 The recording is on. There we go. This is the Fellowship of the Link call for one of the December 20, 2023. We're getting very close to the end of the year. Very close. Go ahead with your proposition. No, good time for planning. And like, yes, I guess for Cecil Reels as well as we were discussing before the recording started. So we have some notes for anybody looking at this in posterity. So I will, I think we can represent them if I can find the right tab. Yes. So this tab is a Fellowship of the Link in the Aura and the notes are in the Aura proper. You can also go to docanagra.org slash a Fellowship of the Link, but yes, in any case. And yes, we discussed Cecil Reels and had an interesting example and the potential of like a storyboard generator, maybe. Yeah. That's the main step. Yeah, on topics, I wanted to maybe actually discuss, like I show you a quick demo. Please. Of something I've been working on. That sounds awesome. As part of the similar adventure, which is something I've been doing after work and on weekends and so on, this month as you could expect. I think I mentioned it like the previous time I was, last time I had to skip it, the previous one, is the idea is just like I've been for cold, a bit of cold, but like a chiller. So essentially you just call something like every day. Cool, sounds great. I love challenges like that. And any other topics or like things for agenda? Michael, any topics that have come to your mind that are fellowship with the Linky? You're muted, I think. I mean, nothing specific. I've got some Linky thoughts about objects that I'm working on, archiving, personal archiving. And then we move to that. Cool. I'm interested in the topic. I'll call it thinking like a Neo book. I was firing. He's running around to their feet, little baby. She's off the charts. He's like a Neo. Michael, we're hearing the video, yeah. Oh, I think it was coming from you, Jerry. No, no, no. It was me. It was me. I don't know how it got unmuted. I had been watching April's Superreal. Cool, this isn't real, yes. Cool. Good stuff. A lot of cuts, a lot of cuts. It's very, like they're really good editors and the pacing like a minute and 15 seconds in you have to assume that most people are not going to go any further. So the whole package of why you're trying to sell this speaker has to show up in that first minute and a quarter or minute. And then after that, there's like a shift of tone. And then there's a reason to kind of stay in because, oh, oh, I want to know about this too. And it's very weird how much attention you have to manage for five and a half minutes of video. Yeah. So as I was telling you, I think before you got on, there's also a two minute version which cuts out a bunch of the story, but it's like for the people who don't want that much, there's also a version with subtitles because some people will need subtitles, blah, blah, blah. Mm-hmm, good stuff. So you want to do your demo, Flancian? Sure. I mean, I'm doing the demo, yes. I can do the demo in it and then actually ties into the Neobook aspect, maybe. Fabulous. So, yeah. So I just want to do like a show. Let me share this down. Actually, it's Neobook. I don't know if you can see this. Awesome. Yes. Well, we're seeing your Gora. Yes. And this is a Neobook. This is the node on Neobook. Oh, cool. On how it looks nowadays. This is the beta alpha, actually, of the Gora server. This is not yet deployed to the Gora link, but it will be soon. I'll be working on this as part of the similar adventure. Some UX simplification, hopefully, to try to make it a bit clear, are also some features. I want to show you a few. So, we have a link to the chat for the link to the particular spot in Gora. This is a Neobook. Yeah. Okay. In general, the node, for visiting node, whatever, on an Gora, you go to another Gora slash whatever, and you can have spaces, pluses, dashes. You will try to parse while you give it. Cool. Yeah. So, here's an example of Neobook. And so, I go just quickly over these. So, up on the top, you can actually search on on different services. So, in case you don't find what you were looking for in the Gora, or if you want to move on after that to do further research, of course, you know, on Google, which I use, and like, you know, as you know, as we may have nowadays, being all others, but in Alia, it's a nice arranging for the indie web and blogs and so on, in general, ready, we also are ready, I think. And then some, you know, X event, took X for like, you know, people who are like, seen Twitter, which makes sense, and to find a format stone. So, essentially, this is very simple stuff. You just forward to the research. Then, the Wikipedia article that we have, that Wikipedia returned for this query, Neobook. In this case, it's funny because it's list of fields considered the worst. How did that happen? I have no idea. Usually, it's this relevant. Actually, if it's very relevant because it has the same title as what you search, it will just automatically be embedded. There might be a movie with the word Neobooks in it. Yeah, and maybe sometimes it's just like the mystery of the Wikipedia search, but I have to say that when it's not relevant, it's funny, which I like. That's good. That's cool. Like, in this case, maybe. Then, a new feature, which is generate the BI. So, the algorithm will actually forward the query with a prompt to generate the BI service. In this case, I'm using Mistral. I don't know if you mean Mistral. Open source, open weights, they're just in stuff coming from France. Supposedly, a quite recently efficient model called Mistral. It has performance around GPT 3.5, and it's open source, so