Episode 257: Slop Factory

0:00 Podcasting 2.0 for April 17th. Two-fif- Welcome to Podcasting 2.0. We're back after one Friday off. to give you all things podcast. The real podcasting. All things RSS. board meeting of Podcasting 2.0 and the only boardroom that wasn't a sneaker company previously. I'm Adam Curry here in the heart of the Texas Hill Country and in Alabama, the man who will never Say hello to my friend on the other end, the one, the only, Mr. Dave!

0:46 Jones! Don't say never. Anything's possible. Don't tempt me. This is true. This is true. This is true. How you doing, brother? Good. Yeah. Yeah. Doing good. I'm Off work today. What? Well, of course, we already had tax season. Yeah, tax season has come and gone. So you get three weeks of, what was it, two weeks of rest, and then it starts all over again? Two weeks may be generous because all the things you can't do during tax season you have to do as IT. You have to do after tax season. You can't make no changes during tax season. How was this tax season? Were people happy? Did they get bigger refunds because of no tax on nothing?

1:34 No tax on that. People did get bigger refunds. Okay. It was the headlines I saw were humorous, though. It was like. Yes, tax refund sizes are up, but not by as much as previously thought. No, it's maybe 8% or 10%, I think. Like 11%. That's pretty good. That is pretty good. Yeah, that's not bad. I would like 11% more money. I'll take it. I'll take it. Yeah. I mean, like talk about making a negative out of a positive. Oh, well, typical. I was listening to. First ring day. So what? What show? First Ring Daily, it's a very...

2:18 in like a niche. podcast about It's daily about. Microsoft stuff. It's only last like. Oh, okay. It's just, it comes up. eat every day. Another one of those things I listen to for work, but Paul Theriot was talking about. The insanity, just the craziness of headlines now. And like one of the ones he was quoting was. this headline that said um these seven things on mars will kill you instantly And the subtitle was, Number 5 will truly blow your mind. Oh, that sounds like AI slop right there. Oh, it totally is.

3:00 There's seven things on Mars that could kill you. I mean, like the number – just the fact that they're – like number one is all – it's the whole thing. Like just the lack of oxygen. Well, yeah. Everything else is just a stupid gravy on top. I mean – The world is insane. Well, the world is – the internet is breaking. The internet is – It is, and I'm so happy about it. Oh, well, let's start there. Because there's a couple different angles. I know you've been fighting all kinds of things. I'll save the CICD prompt injection for a little bit later on, which I find very interesting, actually. I love – like I don't know how many things can happen in a single week, but this week may have been a record.

3:46 Well, tell me about the Claude Code Gemini CLI credential theft because I is a developer now. until everyone got the 500. from the API and then suddenly I wasn't a developer anymore, which was kind of fun. What is happening? Is that because people are putting their credentials into Claude and it's getting sucked in there and spitting back or what's happening? You're catching me off guard. I'm not sure I know what you're talking about. Exposing GitHub credentials through PR titles. Modern CICD pipelines are a train wreck, says Dave, post Dave on podcast index. My bad. Yeah, you got it.

4:31 Yeah, so... Pull requests. Yeah, the pull. This is a continuation. GitHub actions are having a bad day. What are GitHub actions? They're having a bad month, actually. What are GitHub actions? So GitHub Actions are a way that you can trigger. for non-developers, non-SRE sysadmin people. There's a thing modern code is distributed to a production system. typically based on... call a CI-CD pipeline, which is continuous integration, continuous deployment. Right. A typical flow would look something like this. It would be...

5:17 there's a development server or a staging area where things are tested, then... Testing's for pussies. Live, baby. We'll do it live. Oh, you're talking to the king of modifying the production server in flight. Yes, I agree. We'll leave you with a – I can't do it. We'll do it live. We'll do it live! F*** it! That f***ing thing sucks! That's Dave. Dave in the dev department. Yes. I'm just less angry about it.

6:00 um, Yes, the... So what happens is you have this staging environment or this development environment and then things are tested and they go through, tested by people and by automation. The changes are considered valid and good, and then you promote it to a release. So then you would typically do some. it with a version number. this would trigger a release or sometimes the release trigger is manual and then that goes once that happens this whole it kicks off a whole slew of downstream functions that take over.

6:47 So once the release is finalized. All this, this whole, the whole process of getting it over to your production servers that are actually serving traffic to the world. It's almost out of it. Practically out of everyone's hands. It just happens. Mm-hmm. This is. Kicked off by GitHub actions. And so these GitHub actions can. you know, uh, build the project, do things like build a Docker image off the project. And yeah, then perhaps do a security scan to make sure no malicious packages were included as part of the build because the build is pulling from package managers. So this is all coming from a large language model triggering this through an API on GitHub?

7:36 No, it's not large language model. This predates LLMs by a long shot. These are all just, it's code flow. It's DevOps. And it's a good idea. It's considered, you know, you could say like infrastructure as code. Where you're defining a set of things that need to happen every single day. single time. And instead of a human being running, just sitting there running these scripts. The automation takes over and It does, and it runs tests at each step of the way to see if anything fails before it deploys. You know, the idea is it can catch problems before it gets to production.

8:16 So, I mean, the ideas involved are not bad, but. course there's going to be risks because the risks of automation are always there. What if somebody gets in – what if somebody figures out a way to get into your pipeline? And the more steps you have in that chain, the more opportunities there are for something to have. something malicious injected into it. And this is what happened. earlier Let's see, like earlier in March, so the first half of March, the Axios breach.

8:57 Right. Yeah. That's what that came from. LLM breach, all of that stuff has started with, with. It wasn't GitHub Actions itself that was the problem. The developer accounts. of other Package repositories got breached. And then they were able to update the things that were called in the GitHub actions. those injected the malware and so or the info stealers and that sort of thing so um the the biggest one was Smarks, I think is what it's called. It's a security. It's a GitHub action security scan.

9:41 And so the idea here is that you build your production. build your release bundle. then it goes through and looks for vulnerabilities in what it just built. And it's, so it's a security scanner, like an antivirus scanner. And. But somebody compromised that. And so... Every time that thing ran. It was deploying an InfoStealer into the code that you just built. That took place across. dozens of huge major repositories. I mean, we still don't know the extent of that. I think if I'm not mistaken, that's what took down Cisco's. I don't know if you heard about that, but Cisco's entire GitHub private GitHub repository got stolen.

10:31 All of Cisco's proprietary code is now leaked. It's been hacked. That's it. came from that breach. So this is just a continuation of the same thing of attacking GitHub Actions. And this thing figured out that. There was a way that you could. Use the titles of – pull requests. to execute code. So you could inject a malicious pull request. And just the presence of that pull request would trigger the action. So like somebody, for instance, our web UI repository for podcast index. It's a great it's a good example of this. When somebody submits a pull request.

11:24 to our website repo. A bunch of automatic checks happen to make sure that because the primary use case for that is people adding. themselves to the podcastapps.com page. And so... That's all based on a JSON file. And the GitHub action that triggers when somebody submits a pull request on that JSON file is it checks the syntax, makes sure that they did. everything right. end up pushing to production a broken JSON. So now, so this means that.

