 published in the blog, then it doesn't fit. Do I, I don't know, you know, it's confusing. So let me show you what I got here. So the, I played around with a couple, couple prompts and this prompt actually follows a question where I said, so what's a pattern language and it did a decent job of explaining it. So then I gave it, and it made very short versions of the patterns, but still it's pretty useful. And so I can imagine doing this a little bit bigger. And to me, it ends up looking again like a massive wiki, a whole massive wiki on its own subdomain. Is that overkill? Is that crazy? I think this makes sense. Yeah. And we're like, you know, like, I was thinking, even like as a general rule, like, I mean, for the hour, I would just like same shape. I think that's a wiki, we have said, like, like prompts could be not, those could be prompt. So the page could be actually the full prompt unless we hit any limits. Peter, are you gonna close this any place? I like, I'd like to flesh it out a little bit more and make it. I know a bunch of people would be interested. Yeah. It's funny, huh? Yeah. Isn't it? And then the funny thing is it's like, oh, I don't know if those are actually the related patterns I would pick, but that makes me think. Yeah. Who'd you want? Yes. Yes. It's like, it's like this thing is sucking us into system three thinking, right? System one is like a knee-tock response. System two is engage the gears. System three is like, oh, gears just got sucked out of my head into some device which is collaborating with me. What? I mean, this comes back to what Aaron was saying. Is this useful? The thing that people are using chatGPT a lot for is to generate lots of content cheaply which is not really a problem that needed solving because there was already lots of people generating lots of cheap content and lots more of it is not gonna necessarily help. But they're gonna need to put that somewhere. They're gonna need to put that content somewhere, right? So that touches on what Pete was saying. Where do I put this? Do I just pull up all massive wiki for each thing? Well, you know what my answer is. Put it on your hub. Yeah, I like it. I said it. How does my hub make each of the little things or the big things accessible if I may ask? Well, I shared in chat, in fact, I've written you, I'm writing a blog post which at the moment is a briefing for you and Bill and Tim, my developer for next Monday. I've shared it in the chats of anybody on this call can look at it, but don't share it any widely because it's still a draft, it's still just a briefing. Eventually it might become a blog post but this is the only way I can really think things through is to write it as a blog post. And I would like to see how chatGPT could help me with that, you know? That's my personal interest. And so what I would like to see is to be able to go into my back office and access chatGPT whilst looking at all of my notes, private and public notes. But also on the front office, people would be able to ask my hub, what does Matthew like about this or, you know, write a blog post in Matthew's voice using the content that he's hubbed about X, Y and Z and just buy me a coffee for that sort of thing. So, yeah, that's basically the answer to your question is in that blog post, big. Okay, thanks. For what it's worth, by the way, Jerry's got the same kind of wish to open the contents of his brain. Yeah. Yeah, it's a little bit like free Jerry's brain what I just described, isn't it? Could we maybe start, this is the moment, maybe. There's a moment there we go. Maybe we could start like just writing all these and using the proposal as just as the commons or the knowledge commons or the genetic commons which are like, you know, like Kassam came up because I think that will be the lingua franca to some extent, well, it's not, it's not a lingua franca. It's more like a common term. So the commons itself will be, yeah. But just so we make it more platform-agnostic to some extent, that would be my recommendation. Yeah, I guess. So then we can think of like, you know, how to write pattern languages in the genetic commons. Well, I wrote for a massive wiki for us to all play with Chet G P T. The formal Chet G P T, massive wiki, what a mouthful. No, but I think that would be fun. That way, all we would need to do would have a very simple syntax for, you know, this is the prompt, this is the output. Yeah. Just a little bit of structure so people would find their way. But you know, I think Pete, you described your heart air as a hairball, it's not a hairball. It's a website with chapters and you go through each chapter, you go from one to another, that's not a hairball. But lots of things are hairballs, but that's not it. And TFT maps not a hairball. And I think we