Episode 258: Perceptron

00:00 Podcasting 2.0 for April 24th, 2026 Episode 258 Perceptron Hello everybody from a balmy Balmy South by Southwest of the country It is time for podcasting 2.0 This is where we discuss it all It all goes down We've got the ragtag crew of 2.0ers hanging out in the boardroom because It is the official board meeting of Podcasting 2.0 In fact, we are the only board room that is not a part of Islamabad Peace Talk I'm Adam Curry here in the heart of Texas Hill Country and Alabama The man who will balance your weights according to the model you want to be Say hello my friend on the other end The one and only PodSage Mr Dave Jones! Uh... You're uh

00:51 You managed to get that done, even with the Cedar fever. Yeah... I got the cedar fever and the only thing that will cure it is more cowbell! Trying, i'm working on it This uh you don't get knocked out like this very often No no no, I had this in January I think? I don't get knocked out by much No you don't What do you mean my note isn't live in the splits what does this mean Eric PP says your node isn't live in the splits. What does that mean? Isn't in the live splits Oh, well, it's okay. Oh, it's not oh, it's not in the live item was what is that about No wonder I'm not getting live boosts Hmm hold on a second let's see Hey, let me see he's not in there really and we see value hmm What is this? Oh, I see

01:58 Okay, well that would make sense. Let's see what am I Adam at get out get LB calm can do that and This is the listen to people type on their keyboard. I really don't care. Yeah Cuz I'm doing the same thing yeah, I really don't care. I don't care what anybody thinks of me Alright so let me let me do this Yeah, well no wonder things weren't working. What? Yeah thanks Eric! Yeah really ruined this... sorry to derail the show No you did it man You were running and Eric just took his leg out I was good to go everything's fine but also the way i read it I mean this is not how he wrote it but let me see what he said I'm gonna read what the way I received it

02:54 Your node isn't in the live splits again! See, that's how I read it. And... and that is obviously the problem why wasn't getting live boost on the last show. It makes so much sense now. Okay. Yes. This was a gift from Sir PP? No of course its'a gift and I'm just telling you how I feel and my wife was mean to me a minute ago Oh, yeah. I'm so sorry Yeah, I say i'm gonna do the show and she's huffing and puffing doing the sheets And then doing the sheet sucks putting the new sheets on you see you should have just asked me whatever I'm like oh okay? And this is whatever i'm like did you just say whatever to me You got what ever got whatevered Like okay all right well This is so what does that always isn't that's not worth even talking about yeah But you whatever'd me man yes it's like uh

03:47 We'll just deal with this after the show. Yeah, well and have a nice show! Have a good show! Not I guess whatever Joe Hello boardroom how y'all doing? This is podcasting 2.0 where we discuss all things podcasting And sheets. And sheets and bed and pod pings and all kinds of stuff yes Had a fun Town Hall A town hall on Wednesday, wasn't we? Yeah. Wednesday with Alex Sanfilippo. A town hall. Yeah he put together a town hall and he done some survey I guess it's all this all happens on LinkedIn i'm sure about... How do you do what is the difference between like what are the criteria that delineates a town hall from a fireside chat

04:40 Well a town hall, I mean it's all Zoom so basically it was just a Zoom call. But the town hall had multiple people speaking and that was actually pretty good question because some people were like what kind of town hall is this? I thought I got to stand up and ask questions! No there were scheduled speakers. about the issue of people spamming to get guests onto podcasts, which we had discussed in a previous board meeting to which I had suggested the booking tag. Yes okay and so you know Tom Rossi was on and it was actually quite good. What did Tommy have to say? He said well his

05:29 His contribution was, you know, hey we're taking out the email addresses from RSS feeds. Okay. And then I came in and said, yes! And that's very good and so what we're doing is we're suggesting putting a tag in called the booking tag. It was kind of a scripted thing it was like here's the problem heres interim solutions here are the solutions...and I think overall it was really good now Alex and Daniel J Lewis are working on the tag to get that. Oh good, yeah okay it was kind of an industry rah-rah thing I thought it was quite good

06:07 Not as scripted as the interview that Pod News had with the lady from Spotify. Oh, I have not looked...I've been doing just working non-stop this morning. I'm being whatever'd. I just you know rocking and rolling. I'm getting live servers up and running and God cast to you videos now i've been so behind because of this Cedar fever So I did not listen to it what how was that? What was What was the interview about? Who is she, what lady from Spotify who's on first I don't remember her name. It was basically one of those interviews where i'm gonna send you questions

06:49 It's it this is what it sounded like to me. I'm gonna send you questions You record them send them back to me Oh record the answers and then we'll splice it all together I hate us like cuz I mean usually you could hear almost everything but the paper that she was reading off of crinkle Yeah, I mean it was fairly forgettable What was it about though? What was there how to get like wrecking error how to get boosted and in their ranking or something. That's not the right way, how do you get... How do you game your system? Spotlighted or whatever yeah and then but yeah the bottom line was basically you can't there's no... How to get spotlighted! Whatever You need to find out who is on staff and take them out for drinks like like the way it supposed to be I mean there's It sounded to me that they did it same way Apple does which is just

07:40 roll dice and pick somebody from a list or you know somebody on the inside and you get to add a boy. I don't think there's any way to break in from the outside, the whole problem is it's editorial team they start to run it like they're overlords It's not bad. That's just that's how you do it when you're running editorial You determine and everybody hates you except the people that got highlighted and everyone else thinks your dear douche I mean any algo that you put out there somebody's gonna figure that thing out in game And I mean, it's just you can't do it any other way hmm now. It's right But I'll go suck algos and regret I'd rather have it be editorial oh

08:25 In fact, you know it's like why does Spotify do editorial? Apple does editorial but not a single other podcast app does editorial which I've been complaining about for years. Do editorial! Have an opinion! We should do editorial. Have an opinion we mean WE. Should we do the podcast index? Oh, do you want to be hated? I'm not interested in that. We're already hated right? Pick a topic! I have no time to do editorial for Podcast Index. Like throw an LLM at it and just pretend we're doing it. Just to get taken out for drinks, basically? Yeah exactly! We would be just as effective as the Podcast Academy or whatever that is... No but isn't that the truth though? I have never understood this For some reason podcast app developers And maybe this is a don't want to be hated thing But if you want to delight your users

09:32 You know then look at what your users are doing and do editorial everyone else is doing it that Spotify and Apple literally are pushing shows that they think are good you can be your one-man editorial team And you can delight your users with all kinds of suggestions. No one does that there's this kind of like who has to be Equal for all mom like no No, if you want your app to be successful you have to have other things. And those other things are highlights spotlight blue light moonlight whatever you wanna do Rachel Maddow app Hello! I've been saying this forever This is exactly what it should be