12:05 this particular vulnerability. could have made somebody deploy a malicious pull request to your repo. The automatic checks happen in certain circumstances. If they put code in the title, that code would execute. could scrape your data and send back. that they're not supposed to have. It's a disaster. This whole thing is so fragile. We don't... We built systems over the past decade. that are not ready. for the way for how fast things are.

12:48 Deployed. Deployed and yeah, because like. It's that famous phrase like from Jurassic Park, you know, you. All you thought about is whether you could. You didn't stop to think about whether you should. Yeah. And that's what's happening really here is these package managers or package repositories. Like now every time I build something, I just hold my breath. I'm not kidding. Like I was building something this morning. I was doing a – I was doing a – some work on the local aggregator for Godcaster this morning.

13:29 I needed a composer package. Put into. because that runs Laravel. So it's all handled with Composer, which is PMP. PHP. I did composer, you know, install. I just had this anxiety the whole time watching the screen. I'm like, what's about to take over my computer? This is code coming from some... In that sense... Honestly. What we've been – we complain about AI creating code that we don't read. or that we don't understand or we never look at? Oh, I mean, that's, if you think about package repositories.

14:14 That's been going on for 15 years. We have no idea. What is in most of these? Just all these apps install. Boom. Well, you know that. The apt-get The Debian repositories, because it – I feel more comfortable that those have been – vetted more thoroughly and they're also older packages you don't typically like especially on an lts release what you're getting is typically a package that's been frozen at a version that's actually pretty old Comparatively speaking, so those Linux repo, Linux distro package repositories are actually Those do not bother me.

15:03 much i would much rather get one of those than something off like npm or p or pip I mean, or PiPi. Those things have, oh man. I mean, like those things are a train wreck, right? So I've been thinking about all of these and you'll tell me which one you want to talk about. But all of these different clearly vibe coded systems that have been been brought up to create podcasts. And, you know, it's like the big topic. The only thing it's bad is just it's just it's flooding systems. You know, podcasting doesn't have a discovery mechanism. You know, I really.

15:46 I do have some thoughts on this cloning of titles or close to cloning on titles, close to cloning on images. I have some thoughts on that. But in general, all of these, you know, so, okay, if someone happens to be searching. like, oh, I have my podcast app. I'm going to search for a podcast because I don't have anything to listen to. I'll bet you that that's a very low. amount of people that are doing that. positive. Yeah. I mean, so if they're searching for a particular podcast, yeah, that of course is a problem. If there's a whole bunch of clone feeds and things that look like it sounds like. But I don't think anyone's really, I mean, this is a business that is not going to last very long.

16:32 You know the the true way to stop it is to just stop giving away free accounts Or or you know free account, but not if AI or whatever Um, Uh... If anything, it's going to ruin podcast monetization, download-based monetization. Because advertised like, well, it's all scam. I don't think anyone's listening to this stuff. So I. I view that as generally good, but not good for people like Dave. It's because I was trying to update. We get to do this, the podcast. And I'm like, it's not happening. Why is it not happening? And then I waited for an hour, 45 minutes, an hour. Like, this is ridiculous. I don't know what's happened.

17:18 Of course, I thought it was my fault, but it's an RSS.com feed. So those are usually good. I check it. The feed is there. I check the raw feed. Feed's good. So then I wind up issuing a reset, which I don't like doing because I know that it takes a lot of extra resources. And that did get through and that did update the feed, not in the way it should have. And then you come back later and say, well, this is the reason, which is crazy process that was, was it searching fountain feeds? What exactly was going on there? I don't – I can't give you a step-by-step of exactly – so what I – you mentioned something Wednesday about we get to do this, not updating.

17:59 This being very or being stuck scanning. Yeah, that's why. Well, I couldn't tell if it was stuck. It just it said scanning and it wasn't doing it. So. And so later, a little bit later, I looked at it and I was like... By that time, it had gone through. I don't... really see anything. That was probably after my reset though. I don't. don't know to see this is all very hard to understand because what so i look of course the first thing i do usually in a situation like that is i go look at cloudflare and see if we're just getting you know And I didn't see anything that looked out of the ordinary based on – what I mean is out of the ordinary based on what I have seen lately.

18:41 I mean, we're serving a lot. I wish we had kept I wish I had kept historical traffic charts. because What we're serving lately is a lot. What's a lot, Dave? Let's see. Thank you. analytic. Here we go. I'm just looking at our Cloudflare stats. So over the last 24 hours, we've served.

19:22 322 gigs of 9.2 million API requests. Let's see over the last seven days. This is down significantly. from what it was yesterday. You can see on the chart it just – when Oscar fixed their issue, it just – it fell off. Over the last seven days, we've served two terabytes of data and 75 million API requests. I How many podcasts? are there. Well, most of this is just a handful. I can promise you. Yeah.

20:07 But they – so – The that's what I mean, that's so up from what it used to be, let's just say a year and a half ago. In one sense, I'm really happy that we hold up so well. Stop the hammering! I mean, I'm really, it makes me. I'm glad that we're able to handle so much traffic. I mean, that's pretty. That's amazing. That's pretty awesome. Yes. Yeah. Considering the shoestring budget we're on. For sure. Yeah. And. But then at the same time, it happened so gradually. It's like the boiling frog thing. I never really thought, well, maybe some of this stuff coming from these other places is not organic.

20:57 So, like, fountains... Traffic had been going up steadily for a while, and I had noticed I had noted that in the back of my mind. And I just figured. serving lots of data. And I'm like, okay, that's fine. Then what happened was I went. Oscar was like, hey, you know, this... Podcast is not updating. I looked and I was like, I wonder how many. I wonder how many things are in the queue for the aggregators to poll. I wonder what the queue length is for weight. I go and I look. That's where you just discovered it. 500,000 podcasts. Waiting for what? To be updated?

21:40 to be pulled, to be aggr- - Oh my goodness. be requested and downloaded for aggregation. Whoa. I mean, we. I mean we pull something like – something roughly around 2,500 podcasts a minute. That should not – the Q should never be that big. Right. It should not even – I mean the most I ever really see it get up to is about 1,600, 1,500, and that's just when we get a flood of things all in, and that usually clears out. minutes and then it's back to its steady state. This was off the chart. And so I was like, whoa, what in the world? So I go into Cloudflare and I start filtering by what is hitting.

22:28 Like, where's this traffic coming from? So looking at all the paths that are getting hit and the one one I saw up near the top was the. Hub notify API endpoint. Hmm. Which originally, so this goes back a ways. This is the. Mastodon stuff No, so... If this goes back to the very beginning of the index. 2020. This endpoint was created to support WebSub. Right. Okay. Yes. PubSub. WebSub. Yeah. PubSub. Yes, exactly. That's why it's called PubNotify. So it's notified. This is a mechanism for us to subscribe to WebSub servers.

23:17 for those web sub servers to then ping us when a feed updates. that Pub Notify has to be an unauthenticated API. just by default. And so. That is for web sub, but over the years we've said, okay, well, you can use it for other things if you just want to ping us and let us know that, hey, a feed changed and we need... Just send your request to here and you can either use the podcast index feed ID or you can use the podcast index feed URL or the podcast feed URL. Either one of those will work and we'll just stick it in the queue and pull it.