could probably do something quite quickly in massive wiki for this. Wouldn't require any custom code. Just a custom syntax. Yeah. You can imagine a setting where you are like, each wiki link is a prompt, right? And like, to some extent, the pages that exist on smart ground are, you know, providers for a wiki link and you could hook up like a majority models as providers as well. So if you own one day and then you have like a perfect activity which is like, you know, the relink in the wiki, you are like, well, there's no relinks anymore. And to some extent, that's a great opportunity for the person who was a link to exist to choose from the generated inputs and say like, I'm actually going to store it. And then it becomes a wiki page, which of course gives the comments which can be used to train the models as we have seen, you know, like this fine tuning, you know, the fine tuning experiment based on shared gbt. Yep. So that could be massive wiki for sure. Like, I think, yeah. Yeah. I like that a lot. It's interesting, Pete and I were on free to raise been on Monday and we didn't have a lot of the people, a couple of people couldn't make it or whatever. And we had a bit of a crisis of free to raise brain and we're like, where do we go? Cause the experiment to export my brain and feed it to Kyle. Go ahead. Don't cancel it before I go. Yeah, yeah. Exactly. Okay. And we ended up with, maybe we need a recharter for this. And I ended up also saying that there's stuff coming from the brain and integration with GPT that's actually really, really interesting. They sort of puts a twist on it. But this conversation sort of puts a different twist on it as well. And I don't know, I think we got a, I'm inspired to figure that out and see where this is. I have to go guys. It's like, well after, well after my bedtime. Thank you for hanging this long. Yeah. I really appreciate it. See you soon, I hope. And have a good day over there in the U.S. Sleep well. Aflansian, when we're done, if you'll send me a link to the video, I will upload it in the same playlist. Where it's better. Thanks so much. What else? Too bad we got nothing to talk about. It's fun. I mean, this is like Kyle, our friend Kyle Shannon started up an AI salon because he felt like this moment was very similar to early web days when he started the World Wide Web Artists Consortium in Manhattan, which is how I met him. And then how I met a bunch of buddies who were still friends in that community. And it was like a place where people were sharing, how do I build this web thing? Like, and avoid the blink tag and do cool stuff. And we're at that moment again, it's very, it's lovely. I have to say, of course, in capitalism, which I, as a consignor, as a, you know, just as a banner, I guess, I think you're right. But, you know, like, I keep thinking, how can we leverage this moment for like the commons, right? Which is why, you know, so, and to some extent, you know, you will, you can imagine that some, at some point people, like more people, I mean, we get up to the fact that what they have written online, what they have, you know, posted online has a lot of value for them because it will allow them to like fine tune, it will allow them to like, you know, teach models how to do tasks that are interesting. And also as they see people selling them back their data, which is, you know, what happens, you could imagine that leading people to wanting to have their data, but not having it because they wrote it in, you know, wall gardens, like Twitter and so on. So maybe that's an opening to, you know, like prepare for that and say, like, okay, how can we actually make it easy for people like to, you know, to create their own data into the commons? So like, siphoning, bridging, importing and so on, that could be one direction. And it goes back to the interoperability, to the thinking tools. So maybe that's the common thread here. Yeah. So, you know, beyond free Jerry's brain, free everybody's brain. Yeah, free. I want to liberate everybody's brain. I mean, part of my beef is with how terrible intellectual property overprotection has become and how we're denuding the commons and how we don't know what we know and how we're unable to solve problems together, partly because we're not sharing the solutions other than in these different little separate nuggets that are not connected. And then the web is beautiful because it creates the beginnings of those connections, those links, those little filaments. That's lovely, but there's so much more to do. So this is what, I mean, of course, I think of things as I got a shape, but, you know, like I understand it for the commons, but like, if you could imagine, like I saw just today, like I think there's many of these going on about like fine tuning