I really like it. So, what this will do is it will generate text and links. So, for example, these links, interactive e-books, video publishing, multimedia content, and interactive elements, these are things Mistral, in this Agra Assistant role, thought could be interesting links related to Neobook. So, you can just click through, and it will, of course, go to the Agra location, by the app publishing. So, wait, are these things you already had in the Agra, or are these new Agra pages that Mistral just created for you? These are just new Agra pages that it revved up. It doesn't know they are actually there. What usually ends up happening in my testing so far is that a lot of these make a lot of sense. So, even when there's no other content, there's a Wikipedia page that is very... So, just so I understand a little better. Interactive e-books is an internal link for the Agra. Did it know enough to put double brackets around that, or did you put that in afterward? Did it? No, yeah, I did. Well, the prompt says, I can show the prompt if you want. Oh, cool. So, the prompt says, hey, put double brackets around the thing, and then go create the page, or the node? No, it doesn't create the page, it just creates a link, and then it's up for us humans, in further, for example, if you visit. Yeah. So, I think the cool happen, for example, is... Wow, that's cool. Yeah, I found this a lot of fun, actually. Yeah. I can show the prompt in a bit. I think the prompt essentially says, you're a friendly assistant in a knowledge base called the Agra, and the user is asking about an entity, which can be a thing, a concept, and when responding, please put a few entities that are interesting in those brackets, a weak link, and then it just works, because the Agra passes that as essentially as links. So, this will be interactive e-books. The Wikipedia article is just e-book here. Not about time to read e-book. Usually, I have read Wikipedia a lot more, which I think is a good thing. And then, if you want to know what it can dream up on interactive e-books, you wait a few seconds, it should be done then, and voila, you have like a generated node. Now, you're storing this in the Agra, so whatever it auto-generates right now is now permanently part of the Agra. Right, so good comment. Not right now, because it's a prototype, but that's of course the goal, the idea. So, for now, it says, to save a generation, please copy and paste into this document store. So, this I've been doing. So, I mean, this is the manual, this is the proof of work, where some human copied and pasted, this is not the idea is to have a ping button or a save to our button. Yeah. So, I would love for people to still select what they really liked. Yeah. So, if you think about, yeah. Very brief aside, I've turned on Google Workplace Labs, and so, when I open a spreadsheet, I get this little thing, hey, can I help you build the spreadsheet? Which is pretty fantastic, it's pretty awesome. Yeah, it's very good. And Michael, I don't know if you've seen this, have you seen it? You're muted. I'm not sure if I have. Okay, so there's a whole new window that shows up on the right hand side, and it says, I need a marketing database for a publication or something like that. And then it will generate columns, name them, put in fake data to show you what it was like, if there might be pick lists in there. And then at the bottom it says, do you want to just discard this or accept it? And if you accept it, that's the start of your spreadsheet. It's really cool. I know that people are working on it. We are honestly quite excited about it, because putting my hat on, disclaimer, Google is my employer, I think it's pretty cool, and Google is in a position to focus on integration. Yes. And it's responsible for integration, useful for integration. And yeah, I think the flows for like trying out things and then this idea of like, okay, bring it in, like a slides integration is very, this is where I assume, I don't know how much of this has launched, but like it's very nice. Cool, so keep going, sorry. I just wanted to say that I totally understand that you would want to have a little option to discard or approve. Exactly. So here, I just copy and pasted, the nice thing is that you can choose whatever you want. The links get maintained, because again, there are weekly links in the square brackets. So copying and pasting works, next time you, within 30 seconds, this will be in the hour. Cool. So going down, going back to Neobook. Yeah, for now, I have to say yes, the actual generations are cached. So if you're refreshed, you will get the same one and that could be also another mechanism in which you can save generations. Hey, Adam. Hey, Adam. Hello. Lancen is doing a demo of sort of a Neobooks and generative AI using mixed trial trial inside of his Agra. Yeah. Okay. It's really cool. Yeah. You know mixed trial, the mixture of experts? No. Yeah. So mixed trial AI, surely French company doing open source, open weights AI, and brought up mixed trial, which is a mixture of experts approach, an ensemble approach where like the cost of inference is significantly lower than the performance yields, which is around 3.5, a GPT 3.5, I mean. And yeah, I prototype some AI integration in the Agra. So for example, this is a content generated for Neobook. And you know, the system prompt says, you're a system in a knowledge graph called Agra, please link interesting entities. And I found that, you know, it's quite fun how it generates links essentially. So very much to the topic and to the topic of the meeting. Super cool. Yeah. Yes. Another, so for example, I go into the link, this