24:01 And that thing should never be at the top of any list of traffic. And what it was, was. Fountain was hitting us, was hitting that endpoint with feed IDs. At the rate of like. millions. every 24 hours. And so this you know, Oscar went in and I started checking his system. He was like, oh. He's like a bot is just killing us. And so he was getting slammed with some sort of automated bot crap. Someone trying to scrape, scrape podcasts through him, which indirectly came to us. Yeah, exactly. It's my my I don't know the way their system is set up, but.

24:52 I can imagine designing a system such that. If. certain sequence of events happens on It gets requested or a certain number of things happen and it just triggers – it makes your app think maybe there's a new episode. And so then it just sends a request to us to – for us to check. And normally I'm cool. I'm totally fine with that. Like if somebody has a reason to think that there is a new episode that we need to go get, hey, ping us. Let us know. But this was just somebody abusing their system, and it had this downstream effect.

25:34 So anyway, so I blocked that traffic and then he fixed, you know, he blocked it on his side and then we got it. other in this. through and just cleared all those backlogs because I'm sure. thousand of those requests weren't even real. And so then I cleared all that out and then just kind of sort of reset everything back to zero. everything's working again. There's. This is the sort of stuff that... that you have to deal with. And it's been every single day. And you have a wife who has needs.

26:15 She does. Funny enough, she does. Yeah, this this is what I was afraid of last year. I was afraid that this year was going to be the beginning of this. It has not – it is panned out exactly. It is not disappointed is what you're going to say. the way that I thought it would. This is the year of slop. Just abusive bots. see What I've had to – there's just been so much that's happened this week, man. Like I had to – finally just started working on the I'm like, uh.

26:55 classifier stuff. Yeah, I saw that. Yeah, the LLM is classifying it as spam? Yeah. And I'm not happy with the way that's working, but it is, it is working. thrilled with it and I'm going to be making this is going to be a long process. Well, it's hard. I mean, we used to do that on heuristics, and now it's on who knows. I mean, I have a... I have a Quen model running. that ingests the result of RSS feeds that I have set up. And so I probably get about, each run gets about 350 stories. And the Quen model then has to determine if the stories are worth even looking at. And that's hard. It's hard to tweak that just the right way.

27:49 Oh, oh yeah. I don't know if you want to go ahead and dig into this. I mean, I do want to talk about the spam classifier stuff, but maybe you have something else you want to hit. No. Well, actually. kind of flows into uh what james and sam were talking about for the first half hour of power this power I haven't listened yet, so I don't know what they're saying. This Light Knot Studios, who copied more than 75 podcasts and created AI slop versions that are obviously different audio, different knot copies of.

28:29 the podcast themselves, but you know, uh, history podcast names, you know, all the, uh, very close to the original, uh, successful podcast names. So the only thing I wanted to say, because I heard. You know, it's. It's not nice, first of all, but I have some experience in this that I just wanted to share because I don't know if James has mentioned it, but I kind of want to turn that upside down for anyone who's dealing with this. Well, actually, two experiences. One, a lot of people. Start naming their podcast No Agenda.

29:10 And I'd say most not maliciously. He's like, yeah, this is a great idea. We'll call our podcast. No agenda. We'll talk about sports. Well. I mean, if you are serious about what you're doing and you have a podcast, you need to trademark the name. You just need to trademark the name. It's not. Y'all do, right? Oh, yeah. Because now, I mean, every single time this happens, we have to send a DMCA takedown request. We just did one yesterday on Spotify. Spotify is reasonably quick to respond, but we have had to get lawyers involved more than 10 times at least, maybe 20 times. Because people are like, well, I don't want to do that. And, you know, I've done it to mainstream. Even my neighbor. She's like, no agenda with Laura Logan. Like, no.

30:00 No, you can't do that. Oh, yeah. That was a you sent that to Fox News, right? Yeah. That's when she was on Fox. And then all of a sudden it popped up again in our podcast. Anyway, that's one thing. But I have a lot of experience with something called the Lanham Act. And if you, as long as you have your podcast trademarked, and so that would be the name, that's most important. You can also do some trademarking on imaging, but it depends. You have to have kind of the same image. But I have experience with this from the 1994 MTV Networks versus Curry lawsuit, which... was filed by MTV, because I had MTV.com, which they gave me permission to use. They didn't care. They didn't care about the internet. We have the AOL keyword. Curry, do whatever you want. That was literally what they said. We have the AOL keyword.

30:53 interested in this domain name stuff. So I had the domain name and I used at MTV.com. personal site, but it was definitely MTV oriented. So the lawsuit that they filed was under the Lanham Act, which was correct, and they were right. So we settled because, you know, it turns out they had said you can go ahead and do this and they hadn't covered their their backsides. And honestly, if they just said, hey, can we have the domain name back? I would have given it to him. So that turned into them suing me under the Lanham Act. I countersued. We settled out of court. Neither party has any further comment. I'm still smiling. But the Lanham Act is there specifically. In fact, I'll give you the little paragraph about it. Federal trademark law protects brands from consumer confusion. Cortez, would a reasonable consumer be confused about who made, sponsored or endorsed a product?

31:47 Courts look at how similar the marks are, how related the goods are, evidence of actual confusion. And it covers domain squatting. That actually came from my, you know, it was jurisprudence, my lawsuit with MTV. brand impersonation, cloned content designed to look like the real thing. So the Lanham Act is the way to go. Forget about DMCA and all this stuff and going to the hosting company. You have to go to a lawyer and you have to threaten them with Lanham Act litigation. And that is the end of it. If not, then you have to follow through.

32:28 So I just want everyone to know that there is in the United States, at least I'm not sure about outside the United States, but the Lanham Act is the way to go. And any of these podcasts clearly can create confusion. It is obvious to prove that. one legal letter should make them go away. If they don't, then you'll have to take it a step further, but you will win that. And unfortunately, if you're in a business, and I've defended all kinds of copyright cases, including the first Creative Commons. copyright case. I was the first one to fight that in court. I won. I got zero money, but I won. Wasn't really the point. Yeah. So the Lanham Act is the way to go for anyone who's dealing with this. And we just have to do it. The thing that.

33:15 that, uh, that I'm worried about. And. I know that Spreaker has a lot of this going on. And so when you combine creating Lanham Act offenses, which a lot of people just don't know how to defend, won't defend, shrug their shoulders, whatever, combined with bots submitting hundreds of thousands of podcasts, the only thing I'm a little worried about. is that at some point a company like Spotify will say, well, that's been ruined. You should just come to us and we'll take care of you. And we're safe here because you can do it with our API and trusted partners and you're good to go.

33:58 That's the thing that I'm worried about. And maybe I'm giving them an idea, but it's also possible they're thinking about this themselves. I'm not clear on what you mean. If the RSS infrastructure becomes so polluted with similar names and similar topics and, you know, and we just can't fight the spam or the slop or. We don't have a way to help people. find what they're looking for when they're looking for it in that particular case. I just I don't want. companies to go, well, you know, like we're the guys, you know, come to us, we'll protect you. And it just moves away from RSS. That's what I'm talking about.