different models. If you could imagine, you know, like, like targeting, maybe we're targeting a few months, you know, a tool chain that lets you fine tune language models with like a Masi wiki or an Agora or any kind of like a knowledge commons, then maybe, you know, that could also make it so that then people start thinking about, okay, so first I get something like a commons and then I fine tune it essentially a repository and so on. So maybe it's like a sort of like a enticing for people to like move closer to wanting their data. My delivery. But, you know, I don't know, like, so essentially how can this help the revolution as a question. And I guess, you know, like problem solving and, you know, like sketching solutions. I asked Yadvidi how to run the revolution and he actually had pretty good ideas, yes. Generic one. So how do we run a pro commons revolution would be a great prop? Yeah. Yeah, I mean, for every day, yes, yeah, completely. So I guess we have discussed this email before, and it makes sense to discuss it in this context because of the leverage point and so on, but like, I guess we discussed it at times, like how can we build something together that has value for, you know, not taking real users or like for more users and so on. And like, I think that's where we have been working towards that, you know, but I guess I go back. I don't know, maybe we want to like give it a few more sessions to like, you know, think through about like the implications of GBT and so on, but then like try to shoot for something this year. Still like something we will ship, I don't know, together. That'd be cool. I have trouble mentally bounding that exercise, but I think we could have conversations to bound it and we could ask, should GBT have to bound it? Yes, the podcast, right? Yeah, yeah. Oh, and then there's so many places to talk about this and share out what we're figuring out. Right. It's cool. Okay. Any other topics or are we spent for the day? I've got a quick one, which is I moved, I haven't been posting to Twitter since December. And I won't start, but I guess I want to start posting to Twitter if you want to read the stuff that I would have posted to Twitter, but aren't anymore, go here. And right now, the place to do that is to Mastodon. And Mastodon is interesting and fun, and it's not quite a replacement for Twitter either. So I guess I'm thinking of, I don't know. The best solution I've got right now is either go to my Mastodon or go to my Listmonk instance. So Listmonk is a open-source, simple version of Substack, more or less. I mean, it's not nearly as complicated as Substack, but it's a newsletter publishing solution, email publishing. So the problem with it, actually, it's a really nice tool. The person who wrote it, it looks like he wrote it for his company, and then it's open-sourced, and it's all maintained, it looks like a decent system and stuff. The main problem with it is it was built to be kind of one-list instead of hosting multiple lists. So it's got a hack in it that will let you do multiple lists, but it doesn't do the archives in a multiple list format yet. All the archives are in one list. So anyway, that needs to get fixed. But if I'm not using Twitter, and I'm not using Facebook for philosophical reasons, Mastodon is part of the solution, but not all of it. And so I think another part of the solution is like a personal email list or something like that. I don't know. And for reasons that are sort of beyond my understanding, they're promoting Substack Notes as a Twitter replacement. Yeah, yeah, whatever. I got, I remember you're muted if you're trying to. Oh, to be clear, they weren't promoting it as a Twitter replacement. Elon decided it was a Twitter replacement. Oh. And now it has been positioned. No, it is. Because it's treating it that way. I mean, it's a different social model. It's a different network model. It's a different thing. Notes are fine. But again, like, none of these things are really a Twitter replacement because Twitter is not software, not really. Twitter is a system of moderation and community and interaction and brand value. And reach. You cannot, what? And reach. Yeah, and reach, which is not easily replicable. But I'm actually, I was surprised to hear how few users Twitter seems to have in relation to the major platforms. I was like, wait, wait, that's all the number of people they've got? Okay, so this is actually really interesting. I did a Twitter thread on this this morning. How did I land on that? I feel like I just hit and sunk a sub. Yeah, so it's, the thing about Twitter is that it is an amplification platform, which is to say, Twitter does not have many users and Twitter does not itself supply a great deal of traffic. But Twitter is the origin point for