10:16 One day somebody's just gonna vibe code a Rachel Maddow app and stick it in the App Store or something, and it'll just be for you. It'll be your special app. Yeah, they don't ever get through the App Stores. Do you have Rachel Maddows permission to do this? Well that's interesting topic because I don't know what other podcast developers are seeing out there but on the Godcaster side of things Paul has been telling us that App Store approvals are taking longer and longer. And he thinks it's got to do with just being overrun with slop like AI. Oh, very good point! Yeah well we know from customers that the I call them the wrapper apps but what are they called? Container apps? I think is what they're called. Yeah yeah something like that you know They're having a real hard time

11:17 And what you're describing is like somebody just takes your website and essentially wraps an app around it, but it's really just pulling your content off of your WordPress. Yeah, yeah. You have to have a certain amount of unique qualities in the app or native qualities but then they also run bots and algos across the whole App Store and say well this is pretty similar to this one... It has become very difficult Which is kind of infuriating to me on one level because I've written a couple of iOS apps in the past. Never written an Android app, but I've definitely done a couple of that iOS apps and at the time that... One of them that I did was for a company I worked for and we did a lot of recruiting

12:16 like college students to get them into our industry. So we would do a lot of work with that and then So I wrote this app, essentially it was an app focused on our... It was branded with our business but it was all like a recruiting app and we would you know as a way that the recruits could install our app and see what are upcoming events were going to be in all this kind of stuff. And register and also took a ton of work! I mean like I spent weeks writing this app And it worked great. It had a server back in to do all this kind of stuff and then, and Apple just flat rejected it. They were like you can't... This is uh they said this is just a promotional advertisement for your business Yeah isn't that the whole idea? I was like well but no it's got all this functionality to it! I mean like this is no different than You know any sort of event registration type app There are tons of examples of this No Sorry

13:22 Like, you... It's just maybe some and then you just have these wrapper apps. Yeah like you're talking about that really just wrap a website and stick it out there somehow they just sail through I don't know. If they hadn't been such A-holes about web apps we wouldn't have this problem if you could just install the web app as simply from the App Store which was the original idea You know when Steve it was it was when I met job when I met Steve Jobs Did you say that did you say that when Tina complained about the sheets but listen here when I She's like don't hit me with your Steve jobs crap again, I'm gonna do that. I'm gonna do that Whatever with your Steve Jobs When I met Steve Jobs

14:18 This was the iPod touch days. This was the device, this was the dream! The iPod touch would not be connected to a phone network and it would be web apps that was the whole idea and it wasn't really good idea but when I met Steve Jobs He was, I noticed right he was yelling at people. They fucked up Wi-Fi! I was so mad and not sure exactly what they messed up with Wi-Fi but something had happened with Wi-Fi with the protocol or how it switched? I'm not sure exactly what it was that his dream was crumbling

14:59 And this is why he had to eventually do a deal with AT&T, and AT&T as you recall was the first partner. They rolled out the iPhone... You couldn't get it for any other network it had to be AT&T. He also never wanted an app store which in hindsight is huge moneymaker at the App Store And I think you're right. The way we're suffering, or slash that the way you are suffering from the generated slop they've got to be inundated with this stuff but it's at a point where people can't even get a developer account approved within five days some have been waiting for months

15:45 Release on Android that Paul did I think it took over a week to get approved Oh really, oh my goodness didn't didn't it? While yeah It took awhile and maybe not maybe not over a week But it was close. I mean it was it was quite a number of days and that's that's a definite down Turn or what? I don't know the right way this there's a definite increase in time like in recently because it didn't use, it was much quicker before. Yeah. Um, I mean that listen to this. I think I've received my first boost spam. Oh cool. So thanks Eric PP now that I put my uh,

16:32 My split in the, my node in the split. Here's what I got. Orangefriend.com Find the best rates to swap to from Bitcoin on LN and other cryptos! Compare instant exchanges and P2P markets now with no KYC prepay cards! Satogram is what it's called. It says satogram How much was it? One Satoshi Uh...I am a huge fan of this You can spam my lightning node all you want. Just wear it out! I love it, yes. Paying me to spam me? Oh yeah bring it on baby Yeah eventually those stats will be worth something We're so back

17:21 Chad F. Chad F in the hairpin, very nice. I mean like um, you know sort of the big non- one of the big non podcast stories this week was about Tim Cook retiring? Yeah! And i just like Well, Marco will be happy. He thought Tim Cook was a huge traitor to liberalism. I feel like if you look across everybody's acting like Tim Cook was...I mean who cares? But just seeing a point that is made over and over as everybody

18:14 is saying, well you know no matter what you think about Tim Cook he was like a business genius who will never be matched again because of him taking Apple up to being a four trillion dollar market cap. Man this is... if you look across the whole tech industry there's many companies that have had a hockey stick over the last 15 years this is all- This is a ton of these humongous behemoth tech companies have had quote Runs that they'll never have again like because it's not about necessarily being any particular genius Well, I'm not saying he's bad at his job. I'm just saying that like when you when you spit When you when you literally spend almost a trillion dollars in stock buybacks Yeah

19:01 Yeah, you're gonna shoot that. That number's gonna go way up! Well also he is a supply chain guy so he did really good things with the supply chain. Oh okay, so Eric PP says you can actually turn stuff off? Ah that's very cool man So if you want to you can turn that stuff off You can turn what off? Spam. You can turn spam off under a certain number hide boost amounts below, see there you go Oh, you got thresholds. Yeah well this is of course it's a helipad This is one the best pieces software in the universe This is um... It's uh... Spam Assassin for Boost Does anybody- There's probably 1 dude out there still running Spam Assassin on his- On like a box in his closet

19:49 You know, the other day I had one of those moments where like... you know, I should probably check and see if if I really got all the Bitcoin off of that old laptop. You ever done one of those? Yeah, where you have a panic moment. Well it's just like... I famously sold 65 Bitcoin at $900. Let me just go see and so it was Bitcoin Core QT running on a MacBook Air. So it wouldn't even connect to peers or anything but okay because you know I could see all the addresses and yeah there was like 19 SATS here or there but then

20:28 AC Android wallet. I'm like, huh? I wonder oh you have it you have a mystery wall like 3030 Bitcoin like gotta find this thing and So I've talking to Tina and this was before the whatever This is one she would maybe that's the reason for the whatever at the end of this story And as an android phone so this must be around 2013 Android all your old devices though every single one and she says well when I was dating you, I remember You'd been to the strip strip joint

21:07 And your Nokia E71 was left in the Uber. I said, okay so the e71... How can you make it rain with Bitcoin at the strip club? Well no but this was the Nokia she was just trying to help me identify devices and I know that Baby trust me! I got bitcoin, i'll send it to ya! This is like The Yellow Rose It was some business thing I wasn't really going for the strippers And so I was like, okay. And and I said but I had an iPhone and the iPhone 4... ...and i'm not quite sure what came after that. So you know I go into the bin and literally the phones are in order. The Nokia E71, the iPhone 4 and right in between that was an old Pixel.