34:42 The only – I can see the logic of how you're getting there. The only thing I would say that may... is I think that the closed platforms are also fighting the exact same battle. I don't think they can make that guarantee. They could say it, but they couldn't live up to it. OK. Because like YouTube is is full. I mean, it's just a slop factory now. Yeah, it is. Spotify is a slop factory. They're all these. Yeah. All these these closed systems are battling the exact. problem. And. I want to be fair to – all of the hosting companies.

35:27 there It may seem like I'm giving Spreaker a hard time because they... They have just become such a factory for slop. Slop factory. They're a slop factory. They've become a slop factory. They're a slop slut. This. The problem is. that They are really between a rock and a hard place. This is a difficult thing. The light not. Studio thing I think that happened to RSS calm. Yeah the and this morning I saw I taught a whole slew of of just scam garbage With phone numbers to call and all the others a full-on scam. Yeah fishing campaign.

36:23 I saw this hitting multiple hosts, Buzzsprout, Podio. Podbean. It was like three different hosts. This The rock and a hard place that the hosting companies are in. You have to do a free – Even if you don't have a full, a true free tier, Black Spreaker and RSS.com and Red Circle. Even if you don't have that. you still at least It's table stakes now that you have to have. trial period. 14 days. 30 days, whatever it is. You just have to.

37:03 A lot of stuff can happen. a 14-day or 30-day trial. a whole lot of abuse. I know for a fact. There are people in the hosting companies I've taught. doing that are Every single day they spend, who knows, an hour or more? scraping through the trial accounts trying to get rid of this stuff. 404, these scam feeds. It's hard, man. It's hard. What I'm doing on my side is trying to make that job for them easier. I want it to be.

37:54 I want us. I want our visibility across hosts. a benefit to the hosts. Of course. Yes, of course. Because we can see something happening over here on Podbean. Before it hits buzzsprout. Exactly. Before it metastasizes. Well, I think, you know, we have hosting companies like RSS.com. I'm just going to be a douche about it now for a second. RSS.com, Buzzsprout, the boys and girls over there, Blueberry. who support the index. So we should be transistor. Yes, we should be supporting them.

38:38 The other ones, I don't know so much, Dave. I mean, I'm just saying. I understand the sentiment. I just – The evil, mean sentiment that cropped up in me? Yes, yes. That's what that was. Yeah, I did not – that's not what I said. That's what I said. That's what you said. I understand that sentiment, but also, like you said, the larger goal, though, is that we want to get some sort of traction on this. Most of this is coming from two motives. obvious one is Add.

39:23 Revenue. You throw up a slop cast, you get ad revenue. Slop cast. Yeah, from DAI. Yeah. You get that? That's one. That's the obvious one. The other one is SEO spam. Yeah. And this is just plain Jane. SEO web traffic scamming that has been with us forever. If I could make one recommendation. to the podcast hosting companies. be this. If you're on a free tier. you or you're in a trial account. Do not give those accounts any web presence whatsoever.

40:09 Right. And also and also mark them as such in the feed. Hmm. So that everybody knows to ignore them until they become an actual. paid. some sort of like a legit podcast getting You've verified that they are an actual. Human of some sort. Because I know some of the hosting companies try to block the SEO problem by just having their – Podcast web pages that they generate automatically by having them blocked from Google indexing but that doesn't stop it because Those podcasts feeds then flow into other places like Apple podcasts and podcast index They flow all over the place.

41:02 And then they'll pop up in other spots in the web, get indexed and turned into HTML. Thank you. It's not just the straight SEO from the page itself. It's like second-order SEO down the line. And what I'm seeing now, I mean, it's not new, but it's been going on for a long time. And I'm looking at this stuff for like... every single day for a while. And what I'm seeing. is trends. I'm beginning to see the trends myself. And the other SEO, it's not just scam SEOs like online gambling sites in Vietnam.

41:48 It's witches. deal and Asian gambling sites are a big thing. It's not just that. It's also – Is that? I'm starting – I'm seeing – Like local businesses. Oh wow. There was one, it was like PJ's heating and air service in Miami, Florida or something like that. And it was on like four different hosting companies. Well, you know what is big now with the Claude Code type? things is people are selling. In fact, one of our no agenda producers, not saying that he's doing this, but they'll sell you a marketing system in a box.

42:32 And so you log in, they got a little front end on it, and it's got some Claude code stuff underneath. Claude bot. And. So it'll build you a website and then it'll do a marketing campaign for you. And I think that's what we're seeing here is, oh, okay, I can get a marketing campaign for five bucks a month. All right, so I'll use this. And then it's going off and it's doing its SEO business. And it's doing it through stuff like that. I just dropped a search. in the boardroom is a search for You just search in podcast index for PJ slash PJ space HVAC. I see it.

43:18 That's what this is. Wow. One of them's on, let's see, that one's on Buzzsprout. Some of these have been 404'd and we just have We just haven't killed him yet. This one's on SoundCloud. This is all the same local business. Let's see where this one is. This one's another one on SoundCloud. Where's this one? This may be on speaker. There's three different ones on SoundCloud. But I saw this same one, PJHVAC. I know I saw it on Spreaker and at least one other place. So this was across four different hosting companies.

43:59 just a trial. It's abusing the trial account. Right. Well, I'm glad to see that SoundCloud finally has some traffic. PJ Mac, HVAC service and repair company in Pennsylvania. Wow. So, I mean, like there's clearly something that is. Yes. Somebody like you, like what you're talking about out there that is just sort of like SEO in a box. Mm hmm. Part of what we're going to do is we're going to throw you out on all the podcast platforms, and you're not really a podcast. We're just going to make one placeholder episode with – that just says – with our phone number and contact information. It's just – it's abuse.

44:40 It's – It's service abuse because it's not what. intent of these services are. Daniel, that PJ Mac HVAC show, that may not be an Apple podcast, but I can tell you for a fact that there are. Thousands. of shows. Of this of varying scam types in Apple podcast because I send the links to them. I find them in Apple Podcasts also. So this is not – it doesn't – it's not just a matter of the way we do things by scraping up everything.

45:24 This stuff is getting into Apple Podcasts also. And I can bring receipts because I've got them in my email where I've sent them to Ted in the past. So these people are good at – these scammers and SEO people – They're very good at this stuff. They're very good at it. And if you can get yourself onto Apple Podcasts, you've hit the jackpot. Well, a lot of that's automatic. The automatic is... you know, through a lot of the hosting companies. That is SEO gold.

46:05 Because now you're coming off of an Apple.com top-level donut. Right, right. So, I mean, like... What I'm going to have to do. Let's see. Let me go over. just sort of what is happening right now, what I've done this week to try to get some sort of on this. So. What's happening now is. There's a client process that runs on.

46:49 my local system here. because Obviously we're running on a budget. I need to do local inference for this. So. have a client processes running is looking at the recent slash new feeds. It pulls the most recent. shows added to the index in the last Yeah, it's very cool. It's a cool page. Yeah, it runs every one of them through the inference.