sharing that happens elsewhere. And so Twitter's influence is more as a routing tool than as a social network or a media platform in and of itself. There, I put a link to the thread because I did just this morning. The issues that is of replication, right? The reason that I have this in the thread for, right? Like the joke in Silicon Valley is anyone could build Twitter in a weekend. And like from a technical perspective, sure, but that's not what Twitter is. And the question is like, is it going to be, I won't just repeat the thread, but basically like what Elad is doing is a mistake in terms of understanding what Twitter is, right? He's treating it like it's medium or he's treating like it's suspect and he's treating those other platforms like their competition. But it's the exact opposite, right? If Twitter retains itself, which seems more and more unlikely, Twitter is an amplifier for those platforms but is amplified in return. I say like Twitter is not a social network. It is an event space for dropping new pieces of content. So I will say that it is a social network, but it's also like, yeah, I will say it's like a connector. It's like a connector or something like, it's like a RSS feed, it replace RSS feeds to some extent for most people. So to some extent it's like a bit of a parasite of the internet in the sense that it's an issue. Yeah, what happened when Twitter started was I realized that my attempts to use Google Reader to read stiff stuff in RSS were terrible because I was flooded with way too many articles that were way too long. I would read three of them, give up and move on and feel incredibly frustrated with the experience. And then on Twitter, because everybody was forced to compress into basically Haiku, like I was like, oh, now I know I wanna go read a long read, but this is a much better way to get through a lot of things because there's a rich, it's like you landed in the plankton stream in the ocean. Right, right. And it's also like a bit of like, it forces, yeah, it forces summarization, which now you would say maybe actually generated, but you know what I mean. So on the Twitter exit bots, I think, yeah, what it's all in this, if it's true, like what Adam says, right, it's like a mast on itself won't replace it, not only because it doesn't have the same connectivity, but also because it's not sufficient. It's also going on the internet, but like, I don't know if you've seen Lemmy, Peter, which is, it's like Reddit, but on activity path. It just shows like, you know, you can have like, I haven't seen it. Yeah, you have, okay. I haven't seen that one. Lemmy, yeah, I put it in the notes. I know of the run one. And of course it has a thing that you can still see it from mast on, but just sort of see the headers or the titles. And if you go to Lemmy, you can actually get through and see the whole content and the threads and so on. So, so I guess, you know, that could be, it could be nice if whatever your exit path is, I guess it's activity path days just because it will give you like the network, right? Agreed. That makes a lot of sense. Is Lemmy what delicious could have been become? I have, well, it's like decentralized delicious. Right, which is good. Yeah. There's also decentralized good reads, by the way. You know, if you missed the path, I guess it's like a universe where good reads didn't go to hell because it was both by Amazon, I guess, and so on. Maybe it's a little. In the alternate universe where it's still free and alive. Oh my God, I've been doing alternate history stuff with chat GPT. Completely. If we lose you into one of those dimensions, I am going to get sad and miss you. You know, I going back to the bootcamp or like I don't know. Or you imagine like for the podcast, I'm just restored me now, so it's still me. But it's like, you know, we all generate things and we are like a school and we share it. Maybe like a format in which like, you know, N people come and they show one thing. It's like when you play your YouTube video in like a YouTube party, you know, when you, everybody plays one. Like, you know, show the best result you got from a GPT last week. I like it. It could do an exquisite mega corpse performance art thing where everybody contributes something generated between them and generative AI. Except their, except their sequence together. Right. Right, right. And Aram, you had something I think you were going to jump in with if you still remember it when I was asking if we were ready to wrap the call. Oh. Which may or may not have been the thread you just pointed us to. No, it would have been my thread. Unless somebody brings it up, I don't promote Twitter stuff anymore. It's just my brain works better in that format for whatever reason, writing in that format. In Twitter