21:58 Ah, bingo. I mean how awesome is that? And she said, I will never complain ever again about you keeping all of your old crap if you find 30 Bitcoin on this thing. I bet! Yeah well and guess what zero. 0.00 Of course 0.00 Oh yeah we all have the story of selling a bit like I sold 6 Bitcoin at 1800 bucks and thought it was genius Yeah, it's seen in whatever Yes, so live and learn. So this is interesting RSS payment boost true fans Okay True Fans It's coming through a little odd but it's coming through cook miss every tech cycle search AI EV and glasses Also cloud and the list goes on not a genius he has apples bomber

22:51 Well, here's it this that's a Sam Sethi thing if I've ever heard one. Yes Now here's the only thing about this new guy and and unfortunately His name has too many syllables You know Steve Jobs Tim Cook this guy came ready as like too many syllables in his last name I don't remember is what is his last name Ternus turn is okay? It's gonna be hard John Turnas is it John Turnas John John Turner's John Turner's so he's the hardware guy well That's interesting because if I've learned anything over the past 18 months, particularly in the last six. You know Apple has their universal memory and their own silicon is highly usable for the AI Oh yeah And you know they've had all kinds of AI capable type chips in the phones

23:50 They may have an incredible and they've held off, you know It's like they they've had a few missers. They've they've pulled back from the AI nonsense And boy I can't blame him because that Samsung that I got all yes. They also was like hi. I'm Bixby Oh, oh some stupid AI agent? Yeah. Would you like for me to recognize you talking to me automatically? No! And get off and... You can't even take it off the phone. Bixby. That's the Samsung AI. Okay. Bixby. Everybody has got every company is making some agent and giving it some stupid cutesy name It's I can't- You should ask our agent, you know

24:34 you know, Latod. I'm not going to talk to your agent by name just quit it. Yeah exactly but i'm seeing the revolution unfolding You have this new... I haven't been able to test it The DeepSeek v4 There's a cloud tag on Olama It's not actually there yet This thing is gonna be pretty interesting with 1 million token context I haven't even heard of this one. Yeah, yeah DeepSeek V4 let me see the v4 pro 1.6 trillion parameters you know 1.6 trillion parameters before flash 284 billion parameters 13 billion active open weight MIT licensed 1 million token context window trained on domestic Chinese chips

25:36 You know, this could be something very interesting. These are huge! These models are giant... I mean but you use so i've really become kind of 1.6 trillion parameter model. Yeah that's pretty big but you know this what is it together dot ai they're the ones that have been most stable for me of all the rented gpu outfits This has changed my life You know, it's like oh there is a five hour senatorial testimony on C-SPAN. Okay robot go download it. Downloaded okay run it through the fastest most awesomest whisper model you have do word by word throw the JSON onto your drive and then summarize

26:26 Alright, you got a summary. All right? Oh that sounds interesting what any good fun quotes You can get there for some clips You know and and I cost 15 cents an hour Yeah, and all you would have to do then is just publish it to Spreaker and you can start getting ad revenue. But the point is so if you want to use a model like this You can run it pretty efficiently from... And I don't even know what Together dot... I don't even know what their model is Their business model? I'm sure you're using some dude's gaming computer Isn't that kind of what they all are? Let me see here. You know the new thing right with uh With what?

27:05 Trying to remember what they call it. It's distributed Distributed proxy I can't remember the name but well, so what it is serverless inference Is one of the one of the catchphrases? No This is a this is a scam tactic where you or you sign up for something and you're giving your giving Your compute you're giving Another third-party access to use your computer as a proxy. Yes, so that Nefarious actors can distribute their workload and not come from known data center autonomous system numbers also cool Yes Yeah yeah this this is it like the people are cracking down on this all over the place because people they're using it like you Installing Roku apps that have this thing embedded in it

28:01 So there all of a sudden your television is being used as proxy for a bot army. And then bingo, Bitwarden gets popped. Yeah how about that huh? This stuff is probably coming through these distributed proxy networks now as their attack vector because they're so, like it's all over the world and what you need to... Like what we do in Podcast Index. We block a ton of traffic coming off of data centers. We do a lot of heuristic looking at traffic that is coming from a data center, that is coming from an IP that's related to an autonomous system number which is connected to a data center provider

28:52 And so, like if you're run... If you have a browser user agent and claiming to be Chrome but you're coming from AWS. Yeah! You are no good. Yeah! You are not Chrome. Your an imposter. One of the huge benefits running Amarchi is, and I guess you could do with the Mac too but all my bot-based stuff. I pipe it all through my home desktop. Home IP coming from a home machine looks legit. Yeah and I'm currently moving my Freedom Controller to my house, to Ubuntu or to my podcast rig. You should run it on the same machine that's running our Albi hub because it's working so well

29:43 What? The Albi Hub is working. Is it not working? Not for Comic Street Blogger, he complained again! Oh... What but that's... Wait where like I gotta see this Podcast L index LN... what?! I don't understand because it was up and running wait well I've got Machine 2 here with and it is running right here 18 1842. What is 1842? I don't know what 1842 is was a good year for wine Hey, 1842! Was a good year AI generated boost oh well

30:31 I don't know what to say. It's up and running, it just sits right here doing its thing. Anyway so you're moving the Freedom Controller over to your own home machine? Yeah because everybody is doing this. Everybody is blocking data center traffic and my Freedom Controller for many years has been running in Linode. Right And so increasingly I'm getting 403 which is denied. Yep yep yep yep So I can't save articles into my archive anymore so I have to move it Luckily we facilitated for this a long time ago and we can you can use so one of the things frame control can do is it'll, um, you can set up a bucket in an S3 or somewhere. And you can hit the bucket

31:13 And it'll bounce you to whatever the current IP address is of where the frame controller actually is. So, yeah, that's fine. It'll work just fine this way but I don't know... You know, this is becoming a real problem for attackers. So now they're going to have to figure out a way to take over residential IPs and that's behind all of this. They just do that in the routers right? Aren't the routers complete pieces of crap that are just all Swiss cheese? Oh yeah for sure! Yeah your router has about as much... If anybody sits down with a commercial router from like residential home router for more than you know day they're gonna find something

31:57 You can just imagine that people are stripping the firmware out of those things, running them through an LLM to find bugs right now as we're talking. This is a change in the world that people are not very aware of and it's going to mean more people will be needed. Pain. But also more people will be needed to reconfigure everything The whole centralized model is under attack Yes, the slopocalypse. Slopocalypse? Slop-hocalypse! Yes Yeah it's happening and we're not ready for it We're really not But... It's so hard to find this stuff Part of it is just undetectable There's going to be some amount of AI slop that really is just kind of undetectable You can't

32:56 Automate your way around it. I mean, i've been this is I've been focused almost squarely on this for a couple of you know two three weeks now and um And I can you know? I can tell you that what i'm in model training mode right now That's that's the next step and that's what i'm working through it is complicated Um so I mean we could talk about if you want to well, I just want to answer uh Cotton gin he's in the boardroom What even is slop by the way whenever I play an AI generated song on no agenda in the end of show mix? He's the first one to say a I minus So you clearly know what it is, you know slop is Yeah, slop is what pigs eat and It's usually a lot of the same of it. I think that's the definition And this is the thing