47:32 through the Gemmaform model. What are you running the Gemma 4 on? It's, uh... Llama CPP on a Mac Mini. Okay. And, uh... So it's running on – it runs through that, does a classification, and there's a big – there's this – script. has the notion of bad feeds. I'm just going to put them in two buckets. There's actually six. But let's just for simplicity's sake, it has the notion of this is a bad feed, this is a good feed. I'm just going to, let's not get into the meanings of those things.

48:13 And so then it runs like you – it has a local SQLite database that keeps all these things tagged. Here's an example of a bunch of feeds that we've done. to nuke here's a bunch of examples of fees that are legit It just makes. It sends it to the LLM, makes an inference with a jag. prompt. says yes, this is most likely spam or this is most likely legit. And so you get a, you get a. decision that's been made. calls the Report slash problematic.

48:55 point. to report that feed with a reason code and a... note about why it was marked as problematic hey is there a way that um here's the thought oh here's a fun thought Is there a way that I can add my Vulcan to the process? help you with that? What's a Vulcan? My Vulcan 3090 card, baby. Vulcan. The Vulcan. Yeah. Yeah, I don't know if you know in Birmingham we have this huge statue on top of Red Mountain. In the center of Birmingham, it's a statue of Vulcan. Roman god of... Oh yeah, you guys are all pagan cult worshippers over there. We are. Once a week we all go and bow to the... The Vulcan. The Vulcan, yeah. So we had this radio station in town called the Vulcan. The Vulcan.

49:54 Rocking, flame-throwing Vulcan, Alabama. think anybody can run this. script. I'll have to think about how that would work. But it's definitely – it doesn't require – any sort of It doesn't run on the podcast index structure, so you can run it in your house. Here's a thought. So here's how people could really help. It does take a lot of VRAM, though. Well. I've really gotten into the rented GPU stuff. Are you familiar with this? I've rented GPUs, yes. Vast.ai, RunPod, Together.ai is the one I've had the most success with. Grok, G-R-O-K, which has no dev tier, but they have a lot of free stuff right now.

50:44 Um, I do a lot with that, particularly Whisper. So I'll do a... Someone says, hey, this is a great podcast. You should listen to this. I'm like, no, I'm going to listen to it. So I send my bot out, my robot, and it grabs. podcast or the YouTube turns it into a an audio file, whispers it, and then summarizes, tell me what's in it, if there's any good clips, anything I should get from it. And that costs like pennies because it goes so fast. on this rent stuff. So if you had like and I think you can set up like almost like I don't know what the term is, but it's like a docker basically where you can have an API endpoint that's stored. So anybody if anybody could just access that. So you could you know, you could just have a whole bunch of machines firing with the same model, the same weights, the same parameters all the time to help.

51:41 in that process. So you distribute the cost and the actual load of the service. I mean, we could fight this stuff. Does that make sense what I'm saying? I think it, yeah, it does make sense for sure. Um, The only caveat I would throw in there is... this one and this is a Because I'm strongly of the opinion right now. to train Oh, and I don't mean I don't mean necessarily a full training. What I mean is like a fine tune. We're going to have to fine tune a model. OK. Because what we really need is not...

52:24 what this thing currently does. which is... It tries to – which is it – What it does is it tells the model, okay, here's a new podcast. Here's the title and the description and titles and descriptions of. most recent episodes and show links. It just gives us a ton of metadata about this new podcast. And it says, here's the same data from about a dozen picked at random podcasts. from the legit table and about a dozen from the spam table. It says, okay, now do your best to make a decision.

53:07 also taking into account prompt that we give it with a bunch of parameters about what going on. That just has context problems. What you really want, you don't want to have to provide all of that essentially in a context window. What you want is for the model to have that knowledge innately because then you can basically train it on the entire index. So this comes to the point of… Love. Me and James went back and forth a little bit on the podcastindex.social this week about this topic.

53:50 He wrote it up. The Crickler. Riddler. Criddled. You got Criddler. So. He wrote it up as a headline in podcast. about the spam issue and the 24 hour feed report and all that. Oh, which is fine. pinged me on podcastindex.social and said, well, you know, and it was making the point. He called it censorship. I looked. disagree with that fundamentally. I think. censorship is. censorship is when I say.

54:33 like what you said and therefore I'm going to block you. like the, the, the thing, your ideas and what you're espousing. This is. This is something completely different. care less about is the form, looking at the structure. and the structure is abusive. podcasting as a whole. It's not censorship. It's self-defense. But I think the larger point of what he's saying, I took some good things away from what he said. He was – because what he said was … One of the unique benefits of Podcast Index is that you have all the feeds.

55:20 And if you block, if you... block some from being ingested. in it. lessens the value of the index and what it is and what it represents. Fair point. And I think he's right. but we can tag it. And so there was a misunderstanding on his part just because the – it's not – well, some of this is not obvious. And that is. We have all those feeds. I'm not deleting them out of the index. The index really doesn't ever delete anything. Right. The only things we ever delete.

56:01 are actually, I mean, like actually remove from the database, like they're no longer even in there. mostly things that are not podcasts. Sometimes things get in there and it's just a web page. I was like, well, this is not even a pod. I mean, this is literally not even a podcast. Yeah, right. Yeah, right. Yeah, this is a text file or some crap. And so like those we remove and then there's just a handful of other things that we removed. very specific reasons that Honestly, I don't even remember. Yeah. But I'm talking about – I mean you could just – you could put that on – Mm-hmm.

56:44 All the rest of this stuff that is getting marked, it's just there's a flag in the database that's just this dead equals one or dead equals zero. If dead equals one, then it no longer gets served on the website or the API. It's just it's like it's invisible. Data is still there. And when I say data, I mean the episodes are deleted, but the full feed record with the title, the description, the feed URL, if it had an iTunes ID, the status code, the HTTP status code, how many episodes were in the feed, all of that data, that metadata about the feed, all of that still exists.

57:25 So wouldn't you want to – a thought. So when you want to have one or more flags, we already have flags like AI slop, spam, etc. have the flags so you can, when you're polling the API, you get whatever you want. If you want everything, then you say, please add these flags. And only if someone actually polls that particular. Podcast Enix ID, do we do a scan? That makes sense. Or was that not useful? Run that by me one more time. I'm not sure I followed 100%. So we have everything.

58:07 But instead of it being dead, if you want to query the entire index, including the spam and the slop, etc., query it and you can get the information, get the top level information about the podcast. But only if someone says, well, I really, really need this Abraham Lincoln AI generated history podcast. Only then do we scan that feed. So we keep the – That would be a challenge. Oh, okay. That would be a challenge because we just – it's just dead one or – Dead Zero. Right, but we're working on flags now, aren't we? Are we coming up with – I mean you're flagging something if I look at this report page.

58:50 Yeah, things are flagged as reasons for being marked as dead. Okay, right. that's in a separate table so that we'd have to check is I think we could do what you're saying, but I think it would be hard. It doesn't mean it's not. Doesn't mean we shouldn't do it, but it would be. Well, I'm just continuously thinking how can we all work together very simple to the gossip system. So we all can help flag things. We all can, you know. just make it useful for everybody for whatever your use case. So if you really want to have, if you really want to search across all of the slop and everything, you should be able to do that.