format, you mean? Yeah, yeah. Something about like the specific level of the limitation that, you know, like what you call it. Mastodon has too many characters and it doesn't work. Queer has just the right number of characters that I can refine paragraphs to ideas, to focus paragraphs, and then create threads in such a way that it really works and I can't replicate it anywhere else. So I haven't quit entirely. But you go around your instant with the same limits. Yeah, yeah, I could build a Mastodon iteration or something like that. I actually have a bunch of home hosting that I'm setting up because I don't want to pay for it. That I'm probably gonna do myself. Well, the Mastodon hosting gets expensive. It is for Mastodon. Mastodon hosting gets expensive pretty fast, even a solo, even a solo one. I have a nice heavy activity. Sorry, I have a nice heavy, whatever the latest Raspberry Pi version. It has a web server. I stock a 300, what is it, 500, 400 gig USB key in there. It's ready to go. I just need to do the work of setting up the web server the way I want to and map some dynamic IPs to domain addresses and it's good to go. A long time. Actually, really I'm interested in Blue Sky. I saw you mentioned that. Yeah, I got two invites today. So I think they're opening. Oh, it's running. Yeah. Okay, I was not aware it was actually running. I know, yeah. I wrote vaporwave at some point. Vaporwave is a general. At some point, in the sense when Twitter was, I guess I was suspicious of it, but then I saw that it was more concrete and now I'm invites are making the rounds. Yeah, I was initially a little dismissive of it, but they hired all of the engineers who were working on hyper and I really loved hyper. So, and I saw when I reviewed the protocol, there's a bunch of hyper stuff that's definitely just in there. Yeah. Paul Frazí is amazing. What, which? Paul Frazí. Oh yeah, Paul. Yeah, he's so great. Everything he does is fantastic. Yeah, yeah. The giga data was like, yeah. Exactly. That's what I started paying attention. So, yeah, I think that one could be really cool. I don't know where it's gonna go exactly, but like, it looks like it's directionally correct. Right. Yeah. I guess I went back also, Peter, if you are looking for that term of the steel, there is a chance, yeah, Blue Sky will work. So, yeah, yeah, I'm interested in Blue Sky in my work. I am into the masts of steel, I guess, because it seems like the best activity part, I mean. You know, it's actually really good. And it's just a different thing. I guess, you know, I've got a semi-disdermalative, a second cousin who I discovered via genealogy research in the late in life. You know, and I'm like, okay, she used to, you know, so I was publishing my hot air thing, and she would love this, but, you know, she's not gonna see it on Maston. She's just not ever going to be the Maston person. So, what do I do about that, you know? It's funny, I feel the, I guess, I know why now. I'm the same way with Facebook. I'd stay off of Facebook, basically, and LinkedIn, actually. But, Peter, I actually knew and loved and spent a lot of time there and got a lot of value out of it. And unlike Facebook or LinkedIn, Twitter did feel like home, and now I miss it, you know? It's not the thing anymore. Yeah, it's important, at least. I've been, there's a really cool project for archiving, so at least my archives from Twitter I've pulled on. My needs to be updated. I did it right when Twitter was initially done, so it's only up to the end of last year, but. Which is, you know, the end of last year, but. Which, is it a service or? It's a tweetback, which is an 11T-based Twitter archiving tool. Maybe I'll make a massive wiki one. Yeah. I have two quick things. I know we're probably trying to wrap up. One of my, so my, one of my observations after playing with hot air is that we're, so a thing, a reflex that we have from the olden days is like, oh, there's a new tool that makes things faster. So you make, you use a new tool to make one thing faster and you go, wow, that was so amazing. Look, it only took me, you know, 20 minutes, so it would've taken me 20 hours or 40 hours before. So the step beyond that, a few steps beyond that, what it's going to be is, look, I can make 200 of them in a day. So write, you know, write that same story, but with different people, with different locations, write the young adult version, write the kids' version, write, you know, the romance version, write the science fiction version, write the, you know, and so one of the things is creators are going to be able to cycle through a bunch more, a bunch more, you know, maybe I want to write a historical alternate history novel or something like that, you know, you can go to TedGBT and say, you know, give me 20 scenarios