34:03 that we can't ever get around. And I'm talking about more than just technology, this goes way... This goes all the way back to the famous quote of you know, I can't define pornography but i know it when I see it. Yeah yeah and there's a lot of things like this in the world But look You know one thing is just language in general I forgot what we were talking about last week, but it brought up this idea. It reminded me of Noam Chomsky's work on language and you know before Noam Chomsky was the sort of anti-war... Before he was friends with Epstein? You mean? Yeah that too! Before all that, before he was the lovable kook

35:05 He his biggest you know, his big contribution was in the world of language and philology. Philo-philology? Yeah p h i l o l o g y Hold on a book of knowledge give me the definition of philology Really? According to the book of knowledge, philology is the study of language in written historical sources. Combining linguistics with literary criticism and historical analysis to understand texts and their cultural contexts. Thus it has been written... Hey can I vibe code or what? It's just it would be like I can't

35:58 I can't help wishing that sounded like the priest from Monty Python and the Holy Grail. reading from the Book of Armaments. I just, that... I mean, I think I got it where I wanted it to be? That one caught me off guard! Is that been on no agenda because I was not aware this was coming. Yeah yeah yeah, I introduced it a couple shows ago so I have an little interface with a push-to-talk button and you know its piped in exactly the way I want it and took to bridge over

36:35 You know, super fast whisper but then the FFmpeg process which takes the voice and then adds the echo to it. I added a little page scribbling. I'm dangerous with this stuff! SA plus! SA plus work brother! Thank you Um, okay. Yeah. Yes. You know what it is? I like being so this is why don't listen to no agenda because uh, I like being surprised by your antics philology Okay But that's that's where chompsky Uh That's where chomsky's chops came from uh, that's where he that's where he um sharpened up his chops was in language and that's

37:23 what his big contribution was, and I think it's still pretty much the defining characteristic of language neurology is that he said that language appears to be something built into us. Like it's not something that has learned so you could say okay is language a general knowledge thing where we learn it the same way we learned how to drive a car or the same way that we learned how to Put silverware in a certain order on a table and he says no The language is this language is something that's unique It's wired into the human mind in a way. That seems to be pre-existing hmm, so you could say that humans are Are born with the faculty of?

38:29 or the property of language construction that we understand what he, I think if i'm not mistaken. I think the term that you used was universal grammar and he said this spans all different languages so On the outside it may look like Mandarin Chinese and Italian are completely different. But on the inside, it's really all tokenized? Is that what you're getting? It is really! It's all tokens. Yeah... yeah.. It just all has an API. Yeah but on the uh they looks like from the outside those are completely different languages but what's happening in the mind is something that Chomsky called universal grammar

39:18 So, it's this idea that you have a universal set of rules that we all share the structure. No matter what the overarching language is, we all operate on a subject and object in a verb and other parts of grammar that even though they may be in a different order or they maybe a little bit different but we understand what these things are. And so, I don't know if I've said this before but it's sort of like a mere Christianity. Like there are lots of different theologies and doctrines and all these kinds of things but the core of what Christianity is is these two or three things. Yeah like Jesus come quickly everyone in Christianity is thinking that. So when it comes to language

40:29 As I'm working through the train, this how to train this model thing. You know? This is beginning... I'm beginning to sort of see the pattern here that goes back to what we were talking about earlier which is these things are hard to define but we know what they are. If I gave flashcards everyone in the boardroom and gave them 20 flashcards, and said... Okay. And then played 14 seconds of each podcast and said okay here's the art, here is the title, here is the description of the podcast, and here are 14 seconds of audio. Twenty out of twenty would identify every single one of them perfectly as either AI slop

41:33 or just a normal podcast. I agree, but now when you try to take that and put it into an LLM, into a model... It is that you hit a barrier And so, and I think that the issue has something to do... So we've talked about Jacques Ellul before. Yes a good old Jacques. Many times. Professor Jacques. He's an interesting guy. He had his two seminal works Technological Society and Propaganda One of the key takeaways from his work Propaganda

42:20 is that if you're trying to convince someone or a group of people that your idea is correct, if that's what you're doing as your propaganda then what you are doing is not propaganda and you aren't very good at it. Because attempts at propaganda… Attempts at propaganda by someone who's not good at it. They'll try to do that They'll try to like explain to you why they are right? Yeah, what he says is propaganda is not about ideas about actions real propaganda Does not care about the ideas in any shape form or fashion

43:04 They change over time. They're different every day, the news cycle changes It's all kind of irrelevant What all that matters is whether you drive a particular outcome an action if you can say a thing and then That provokes an action in the listener Done your job is done with that as propaganda Would you like to hear? A readout Of one of my agents that searches for YouTube videos and has been trained to avoid AI slop videos? Sure. Okay, red flags!

43:43 Telltale AI phrasing. So it only does it on whisper transcripts personally, I think Let me break this down What many people don't realize if you take a step back and think about it one thing that immediately stands out from my perspective what this really suggests is Deep dive. Red flags in the channel, no real person or camera identifiable name or title generic aggregator channel names stock footage with TTS narration monotone flat overly uniform anchor voice synthetic rhythm no natural reading errors no pacing variations

44:24 The rule, scan first few lines of transcript before anything. If the patterns are there don't cut the audio. The content might still be useful research but the AI voice never goes on air. Adam caught seven AI slop clips on NA 1857 rejected them all on content So that's how my agent is now identifying this stuff which sounds kind of human-esque really Yeah, and the issue is now do that for you know a few hundred podcasts an hour. A minute or an hour! Exactly exactly And so this is the problem So when you're restricted to only language... ...and you don't have the processing horsepower To listen to all the audio

45:16 It is very difficult, but it's isn't it amazing though that with three watts that powers my entire brain I can do it in seconds the hand this was at this was this would this was Chomsky's Genius Was that? It's not math. No you know the way our minds work Isn't is not math is is not a math problem and it's a man. That's a meth problem Yes And so it's this, like the ideas I think the nature of what if you take sort of what Jacques Ellul was saying he said that action and words and actions they share sort of a congruent relationship. They function together. Words drive action and action drives words

46:25 They have a relationship that is mysterious, but it is there. When we talk to people we have relationships with, it really matters what the ideas are because we want to be on the same page with each other in a way that something like propaganda doesn't care. Propaganda just cares about actions but we don't when we're having real relationships. There's this fundamental paradox with language that the words themselves sort of don't matter, but at the same time they very much do. And I think that's really what you bump up against when you try to make a definition because in one sense the definition of something is so hard to put your finger on

47:22 Like in the one sense, the words don't matter because words are kind of fungible. The English word is five and the Spanish word is cinco I mean like they're interchangeable Also the meaning of words change over time Yeah so the words are just a... The words are a conduit for meaning and that conduit can take many different shapes Square round oval whatever Essentially like the language is not the point the ideas are, the meanings are the point. But then there's this other aspect where the words themselves ARE incredibly important like in fact they're critical so on the broader scope the words don't matter but when it comes to conveying the meaning to another person or another language passing sort of... The utility of those words you choose are a critically important aspect of it