59:38 But that these distributed systems are alleviating the load on the ultimate endpoint from the index to make it easier or ultimately, hey, you know, you want to have an index with all the slop in it? Your gossip has it. Yeah, so I think James had a good – I forget if his idea or mine. I don't remember exactly, but. We came up with a good idea, which is. index right now The core part of the index from day one has been this downloadable SQLite database of all the feeds. Of everything, yeah. And so that, what I've always done is I've excluded the dead feeds out of there.

1:00:24 Just for size constraint and because want to distribute things that dead because sometimes they're marked dead for a reason. Right. Sometimes they're more dead because – It's theft. It's plagiarism, et cetera. Yeah, and the creator asked us. Yeah. take it down. And so we don't. I wanted to just... distribute that in a different form that goes against their wishes. And so. what I think My next step is going to be to create a second download. And it's going to continue to exclude all of those things that were marked dead in the past.

1:01:10 Because those things, there is. Well... I don't know. I'm going to have to think this through. I told James, I said, give me a few days. I'm going to have to think this out. cause. What we want is a second download that Everything in it. for just for the people And then. But what needs to be in there. Who really wants that? Who really needs that? I don't know. So it seems like a big lift for a very few amount of people who really want all that.

1:01:56 Well, let me let me tell you, I could tell you this. I want. And here's why. Well, let me finish that first off. What we would have to do is in – I think this is the way it would have to work. What we'd have to do is we'd have to export it, do a second export. It has everything in it. And it's going to have to have at least one extra column, probably a few. that details if it's dead, why. The reason would be out of the MF – out of the problematic table. Listing, okay, well, we considered it spam or abuse or slop or whatever. And here's the note on why it was.

1:02:41 marked as dead, excuse me. have that. But then these older feeds. We would probably – what we'd probably have to do is if – The HTTP status code of the last scan is... 200 or if it's a valid. HTTP access code or response code. We would probably have to somehow. still not export that. Because that's probably an – that is most certainly an indication that that – that we were asked to remove it. But then if it's marked dead for some other HTTP status code, like a 404 or something like that, then we could stick that in.

1:03:24 And say, OK, well, it's obvious why this was removed. This was removed because it's no longer validly served. So there's going to have to be some work here to make this function properly. But once we do have that second download, this is what we fine-tune the model on. So what we'll have to do, because when you fine tune a model, you have. You have to create a data set. dataset, it's a lot like training a machine learning model. What you have is a data set. It comes down to here's a bunch of words and then here's a decision.

1:04:09 is it this thing or that thing? And so what it's just like a whole, it's a lot, it's a series of funnels. This is a series of tubes that flow from a bunch of content into what amounts to a single decision. And so then you feed that into the fine tuning. Thank you. There's a bunch of these scripts out there. So we can fine tune a GGUF model or something like that on. on this data from this second download because it's going to have all the valid ones in there and all the spam junk. So what does it take to train or fine-tune a model? That's above my knowledge base.

1:04:59 But what is the application that does that? What is the system that does that? I think there's some – it's easier now than it used to be. I think there's a couple of tools out there. NVIDIA just launched one through Onsloth. unsloth studio or something like that that will help it's it makes it easier But you can still just do it with a bunch of like PyTorch Python scripts. I'll just... research to figure this out right now. As soon as I have this second download constructed, then I'll... headlong into that and try to fine tune a model.

1:05:41 And it's just got a vast. knowledge to pull from to make these decisions. So that's the, when I said that there was a kink to. idea of distributing this. That's what I'm. We'd have to somehow distribute the model. this point, how that works. I like the idea of providing, of being a clearinghouse and providing the information back to the hosting companies. And that's a great idea. Yeah. Yeah, I think, I mean... That's gonna help a lot. It's gonna help a lot. I hope so. This morning. We saw this morning started the black magic.

1:06:23 Black magic? Let me see if I can find one of these. Yeah. I mean, this spam feeds are – I mean, they're entertaining to watch sometimes. I just want a terminal where I can just watch the spam come by. Like, what do you got? Oh, this is interesting. What do you got here? Oh, OnlyFans. Let me see if I can find one of these. It might be a little. What? Oh, is there? What is that one? It's like, I think I posted one on podcast index on social. It was like – it's stuff like how to get your girlfriend back.

1:07:05 And you do it with black magic, and there's a phone number you can call to get them to help you. Oh. Oh. Oh, it's rad, man. It's rad. It's super easy to – yeah, here it is. How to get your ex-boyfriend back when he has another girlfriend. Call 91741-409-7597. Excellent. And it's Black Magic and Val... Vashikaran? Top Vashikaran specialist Vikram Sharma is a very good astrologer and world gold medalist. Oh, gold medalist in what? Black magic, I guess. I don't know. Hold on a second. Let me pipe my thing in here for a second.

1:07:46 Pipe it in there. I'm going to pipe it in. Let me connect my phone. Should my phone be able to connect? I want to dial that number. Let me see. I haven't actually connected. Oh, here we go. Roadcaster. Boom. Pairing. That kind of work. Hey, citizen. Dave, is there an explicit rule banning AI slop from the podcast index? No, we don't have explicit rules about anything. Our only rule is... keep it free and sensory Okay, I got it paired. Now let me go back here. I don't believe. Rules require meetings, and we don't want to do it. What's that phone number again, Dave? Let me get that. Let me dial that phone number.

1:08:32 No, no. This says 9-1- Plus 9-1-DASH. Oh, hold on. That's sketch. Okay, plus 9-1. Oh, yeah. Plus 9-1. 7-4-1. 7-4-1. 409. 409. Seven, five, nine, seven, five, nine. Is this going to cost me a thousand dollars? Oh, I'm sure. Oh yeah, for sure. Let's see if it works. is ringing. Hello. Hi, I'm calling about the black magic to get my boyfriend back.

1:09:14 Yeah, you're good, man. Is Pete. Okay, your professional is arriving. What? What are you doing by profession? I'm calling about this. I had this number. I found it in my podcast to call if I want to get my boyfriend back. Okay, no problem. Send me the details of your boyfriend and your... on the same WhatsApp number named BigTelCity. Yeah, I got to hang up. It's India. I called India. None of that. None of that. How ineffective is this stuff? This doesn't make any sense to me. Why are you saying...

1:09:59 Are you scamming? I mean, the podcast index for this nonsense is not doing anything for me. They're blasting it all over the world, and then you can't understand what they're saying. You can't understand the dudes. This is no good. This makes no sense. Yeah, this is ineffective. Ineffective. Very ineffective. But there was, oh gosh. Meanwhile, I didn't get my boyfriend back. So, you know, this is like a double disappointment. This stuff is going straight. This stuff is going straight into the ecosystem. And then there's some then there's some homeschool mom somewhere trying to start a podcast about, you know, Charlotte Mason education and can't get nothing through.