of alternate history and it says, well, there's the one where the US lost World War II, there's the one where the US lost the Revolutionary War, there's the one where France prevailed over Britain, there's the one where Holland didn't become a superpower, you know, et cetera, et cetera. So you do like 20 of those and then you tell it to do a script treatment or a novel treatment or a story treatment for each of the five that you like best, and you go, well, I don't like any of these, but this one's fun and then you blow that out, right? Just like going over and over and over and doing what used to, you know, you used to, in the olden days, we used to write novels once because it took so bloody long. Now you're going to write it a hundred times and, you know, a thousand different variations, you'll never finish most of them, but you might finish 20 of them and you might publish 20 of them. Okay, so this is actually a really interesting piece of literary history for this. Have you ever seen Plotto? No. Okay, so this was a guy who had to write Dime Store novels. Book of all plots. To make a little, and so he came up with the master book of all plots. Right. And each one is a plot seed that interconnects. It's all predates the internet here, right? You've got plot seeds that inter, predates computers, actually, that interconnect with different characters and different plot points. And it's exactly that, right? It's a formula for children very quickly. And it's fascinating as an artifact, but obviously it's not a sustainable career. I mean, it was then he pulled it off, but nowadays I don't think it works quite as well. Pretty cool. 1928. Yeah, it's a fascinating little artifact. I love, as somebody who's always been very interested in narrative structure, it's like a very fascinating thing to do. I have to hop to another meeting, but check that out. Yeah, thank you so much. I'll talk to you later. Yeah, thanks, Aram. Love that, wow. And I've got a section in my brain about work plots and there's a couple of different models for how many work plots there are. Right, like the classification system for full tales and so on, right? Yeah, yeah. Excuse me. Very nice. Actually, I've got a bunch of stuff on work plots. I'll share the link in the description. The other thing I was gonna mention was, Aram is gone, unfortunately, but speaking of Raspberry Pi, there's a new fun Raspberry Pi thing out, the RP2040. And then people put together, so in this one, this is the Raspberry Pi here, right here. And then this is a little board that somebody put together that has a tiny little bed board thing on it and a tiny little display. And so it seems like a really fun thing to play with. It's getting somewhere. Yeah, it's getting somewhere. So one of the places, there's like Raspberry Pi worked with like five vendors, Adafruit and three or four others. One of them is Pimeroni and I like theirs the best. And then there's, if you're in the U.S., it's a little bit tricky. It's better to order in the U.S. rather than from overseas, from the U.K. or whatever. So pi shop.us.us is actually a real company. I guess that one of the cool things is that there's a W version of it with built-in Wi-Fi. So on a little stick, it's ready to go. Wow. And talk to Michael Bison and all that kind of stuff. Super cool, thank you. My brain is full. For today. Yeah, for today. We'll have to empty it and come back, exactly. Just flush the buffers. Which is what kids do these days when they cramp for tests, right? Old days, no. Wash the buffers, turn to the next subject, don't remember a thing. In fact, Mondays, at Siba, I do a mind-meld and there's a young guy who's been joining the conversations and he was like, yeah, it turned out in school, I learned that I could memorize stuff. And so when he got a test, he would basically, from memory, write down the test equations, the test problem. He would just write it down on the margin and then he would look at that and adapt it to solve the current problem. And his teacher was like, I've never seen anything like this. What are you doing? Where did you get the copy of the test exam? And he's like, I just memorized it. So it's there. I'm like, okay, that's really interesting. But, oh, and the reason I told that story is that he said, and the idea that I would be learning how things work or internalizing the models of physics or chemistry or whatever, none of that happened. Like he didn't, he wasn't understanding how the thing worked. He just learned this memory trick that worked for him to pass tests and that was good. Yeah. It was really interesting. Very much to the side of the idea as well. Gentlemen, thank you very much. Yeah, thank you so much. I'll stop recording before I forget. Stop recording. Perfect. Thank you.