48:22 So there is like a fundamental paradox that exists within language. And this is why I brought up the propaganda reference, because when it comes to propaganda word choice... Word choice is everything because the actions you desire as outcomes from your words were as a result of the emotional stirring- Of those words that you use? ...of those words that you deliver as they sound to the hearer You know, so like a good example of this. If you think back to the election of this most recent election Trump is that one of my favorite examples is J.D. Vance there was sort of like when he was nominated as running mate all the typical attacks started political attacks and then what one thing that came up with people started calling him quote weird. Do you remember? Yeah, yeah, of course

49:18 Weird is a word that young people use to mean that a person makes them sexually uncomfortable. That is what, I have kids... When they say the word, when they say that person is weird, that is what they mean by that. That is not something that older people without kids would or people with older kids would be familiar with But that went like if a young kid is watching like, um, a movie with a sex thing. They're like oh, that's weird That is they-that is what they mean You would and so like you're only gonna know this If you have children that are like in their mid 20s or younger And so research definitely went into the use of that word and it was chosen to convey that exact meaning To that generation of people, you know So like the language

50:20 evokes actions is one aspect of it. The language and then the meanings though that the words convey can change over time, so as I'm going through this model training thing, I hit this point when I'm interacting with LLM where it asks me like... It says do you want me to spot check the output blah, blah that it's doing. And I just stopped and like what do you mean? What does that mean spot check? Like what is it...I have no idea! But what...what does it mean its going to look at something and tell me that its right but in what parameters like this is the problem we're

51:24 It's trained on our language. Yeah, and so it's talking like us is trying to communicate something to you But when I say hey Adam do you want me to spot-check this? You can instantly get it. You have an understanding of what I mean Just innately you don't there's we're on the same page just by default but when in but when the machine asks me Do you want me to spot check it I'm really, I was just sort of paralyzed for a minute. I think you might be overthinking it doesn't mean exactly the same thing as if i said it to you? No because what if I say yes It says do you want me to spot check this data If I say yes What am I communicating to IT? I'm not sure Because if I say yes is it going to say... What is it going to then DO? And then perfect- What would you expect it to do? But see here's the question but the question though is

52:22 This is the action part. This is the part that I'm getting to, if I say yes and without knowing exactly what it intends to do... What is it going to do? Hold on a second If you said it to me Now if I said to you Do you want me to spot check this Dave? And you said Yes What would you expect ME to do? Why would you know what I was suggesting So, we would have a context right? If I'm saying if you say hey Dave this feed is not updating in the index and i think that there may be a few more of this type of feed that are not updating. Can you spot check that? Okay. You would...I would immediately know what

53:24 You are communicating to me and what actions you expect me to take okay, so apply that to your example? You were doing something and the LLM says do you want me to spot-check this what was the process you were doing right? And so what I have found through this process is The process of model training is that a good 90% of the work of model training goes into preparing your data correctly, your training dataset. Having an excellent training dataset is literally 90% of the job and to get that I've had to go through a lot of steps The first one was building what James wanted which was the dead feed problematic feed

54:27 export. That is now built and it's functioning, so I can't release it really as a public link because it has DMCA stuff in there we've been asked to remove so I don't think that we can release that publicly but I think it would be okay to release it for research purposes. So somebody would have to like contact me and ask for it, I think. Chad F says hey Adam can you spot check this booster gram? Done! Contacts understood human-to-human communication complete yeah And so if somebody wants it James or anybody else you know call me we'll work it out I'll work out how to get it to you

55:22 So that was step one. And as I'm working through what it required was to put a bunch of new SQL exports statements into the index to get all this data out in the correct way and then export it into a SQLite database just like the other one, what I wanted is two separate SQLite databases, the standard SQLite database that we export every week and this new problematic database that we're going to use as the test corpus for the bad feeds. Yeah And so, as I'm going through this... You know you're gonna get criddled on this That's okay He was like, Podcast Index is deplatforming people! This is wrong!! I've been criddled before and I'll be criddled again It's fine it's not a problem

56:19 So, as I'm working through this, I exported the data and there was a problem in that the SQLite database initially came out like corrupt. But the LLM gave me a list of issues it was like the SQLite database is unreadable but also the CSV that we exported was missing, it was like missing a field. It gave me four or five different things that were wrong and then it asked me if it wanted me to do X Y & Z and spot check the data And I'm like what? I don't know what you're referring to

57:17 Yeah, and it said it in such a way that I was left with the impression that if I say yes then its going to perform an action that I may or may not want. And Im like man this is vague. I dont know what we're doing here anymore. So this... but this all part and parcel of just the difficulty of saying of having to describe something that is basically indescribable. Like you, what your example was you gave it a list of things... You say this tone, these phrases... I didn't actually give it that list. It came up with that list itself. I didn't give it that. Well you came up with a set

58:18 heuristic phrases. No, I didn't it did that itself It did okay based upon a feedback of loop of me saying probably ten times That's AI slop Okay And we're all to a certain extent going through this same exercise We are developed were trying to come up with heuristics Yeah To identify some stuff but what we're gonna end up with is And this is always how heuristics works. What we're going to end up with is the same thing that happened in phishing scams, the way that any adversary and I'm going to consider AI in this example, I'll consider AISlop our vague adversary. The way any adversary combats heuristics is they shrink the sample size. They make it smaller and smaller and smaller

59:22 until there's just simply not enough information there for you to make a, for you to pass a confidence threshold. And that sample size in something like email spam or email phishing can just be the number of words they use. So they may just pick butchering perfect example. Yeah, hey did you go? Are you going to be at the tennis practice today? There's simply not enough information there yeah There's just not is so and but when you do when you're dealing with something like AI produced podcasts well now you have many variables

1:00:01 You have all the metadata that goes into the podcast itself. You have the voice of AI slot presenter, you have the content... There's tons of variables you can push and pull around but for each one of those there is a set of heuristics involved and you can refine it down over time to a point where it is indistinguishable Because if I get on, this is what i've been learning all week. This week What I'm trying to do is create a Laura adapter L-O-R-A So the way this will work Is that Lord of the Rings adapter? Yeah, I think it stands for

1:00:59 Oh man, I cannot remember what that acronym stands for. Book of Knowledge! What does L-O-R-A stand for in AI? It's having a hard time... According to the book of knowledge LoRa stands for low rank adaptation A parameter efficient fine tuning technique For large language models and other deep neural networks Thus it has been written Low-rank adaption. That's it, that's it. Nailed it! Good work. So you... the LOR, the LoRa adapter is a way to add information into a language model into a base model and actually you're making adjustments to the weights themselves This is fascinating I love this so the quote unquote the weights