1:10:43 No traction. That was completely unsatisfying. No, I always wanted to laugh much harder than that. This didn't work. Because I wanted to say, okay, here's what you do. And then I wanted you to be like, hi, I'm the boyfriend and I'm here. I mean, we could have had a whole thing going here. That's right. I was fully prepared to be your boyfriend. Salty crayon. Did you turn your boyfriend off and turn him back on? Exactly. Yeah, yeah, Metis is right. That dude was like, holy shit, somebody called. It worked. I spent 10 cents on spamming podcast indexes and I got a call.

1:11:25 C. Los is right. He just said that it was James. James Cridland. We can always call back. My boyfriend's name is Sam. We've been having a fight. Oh, I screwed up the script. I can't believe it. Oh, what an opportunity. Instead I come up with Pete. I'm so stupid. This is James. That's funny. Oh, man. All right. Hey, we probably should thank some people. Oh, you have the date. Well. We got people. I have to clean up my studio. Tina made me commit to it. Like one of those, one of those wife commitment things.

1:12:07 Oh, OK. I need you to commit to something. So you have a hard out. I've got a I've got no out. And you have I got I got a hard out. Yeah. The studio is absolutely atrocious. It's like I've got no agenda. Chachkis and hats and. I've got 8,000 books about scripture and I still have a T-Mobile bag sitting here that I got my phone in. I mean, this is bad. really bad. Well, I mean, like we did, we covered basically nothing. But there's like 50 things on my list that I have not even had a chance to talk about.

1:12:49 Well, we can do it next week. We'll do a better job next week. We had fun, though. I want to call that guy. OK, new segment of the show. Call a different scammer every week. Okay. I'm good with that. Well, we've got to have a better script next time we screw that one up. Oh, yeah. No. But it will be the James and Sam script. Every time. Sam and James. Sam and James. Call a scammer. I don't know. Does anybody still remember the – what was it? Not the janky boys. The jerky boys. Remember the Jerky Boys? Yeah, yeah, yeah. The Jerky Boys, man. They were so good at that back in the day. That was like on cassette tapes that people would pass back and forth to each other. Oh, yeah. I think the Beastie Boys made a song about the Jerky Boys, if I can recall.

1:13:37 because they were calling some guy about cookie puss. Oh, hold on a second. The crank anchors. That's another one. Hold on a second. Let me see if I can find this. This was a class. You remember, well, you know, the beast. This was actually, uh, let me see. Peace to you, boys. cookie puss. This was the This was the OG Beastie Boys back in the day. Bye.

1:14:20 17 years ago this. Now it's got to be older than that. years old. you forget this Hello? Oh, my God. the number N-Wing. Last minute. We'll be right back. What's up, bitches? And. Thank you. He's calling about the ice cream cookie puss. That was just one step removed from what was the videotapes?

1:15:09 Faces of Death or whatever. Oh, yeah. Your copy was like an eighth generation. And you always had to skip it at the end because it was the people eating the monkey's brains out of the monkey head on the table with his head sawed off on the top. And you'd have a big line of snow at the bottom that you just couldn't get rid of no matter how many times. What was the little thing that you would adjust, the little dial on the forehead? Tracking. Tracking. Yeah, that's it, the tracking. Yeah, tracking. Wow. So that whole... That whole episode resulted in the one and only boost that came in during the live show. From C-Loss on Linux. 4,329 sats. Interesting. That doesn't sound like a right number. I'm on the new helipad, so I'm jacked.

1:15:59 into oh yeah big big helipad upgrade by eric pp oh yeah oh yeah this thing had this thing had like 75 commits in it i mean it was like oh that's amazing yeah that's amazing let me see he fixed like every possible thing. Everything. Everything. No, Eric P. said he boosted $33.33. It didn't come through to me. No, Dreb said his boost failed. Hmm. Hmm. I don't know why. But C-Lost, was that a key send? Was that an LN URL? Which one was it? Oh, he did. I said four, three, two, one. Somehow I got four, three, two, nine. I don't know why.

1:16:42 I don't know why that was. Yeah, I see 4, 3, 2, 1. Interesting. So I got some extra. The new helipad now adds eight sats for every transaction. You just get a freebie. Get a freebie. Wait. Let's see. What is this? crazy. Did you get it? Random look. hub export this has become so complicated now yeah silas was that ellen url or was it keysend Stay here.

1:17:25 It says it failed for you, Dave. Oh, it did? My note is up, I think. nose I'm re-exporting the Albeho. And I just want to say thank you. PayPal. Let me see if that came through too. No, I think I changed. I don't think I have a, Fountains sometimes send strange TOV records. Let me see. Did it come through? strange fire. LTV. at uncashed payment. I'm looking at... The runway too. Let's see it on runway. Interesting. Shouldn't be on runway. It should only because I changed everything to my Albie hub.

1:18:12 helps working Drebscott just tried again with fancy new customatic. Okay. Well, I didn't. I didn't hear anything. I don't see any. My lightning note is up and running. You're lightning fast. Transaction was 11 days ago, though, so. that's all about. That doesn't sound right. No, this is all just a train wreck. Anyway, thank you, C-Loss on Linux, for giving us the only boost I got. I do see. 91 Sats from Oscar about eight hours ago.

1:18:57 Twice, test one, test two, and then I hit the delimiter. I don't think we got anything else. At least I don't see it on runway either. I'm currently exporting out of AlbieHub to... Excel spreadsheets. This will be fun. It's on 34 out of 79 transactions. Let's see if I have any streams. I don't think I have any streams either. When's my last stream? No, 22 hours ago. That can't be right. Lyceum sent 7777 via True Fans and 1111. Oh, here's Eric PP. Okay. I got 3333 from Eric PP. Yeah, I got it. That just came in.

1:19:40 Oh, it just took a minute. Yeah. Okay, so that worked, but I didn't hear a pew. Yeah. Didn't he make some changes to the sound settings? No, I think they carried over. I know he added some stuff, right? I think. Let me see. I think it does. things like MIDI numerology. No, he's got it in there. He's got the baller. He's got all that. Oh, Eric said he reboosted with Boost CLI. What did you, Eric, what did you boost at before, Fountain? Maybe. Maybe he's – maybe he's sent it with Fountain and – I don't know. Maybe Fountain – maybe Oscar's dealing with some issues. Yeah, he has tests. So if he's testing eight hours ago, who knows?

1:20:29 Yeah, I wonder if... bot abuser thing was it could could totally be that could be okay. He may be fighting a battle, so we'll send up prayers for us. He's fighting a battle that we've already won. It's a worship song. We're fighting a battle that he already won. Why don't we thank some people, Dave? Okay, I'll do this next. Speaking of Oscar, Mary. PayPal. $200. Whoa, hold on a second. Please.

1:21:10 Yay, thank you very much Oscar. Okay. Well that will cover that 200 will cover the new database upsize we're going to have to do because we're about 75% disk space on the current database server. Is that something that takes a while or is that like a quick thing? Uh, it'll take... Upsizing will take pro- Hour? 30 minutes. Oh, okay. And you're going to do that when? Please. I mean, I love you guys, but I'm not staying up until 2 o'clock in the morning. I'll just do it on like a Sunday.

1:21:51 Hmm. How much dispatch are you using? Yeah, our current box has 650. and we're at like something what an odd number 640 gigs we just get a terabyte Well, the new, the new, the. I've been pricing stuff today. to try to figure out what to do, 'cause this is what set this whole spam thing off. I got a warning about disk space on the database server. And I was like, whoa, what? I mean, because 640 gigs, we should have plenty of space. And then I get in there like.

