini splits The mini split world is

1:21:32 The compressor and the internal air handlers, they're sort of married together. You can't mix and match one is sort of tied to the other. You got to replace both? You gotta replace them all so ours were like seven years old. That's not very old It's not very old but new mini splits come out on a yearly basis So ours, we had no hope of finding any parks for our old one. And so they would have to put in a whole new system that was going to be like 10 grand and I was like well for that we can just replace the whole downstairs with the whole new central system and it's gonna cost

1:22:16 A lot more than 10 grand. No, but oh yeah it's gonna be like $16,000 Well why don't you take ten thousand dollars podcast index money and we'll call it a day and then you add the other six I'm fine with that I don't need any money Never no no no Everybody will agree with me You need air conditioning This is critical Look listen to your attitude today This is critical to the future of podcast index. I'm telling you! The air conditioner? Yes, yes yes You need to have cool air. We need keep your blood at the right temperature bro. Yes, yes No this is not okay Just know that I approve it. That's...I'll just wait for the private equity money Oh

CHAPTER 28 / 30 Discussion

The HVAC Start Capacitor Scam

The hosts discuss the common practice of HVAC companies overcharging for start capacitors, which are $15-$35 wholesale parts. Jones recounts being charged $700 for a replacement he could have done himself, comparing the industry's "cash cow" tactics to those of auto mechanic shops.

hvac· capacitor· repair· wholesale· service fee

1:23:02 Yeah, we want a way for the eggs. We need 1 million sixteen thousand dollars for the podcast index No, I'm gonna wait for for the IPO wait for the IPO yeah I'm sorry brother that sucks really does Living in an old living in an old house it costed do a Arm and arm and a leg living in a new house is the same problem But but I've had my air units. I mean, it's always the capacitor and that's usually what it is. That capacitor thing is a total scam too? Oh, is it? Oh man! So... Really?! It is one- so when you take the main way mechanic shops make money is by with oil changes and brakes Yeah sure They'll put they'll take $50 set of brake pads and charge you three hundred fifty dollars to do them

1:23:58 The way that one of the ways that HVAC companies, one of their cash cows is start capacitors. Yeah it's a $35 part and they will charge you seven hundred dollars to replace it No no my guys do not charge me 700 at all How much how much they charge? I'd have to look, but we have a service subscription so we don't pay anything extra for them coming out. I think it's probably 70. I think they're probably charging $70. But now I understand because I'm the guy that is like hey here's 20 bucks thanks for coming out man and they always looked at me like you don't have to do that bro! They are probably feeling bad about themselves at this point Like oh man he is tipping me just ripped him off I can't believe he was doing that Don't feel good about this He's so nice My dad used to work for an HVAC supply company

1:24:46 And so we could always get, I could always get the parts wholesale. So I can get a start capacitor for like $15 and I just plop that thing in myself and then one day I got real busy and I was like you know, I just don't have time to handle this. I'm going to call the company. I was expecting them to charge me $200 for the service fee and all that kind of stuff when they come out. $700 is what they charged for a star capacitor And for just replacing it. It was simply a wire, I was like nope! No they can't do that in Texas man not in the hill country you can't get away with that people will shoot you no. I literally got this text from we have uh is it called aerobic septic system? Are you familiar with this

CHAPTER 29 / 30 Discussion

Texas Property Security and Aerobic Septic Systems

Adam Curry shares a humorous text exchange with his septic system technician regarding "trip mines" and "booby traps" on his Texas property. Dave Jones counters with a story about his HVAC technician in Alabama, who bonded with him over vintage Ford trucks and spare parts.

septic system· texas· booby traps· ford trucks· alabama

1:25:35 Oh yeah, that's where you got bacteria that eat your poop. No! Nice visual but no that's not it. The aerobic... Aerobic? I think it is. That's probably the wrong word So basically you have your own little mini water purifier so there's a little pump there CO2 pump and it just shoots in co2 at high speed and So after a while, then it just... Then sprinklers pop up and it's just shoots this clean water all over the grass. So it's really purifying your poop into stuff that you can, you know, just water. So the guy, Paul comes out. Let me see if I can find this. Oops! What is this? That's not what I wanted. Paul comes out and oh man let me see if I can find his text message

1:26:35 Hold on. And he texts me before, here we go... He says please have all trip mines and explosives secured and disengaged we will be on your property shortly See that's how we talk in Texas Whoa! Are you serious? Yeah Yeah. Holy crap! I said, it's alright Paul all booby traps are disarmed Phoebe is indoors it's good you're good to go Did he think he was going to Laura Logan's house? No no I think they just keep their poop in the ground I don't think they'd do anything with it The guy that uh... the guy thats doing our HVAC though is really a cool dude we used to live about an hour north of Birmingham

1:27:20 in the wilderness and he's from that same area. And so, he rolled up the other day and he saw the old 72 Ford 4x4 in the backyard. Yeah yeah yeah. He was like hey man I got four trucks too! You want some mirrors? I got a set of mirrors I'll give you I was like, yeah, I'll take some mirrors. You never know when my wife will rip one off! And he's like no no it's just one Ford guy to another gotta give you some mirrors. Need a tranny? I got a tranny here for ya! Something different in Austin that doesn't Alabama. Alright brother well what are you gonna do this weekend

CHAPTER 30 / 30 Discussion

Show Wrap-Up and Final Thoughts on the AI Bubble

The hosts conclude the episode by discussing their weekend plans and the heat in Alabama. They reiterate their belief that the AI bubble has popped before signing off and directing listeners to the Podcast Index website.

podcasting 2.0· podcast index· ai bubble· board meeting

1:28:06 Well, I was going to go climbing yesterday. Yeah! I mean tomorrow... Tomorrow but it's gonna be like 92 degrees outside so Oh no it's better to sit indoors without air conditioning enjoy that that's gonna be much better No just sit in the front of the house where there is still air oh man yeah There's air in the front of the house we just don't stay in the kitchen that's what God is like a thousand degrees you got fans? You've got fans going? Yeah All right, well there it is boardroom. We have solved nothing today, but we had a nice chat I hope you all enjoyed it and as always Would you please join us next Friday for another episode of the board meeting podcasting 2.0?

1:29:01 You have been listening to podcasting 2.0 visit podcast index org for more information The bubble has popped