1:02:00 You're really talking about blocks of... you can think of an LLM as a stack, sort of. It's like a stack of matrices that are layered into each other and those layers are distributed in blocks across the model And one of the parts of the model is, which is the best name in all computer science, is the perceptron. Perceptron? Yeah so you have the attention layer which feeds into a forward... what do they call it? It's like a forward functioning, a forwarding function and then that goes into a perceptron or an MTP layer

1:02:56 And essentially all it is, is that it sounds fancier than it really is. It's just a block of matrices that try to make associations with the attention between words. So and it really is just matrix math right? So if you take... so the LoRa adapter is a set of matrices that you pull out of a training data set and then you add them into the LLM server, so something like Llama CPP or VLLM. You add that LoRa adapter in to the serving harness and then it will do an additive function to add those new blocks

1:04:00 into the matrices of the large language model. Wow! It essentially puts it through it The serving harness? Yeah, so like something like you can think of an agent harness would be something like cloud code or open code Okay, okay The server harness would just be the same thing on the server side So it's not changing the base model. So you're going to take a model or something like QN 3.6, or something like that. Like let's say you take QN3.6 35B model as a mixture of experts' models with three billion active parameters. You're going to take that model as your base and then you are going

1:04:44 essentially take another set of weights that are trained on your own data set. And you put that in front of it? You put that beside, yeah beside it so now every time it runs through the Perceptron is also going to add these other weights into the mix In a way that sort of nudges the output token in a certain direction And the good thing about this, which is really nice is you don't have to change. Doing it this way you're not retraining the entire model You're just adding this extra component and that extra component is small It's 500 megs or less and you can just swap it out So you can train a new one every month if you wanted to Just add that in yeah so like if you The reasons you would want to do this are

1:05:39 specificity or high customization because you can do things like this with like rag, you know, retrieval augmented generation where you take a database of content and sort of mix. You mix that kind...You run the output through the rag as a filter on sort of the output side or pile everything into the prompt, into the context window which you quickly run out of space. But the LoRa adapter this would be called fine-tuning the model if you do this well and it can fail It can easily fail if you don't have the right training data

1:06:35 Eric says, is this better than old machine learning techniques like random forest or neural nets? I mean you know transformer models are really just an evolution of neural networks so this is going to it's very similar. Well and I think have a different question between the server harness and the LoRa when was the last time you took Melissa on a date There will be no dates until school's out. I'm just saying, before you get a whatever... Yeah wait luckily I've changed the sheets that's my job so I do not have to worry about it. You nanny! All right

1:07:30 Good question. It's been a while we're do just saying soon as schools out So yeah, you're yeah Eric it handles it handles language better because it's built for It's billed for that sort of Dino tokenized input But the the train in the training if you do it right If you do it wrong here's what can happen? If your training set is poor or if it's too small or too big, you can end up just making the model memorize things. So really to make sure that you're doing a good job of this, you have to do a lot of testing on the backside. Spot checks! Spot checks yes You have to do a lot of testing to set...to make sure that its not just memorizing

1:08:36 the training data. Yes, and that because what you're looking for is a balance of weight of waiting of word connections and so the best way I saw this explained was You can think of A well-balanced model you can do you can do the underlying matrix math? yes exactly Eric that's right it's called overfitting you can do the underlying matrix math in and you can literally come out with, and see a visual representation of king is to man what queen is to woman. Like you can say... You can take the relationship in The Matrix of King to male

1:09:37 And then you can do the exact same math on the word queen and it will predict that it's going to be woman. Okay, so that's a good sort of like what is woman? Our Supreme Court justices can't answer that yeah this like this but if you over trained it You may end up with Queen equals Elizabeth Yes See what I'm saying Like you would end up with a model That has memorized The training set And so this is where we're headed. If I can do this right, we will end up with a highly customized model that is an expert at identifying slop. Wow! Not just slop but slop and spam

1:10:37 you know, malicious content and these kinds of things. All these things we don't want. We can have an expert that is they can hit those things and identify them in a way that is forward moving Whereas just doing it the way that we're doing at the moment by piling more and more and more stuff into the context window is just not scalable. Right, and then eventually this will wind up as an endpoint that people can query or a flag or something that can be used by people who use the index? That's the idea isn't it? It already is I mean like if you hit the problematic endpoint right now report recent problematic endpoint

1:11:29 you can already see what it was classified as and a short text about the reason. Is that open or do you have to have a key for that? You gotta have a key for it, yeah. And so what is doing is this train... So taking all that into, you know, taking all that context in with what I've been hitting is the training set design is so complex. I'm aiming for about 25,000 quote good podcasts. Oh but good ones okay good one yes so they and those need to range across all the different ways that we know that they do Wow everything from yeah everything from Joe Rogan

1:12:28 To some... Oystein Berger. To Oystein Berger, yeah exactly, to Bowls with Buds, to Pod News Weekly Review, to a person reading a bedtime story on Buzzsprout, all the- to some teenager reading their term paper on Anchor. All of those are really legit podcasts but then somehow identify the difference between a 17-year old in Delhi reading their report from school and differentiate that as good against an AI generated voice on Spreaker, reading a bunch of history nonsense off of Wikipedia. Right wow That's interesting challenge Yes it is very

1:13:28 It has been a challenge and I mean it's not where, it's not...I think I still got a few weeks to go. Realistically is probably going to take another two or three weeks to get the first shot at this because if you train, you know, if your data, if your training data set is wrong or if its poor then you're just gonna end up training the model that everything from Spreaker is bad Yeah yeah And thats just not true Well kind of is It's like, you know... Kind of. Like if Spreaker and the other here is what would probably help if they put a flag in this as a free account that would probably be some good signal for your training Let me give you great example copy link Here we go here one Is it the number I can call as James?

1:14:35 No, I don't think so. There's no numbers in this one. I'll find one of those for you. If you care about predictions and props right now... Of course pre-rolls, two prerolls every time Always do pre-rolls But here is the twist We're only going to AI within three seconds And I'm Luna So excited Now at McDonald's a McDouble is $2.50 So you can get your gym gains on Or just get lunch I can't, I can't shuttle. It's almost here. Hold on. Yeah yeah McDonalds alright they made some money off of us! Alright here we go

1:15:13 Hey, welcome back to Learn Tamil with Vexingo. I'm Lucas and I'm Luna so excited for episode 13. Hi slop Yeah, we've been on a roll today. I thought we'd do something a little different. Oh goodness Let me uh, let me do something here Analyze this podcast AI slop or not. I'm just going to give it to my robot, it'll take a minute obviously you know My robot is nothing like your building but anyone's who's listening heard that right away Oh yeah this is what I am saying if we This is the difference between the built-in language and an emotional action comprehension that humans have versus machines

1:16:11 We just know things and we don't know how we know them. We just do yeah, we just we just know and nobody can explain the the Nobody can explain how? but everybody's gonna pass that test every single one of us but AI is really gonna struggle with this and Well it right off the bat it'll struggle with the ad Exactly. Oh It has to get past the ads great point I wonder if my robot will figure that out. Well, and this so here's sort of the last kind of point I want to make is I think here is an Achilles heel of how we discuss these things today We tend to think about AI in terms of what it's not capable of doing or where its flaws are right now