1:22:32 We're just thousands of feeds a day. let's see, I was looking earlier What we're paying now for this VM for the main database is. Let's see, it's got 32 gigs of RAM and 640 gigs of disk. $192 a month for that VM. Upsize it. next size. Oh, it's going to be a jump. Oh, it's a jump. 64 gigs of RAM and 1280, so 1.2 terabytes. Let's basically double the cost from $192 a month to $384.

1:23:17 But I priced it out, and I was like, okay, well, can we just create a one-terabyte disk? And attach it? And attach it. And attach it. And that would be $100 a month. Oh, okay. So you might as well just go for the whole system. Yeah. It's worth it. Yeah. So they just – it's – It sucks. I hate spending money. I hate this crap. You're so good at that. I'm the opposite. I'm like, ah, spend the money. Throw money at the problem. Well, no, not throw money at the problem, but at a certain point, you can only wrap yourself in so many pretzel nuts. Yeah. I think I can trim some stuff other places. It's not going to be much.

1:24:00 Maybe 30 bucks off the 12th layer bill. Think about all the extra stuff we can do with 64 gigs of room. can do more with that. It'll be worth it. Think of all the extra slot podcasts we can index with 64 gigs of RAM. Think of all the different Wikipedia scraped Egypt history. LLM read articles we can index. Yeah. It's going to be great. And the funny thing is, is that now Wikipedia is all run by slop bots. Oh, yeah. Yeah, exactly. So, you know, it's slop creating Wikipedia articles and bots creating slop out of slop. Slop on slop. Slop on slop. Slop on slop.

1:24:43 Okay, Oscar, thank you. Thank you, Oscar. that's Boo! Five dollars. Joseph Maraca. Thank you, Joseph. Appreciate it. $24.20 from Lauren Ball. Thank you, Lauren. Mitch Downey. Thank you, Mitch. So kind. Yeah, that's the pod verse. And then. Mitch has always had a secondary contribution of $10 for him personally. Thank you. Christopher Harbaric. Thank you, Christopher. and Casa Jack. $15? We got Terry Keller, $5. Richard Turoksi.

1:25:28 Thank you, Richard. Christopher, see Chris Cowan, $5. Paul Saltzman, $22.22. Let me give him a row of ducks here. Where's our ducks? Okay. Here we go. There's my row of ducks. Even though it's not booth, but still $22.22. Row of ducks. Derek J. Vizker, $21. Best name in podcasting. Jeremy Gerds, $5. Thank you, Jeremy. Todd Cochran. Thank you, Todd, on the cloud. Todd up there in the cloud. It just... It kind of gives me every time. Michael Hall, $5.50. I want to have all kinds of...

1:26:10 podcast boosts, and I'm going to set up a bot for that. So that even when I'm dead. PayPal's and Boost will still go to podcast. That's kind of cool. Really cool idea. Dead boost. Nice. You just. free. prepay on some cloud hosting provider. Yep. Set up a node and then just set it to boost. to a bunch of different people forever. But with like random messages. Yeah, just messages just creative messages. Yeah, with like, do you want to get your boyfriend back?

1:26:55 Call this number. Gene Liverman. Thank you, Gene. Timothy Voice, $10. And that's it for our PayPals. All right. Let me see if this. your spreadsheet came through yet? Has it done it? Has it done its thing? You know, these are always fun to interpret. Here we go. This stupid spreadsheet is a train wreck. Oh my gosh. So why are you doing that?

1:27:35 I'm sorry. It's just humor. After every no agenda show, I have to do all the executive producers. An associate executive producer and put them in the credits and I have to – you know, for years, for, for. almost two decades. I've been copying pasting. these names from the spreadsheet. And of course, if someone's, you know, a knight, you got to have their knight name and all that stuff. putting it into the credits. And so I finally just like, hey. Claudebot. Can you just do this for me? And it does it. It just does it. It goes into the spreadsheet that's just sitting in my donation folder on the drive. And it knows $200 between $200 and $300 associate executive producer. It finds a night name and just puts it into the credits. Oh, wow. This is the stuff I love.

1:28:27 Well, that's not this. Yet, yet, yet. This is terrible. It's not getting the names. It's not. I've got. OK, I see a Brian. I see Brian Inzminger. The best performing comedy shows for ads podcast. I don't know what that's. That's like a pod news boost. Why are we getting. this. Hope this boost were loaded an old Acer. All in one with Umbral. I have my Albie hub running there. Hope this boost works. Thank you for your show. 73 is Bruce, the ugly cocky 73 is keto five alpha Charlie. Charlie got to renew my license.

1:29:08 You're on the 10 years. Here's some Bitcoin. Let's see, 1842. Here's some Bitcoin. You guys should start. government backup more often so it dips more what Okay, this is unintelligible. I'm punting. I got to figure this out later. If you saw this spreadsheet, you would just laugh. Well, okay, so we got 4321 from C-Loss. You already read that one. Test boost. And then we got the delimiter, Common Street Blogger, 20,000 sats through Fountain. A CompStript blogger, all the other boosts are a train wreck.

1:29:53 Comestrip blogger comes through loud and clear every single week. I know. I know. I don't know why. Howdy, Dave and Adam. Today, I want to recommend a podcast by James Cridley. . Yo, CSB, AI, Arch Wizard, PS, Adam. Yes. You spend many hours talking fictitious deities, not even Cthulhu or Roko Boss Basilisk with Jimmy, who, it's a fact, is a fully paid church employee. So when you'll give me, not a Skynet employee, a short 33-minute max interview.

1:30:36 Well, that was a grammatically okay, but very long sentence. Yes, comics or blog. I just need a day. A day. I got no time in my life. It's crazy. And I thought I'd have time today, and I committed to cleaning up the studio. It's sentences like that that make you really question Noam Chomsky's theories. Okay. We can't get into that, Dave. I got a studio to clean. All right. Is that it? That's it with the delimiter? Thank you all very much for supporting the podcast index. You know, we want this to be with us for a long time. We feel it's important. That's why we're here. You know, Dave and I both have other things to do, but we love doing it. We do it as a public service.

1:31:24 You should consider yourself lucky that it's there. I mean, really, just keep the machines running. Dave and Adam get none of this. This goes into the podcastindex.org account, and it funds the machines, the many machines, this space, et cetera, et cetera. The time and the talent is just a bonus that you get from us because we love doing it, and it's important for podcasts. It's important for the world, for the universe, I would say. We should get a Nobel Peace Prize. Dvorak says it all the time. We should get a Nobel Peace Prize for this. I agree. I agree. um And we'll get one right after Donald Trump gets one. Thank you, Boardroom, for being here. And especially, I really I want to thank PJ Mac HVAC Air Duct Cleaning for all the great work that you're doing. Perfectly clean duct work.

1:32:16 We clean out our ducks, baby. It's great. board meeting. Have a good weekend, everybody. Bye. You did you. been listening to Podcasting Too. in this Go! It's hard, man.