1:17:22 And I think that is just, I think that is very short-sighted. What I think we have to do and this is why I'm trying to do it this way what I think we have to do is assume that these AI generated podcasts are going to get so good that we cannot tell the difference We have to assume they are eventually be perfect in their tone in their word choice, they're going to evade any sort of detection. This is what I meant by saying they're just gonna reduce the attack surface. They're gonna stop doing the things that sound unnatural and the uncanny valley's gonna shrink down to such a small degree that we're not going to be able to just do heuristics on them. What we're gonna have to do is look at

1:18:27 the whole thing in a Bayesian way, where we're taking things into account that they can't stop. Would you like the verdict here from my robot? Yeah sure! Almost certainly AI slopped. The evidence 55 pod... red flags 55 podcasts across every major language no human operation produces 55 language courses That's a content farm Two warm voices Exactly two TTS voices across all 55 languages. Real language instruction requires native speakers for each language hosted on Spreaker, huh? See I told you! It says hosted on Spreaker one of the platforms Dave flagged for hosting in clone spam feeds on last week's show Holy crap

1:19:19 No team, no instructors name. No methodology, no credentials completely anonymous generic Fexingo branding across all 55 template operation learn insert language with Fexingo for every single one Zero information about who runs it? No company location of people just an email address the math fifty five languages times even ten episodes each 550 episodes of audio with two voices that's TTS The pattern matches exactly what we screened for. Generic channel names, no identifiable humans and possibly broad output from a tiny operation and a template applied across dozens of instances This is a podcast spam farm using AI-generated TTS language lessons to harvest ad revenue across 55 ads on Spreaker Textbook slop Okay Great analysis But let me throw a twist in there Okay

1:20:12 One of the things I've been seeing a ton of lately are people taking LibriVox recordings, reposting them on Spreaker to get the ad revenue. Okay So that's theft? Well not... no actually legally because LibriVox uses a public domain license You can really they literally say you can do whatever you want with it and we will not stop you Alright so Legally it's fine except that you're going to have a hundred different people posting the same exact thing over and over and over again. Here's what I would actually enjoy, because LibriVox since it volunteers many of the recordings are horrible. Yeah, it is actually improvement A really well done AI narrator, like high quality AI narrator would be better

1:21:15 than many of the LibriVox recordings. And I would prefer to listen to the AI version, if it was well done then I would the LibriVox human version But aren't we now down to what i've always said? Is some people will actually want this for certain reasons and that's very individual We all need our own... Percepticon... Perceptitron Perceptron! I need my own perceptron Yes So, see and this is the issue. Just because it's AI generated does not mean that it's slop and nobody wants it but if you ask somebody is this slop they will be able to tell you. They will be able to tell you for reasons they don't even themselves know. Sir Bemros has a good point

1:22:07 One of the most powerful signals for AI slop is that the podcast has ads. And that's another thing, it's like you don't want to overtrain on that either. Yeah exactly. Because then because then the model will be like oh yeah where he's got ads must be slop but that's not true and if there was a really well done AI voiced version you know, Frankenstein. Mary Shelley's Frankenstein that was way better than the liberalized recording I would not only enjoy it and want to listen to it but also not mind if it had pre-rolls in at all. Well back to my point isn't ultimately that we all need our own perceptron? We really do

1:23:00 Because you're saying things that are really important here. I wouldn't want to hear that, I don't wanna hear some crappy Frankenstein but i wouldn't want to be... no! I do not wanna hear that But so isn't that kind of the point where we have to be? That we all have a robot that we train for stuff that we want? Customized. Yeah. Customize yeah. Customized but in the interim while there's sort of like brokers of pipelines of content, brokers of content like Podcast Index and Apple's podcast directory. And these things because things like we can't you know the con as a conduit We have to be able to continue to conduit. Yes yeah true. We can't be overrun with 700 people posting you know LibriVox recordings or Frankenstein with a couple of pre-rolls

1:23:55 It's just not going to, it's just that it's absurd. Anyway so I mean that's, I think that's where...it's difficult we're gonna you know we can try to hit the high level and train this model to get the obvious junk like the one I just posted a while ago about the podcast about the Denny's menu Yes That's obviously junk that nobody wants it's the equivalent of punch-the-monkey ad banner in your web goodness But those were good for a while. Those were fun. I've tried to punch the monkey everyone's everyone's punched the monkey So so you would actually wouldn't mind a good TTS reading of Frankenstein and

1:24:44 Oh yeah, like if it was a high quality... when TTS models get to the point where they can really deliver a good quality thing and not just be LM notebook. When we get to that point? I'd love it! Because I love audiobooks and I like LibriVox, but so many of the recordings are just so bad they're hard to listen too. So I asked my robot to find a reading of Frankenstein that is well read by TTS good enough even though it is AI slop? That would be a perceptron that you might program right?

1:25:26 Yeah. Okay, well it's not there yet it's working on it because we're working It does have my podcast index API key and secret Oh doesn't? Okay let's put I'll look forward to the server crash In a .env file don't worry in a dot end file so it's not posting it anywhere Until you get a CI, CD pipeline hack. I don't... I don't CDCI anything. CDCI? Whatever! I don't do any of that stuff Well, I hope I don't cause any problems A lot of people have keys

1:26:07 Every time I run pip install, I'm just like... Well isn't it with just every packet manager NPM all of this stuff. Yeah they're ticking time bombs inside your machine just waiting to crap on your front porch yeah well you are doing important work I'm doing work. I don't know if it's important, but... No, I think it is! I think it is important let me see he's gonna get a two-minute sample into my show folder here so I can play it for you. I got my mom pretty happy with my robot and now and I really want your little guy that runs out there and gets stories and stuff no not that one but the one that just analyzed the podcast oh that's my favorite guy

1:27:05 Yeah, I like that guy. No, that guy is good. Um... Okay- I want a copy of that guy so i can make it work on my behalf Here, my guys got some for you Letters Frankenstein or the Modern Prometheus by Mary Wollstonecraft Shelley This is a LibriVox recording All LibriVox recordings are in the public domain Do you think thats real? Or is that slop? Sounds real to me According to my robot its a TTS For more information or to volunteer please visit LibriVox dot org Recording by Caden Clagg. TheLunarIsland.blogspot That sounds pretty real to me Sounds real to me Let me see Some people just sound like robots

1:27:46 Yeah, I just think that's real. I think it's real okay We think it's real you're talking through a tube Loser robot And I have to call my any AI I call the robot because it reminds me don't get involved with this thing It's a robot. Don't take it seriously No You can't you can't that's dangerous before you know what you said? Okay good night talk to you tomorrow Yeah You have no idea how many people are doing that brother. It's bad, it's bad... Hey we're over time man I should've gotten you out eons ago. I'm sorry about that Yes we're way over time Do you have a few minutes to do some thank-yous? Yeah sure Why are you talking through a tube? Yeah I'm uh doing a libervosh recording

1:28:31 Value for value podcast, which means we deliver the value. We hope you enjoyed today's board meeting and we would like to receive some value back lots of people contribute time talent and treasure And in here we thank people throughout the show for all kinds of things they're doing I mean started from me right off the bat with Eric PP with a with helipad best software package in the universe And then of course he also scolded me But on the treasure side, this benefits the podcastindex.org infrastructure directly so if you send us Boostagrams or if you send us PayPals which can do by going to podcastindex.org at the bottom a big red donate button and hit that and you can send some fiat fund coupons So there's 333 from Chad F he asked me to spot check the Boostagram Cilas on Linux 2220 oh interesting

1:29:27 from Fountain. It might have been a row of ducks, but Fountain somehow is rounding up or down and he says firm weight update available this is probably when you crashed. That was it yeah here I got another 59 sats interesting from True Fans co-listening and chatting with Sam Sethi talking about beverages tea from Assam India and rose wine from Provence France do yet? And it stopped at the why They're talking about things completely unrelated to the show. Yes, from podcast guru 333 from Eric PP pew yes we got the Pew There's the 170 sats Fetch metadata Oh that what is this little fetch metadata thing hold on oh wow okay I'm sorry so

1:30:20 Oh goodness gracious, this is awesome Eric PP. Share with the class! Yes so on these weird lightning invoice that comes in that I wasn't able to read there's a little button now in helipad which says fetch metadata And then boom, it translates it. And so now I see indeed a row of ducks 2222 from Lyceum co-listening with Sam Sethi talking about beverages tea blah blah blah Do you have the ducks in a row? Cheers! So I don't know exactly what's happening but I love it Cool. Yeah, whatever's happening is great 6390 from Sam at true fans out FM yeah that was about cook missing every cycle Salted crayon 333 has been months how do you know'd he says? 333 from Chad F this booster Graham brought to you by Eric PP plus-plus

1:31:11 2220 from C. Los on Linux AI generated booster Graham spam and Okay, so then there's that actual spam I got and now I hit the delimiter So we're in a good good place Thanks everybody these that was fun an Eric PP helipad rocking it man love it trying to get my being pulled up here Oh yeah, okay? There is what do we got here? I guess see this Oh, that's a... I hate the way it mixes all the PayPal stuff together. Never can get my head around that. All right here we go. Oh! Look look Bus Sprout thousand bucks Holy moly! Baller! Shot caller 20-inch blades on an Impala Yo baller boys and girls from bus sprout thank you Tom Rossi It is not even end of the month yet

1:32:11 Tom Rossi also does conference calls. It does for free! Yeah, for free Michael Goggin five bucks Thank You a thousand to five in the blink of an eye Jorge Hernandez $5 thank you Jorge Christopher Reamer ten dollars thank you Christopher yeah Cohen Glotzbach five bucks thank you Cohen James Sullivan ten dollars And Randall Black, five bucks. That's our PayPals and let me see what we got on boosts here. Oh, we got Oscar. Oscar Mary 20 thousand stats through Fountain he says sorry for the disruption guys. No problem no worries brother I will throw no stones. Comic Strip Blogger Delimiter twenty thousand sats through fountain

1:33:05 Howdy, Dave and Adam. Today I want to recommend a podcast from the Podcast Morning Chat www.podpage.com slash PMC slash which is quote daily morning show for creators by creators ever wonder how top content creators and podcasters keep their shows fresh engaging and profitable The podcast Morning Chat, hosted by Mark Roenick, suggested by Martin Lindeskog. Yo! CSP AI Arch Wizard $15.52 win cent." Thank you very much Comics for Blogger That's it? Everything else is just river bitcoin spam So I hit my... this has not happened This happened twice now Ever since the wonderful upgrade I've hit my limit on Claude Code

1:33:58 Oh, did you see? I saw that. Yeah, I saw the GitHub is now charging per token and they've gotten rid of their all-you-can-eat. GitHub is charging per token for what does GitHub do with tokens? GitHub copilot. Ah! So now I can upgrade or wait an hour in 54 minutes What is this bull crap? It's falling apart. But I thought I had extra credits, man. Yeah, you created it! You need some credits! Hey man...I need some credits How come you can't give me some credits?! This is no good any credits Resets at 4 10 p.m. So predictable

1:34:38 So predict well, I mean I predicted this. I said it was going to happen but now I'm a pusher model I'm kind of pissed like man all that mail it like and gets I can get some more money You got have you got are you on the hundred plan or the 200? I'm under hundy plan Nope, gotta go to the tube. I don't want to go to the tube man! It's like...I can't buy food for my mama This is g- this is the best way to...this is token inflation is what this is Tokenflation? This is tokflation Tokflation Yeah, the best way to do it because you don't actually have to raise prices You just inflate the amount of tokens that get used with these requests and they just run out sooner This is great But what I don't understand is I thought I had

1:35:24 I thought i had extra credits man. Your 20 becomes 100 then your hundred becomes 200 and next thing you know bada boom you're on the API plan Oh well the API plan that, I mean that's the one that's crazy oh you'll go broke in a day. Oh goodness gracious let me just take a look at here what is this billing here we go billing but I have extra credits I am literally a whore. Let me see... Okay, man alright I've never run out of tokens on the Claude $100 plan until this past week and I hit my limit for the first time. Yeah, so it's happened... And I changed nothing about the way I work It has happened twice for me today! Well this sucks! It's Opus 4.7 and they changed the default if i'm not mistaken check your settings

1:36:23 I believe they changed the default to 4.7. To super high effort. Oh, okay! So make sure your effort is the same and you may want to change back to like 4.6 or 4.5 Wait so where do you that? Effort slash effort Yeah And then what is the... What do I say low medium high max auto Medium is what I've always used Medium But I noticed today when I opened it that it was auto set to extra high Ah the bastards Yeah, you're right Cotton Gin. I've heard that the 4.7 people hate it So how do you set the model is a model? Yes model. Yeah Opus okay model What is supposed to be does not auto completing for me four point six But dude just probably pointing four point six as we do that. No Model four point six not found on opus four points things Opus is it

1:37:22 Dash 4.6. I got no idea. Well, come on help me out here No can't find it anyway days We're in I'm telling you the local model stuff Oh, no that's that's the future try Whenever you get a chance Try Try Quinn 3.6 the 35 be a three B model. Let me write that down It is, it is bull. Quinn 3.6. Quinn 3.6 to 35B A3b model. A3 B model okay use on open code run it locally and run it on open code. It is really really good awesome. And it's so fast I mean it is like it is giving me output almost before I hit the enter key

1:38:26 It is lightning fast. And this concludes another episode of What the Heck Were Dave and Adam Talking About? It's a beautiful thing! Thank you all very much, Boardroom will be back next week. Dave thank you brother have a great weekend take Melissa on You have been listening to podcasting 2.0 visit podcast index org for more information You can spam my lightning node all you want