is an example of what you get just out of the system problem, it's quite short. So you know, it says like, it doesn't know if it's a link currently, it isn't getting extra information from the Agra. Oh man, our marketing is failing. But that's the thing. If we delay the generation until the, we actually have gotten, you know, all the content that is in the Agra, which is the next step, it could actually read this and summarize. As a corpus. Yes, exactly. Yeah. So, but it's still, you know, I liked how I wanted to know how it was, it managed to like, you know, in the context of the Agra and so on, it says I'm here to facilitate the operation of the Agra and provide insights. It's sort of like guest what it is. And it's sort of like as we write. Wow. First a group of individuals or entities who are closely connected to share interests, collaboration and creation within the Knowledge Graph. And all of these, of course, for example, like Knowledge Graphs, so another, it's really cool. Of course, you get Knowledge Graph immediately because it matches well. And then you just can, again, generate content and optionally save it. You know, seven to eight seconds, but you know, open source and good price. Yep. So, I'll just explain. And in the Agra, it's sort of like assumes that the Agra is about like a classic Athens entities, also, you know, I guess, I guess. In any case, going back to new books and finishing the demo, like I said, yes, this is generated in the current layout. So you get like, I mean, this could be ranked, but right now it's always like search first, Wikipedia, Genai, Stoas, which are just like essentially, like, well, hello again, like I'm looking right now. This is a good time to mention that I found Fama Talk, which is also French, but you know, I swear the French government is not paying me, but like, it's also great. And I will actually suggest that we move the calls to these because it is deep seed, but it's run completely open. So no logins and like not of all these other things we are having suffering. No kidding. Yeah, I'd really recommend it. So just a suggestion that we can boldly make. Cool. And then in Neobook, here's what I guess more specific about Neobooks. This is essentially, there's no actual node in the Agra about Neobook in the sense that I didn't, nobody else started a node called Neobooks, Neobook MD in this hour. I think there may be one in the link, in the other link, but there are a flange on. But still we have all this concept because we have been writing in future link in the nodes and every time we mention Neobook, we are essentially writing at a distance. So this is essentially how without even creating the Neobook page, you can see all the content around Neo. Just because it's a recurring topic that's being parsed out. Exactly, yeah. I mean, essentially, because it has been linked. Yeah. If you don't link it, it will only show in text search, but if you link it, it will actually show about it at a distance. And yes, and then I made some changes in how fussy we get on which other content we surface and I like it because it's a bit promiscuous, which sounds good to me actually. Yeah. For example, I call write a Neobook or three, my relate is already here. And in this recursive approach, you can see all these and do the same thing, like for example, like JNAI for call write a Neo, I'm looking forward, but it comes up. Call write a Neo involves collaboratively creating an interactive and multimedia-rich digital book using the Neo authoring tool. That seems like a fair guess. And Neo book authoring tool, it proposes, that seems like an interesting note to write, maybe. So I guess I already see how just distribution of links and generation, even at this basic state, I found I have had this only for one day, essentially, and I already missed it when I don't have it. That's super interesting. It's interesting to that. So this is generated AI text that is making up what it thinks a Neo book might be, but it's really easy to start mixing that up with text that we wrote about trying to describe a Neo book. And it's like hard to label what is an inference by the engine that might be hallucinating and what is something that some author intentionally wants to write where it's been very carefully crafted to mean what it says, right? And the boundary between those two things is boring quickly. Exactly. And this is where I think having one interface where you can both generate and as a community, select, repeat, and edit is where, and maybe have some traceability, right? If people choose to do so, it has potentially probably will be beneficial. This is where I, even though right now, it's clunky to say, copy and paste into the document, I've sort of been doing it because first of all, it's easy and it maintains all the links. But as you do that, I would add a header or a footer that says, by the way, this page is auto-generated text. Yeah, I think it would be great if, you know what I'm gonna do, I think add a copy button before I add a save button and have it have that header. Yep, yep, yep. The easiest one actually here. Auto insert the header so nobody forgets, so it's consistent, sounds great. Right, exactly. And you know, to which extent? I mean, this also going to, this already goes, by the way, if you write to Astoa, it actually automatically gets to a git repository. So then you could just like take the git file and edit us. Cool, cool, cool. If you want. And that's the demo. That is awesome. Congratulations, really, really cool. No, thank you. I mean, it's been fun. It's just fun stuff. And of course, I would love to add a bar and a LGBT. Right now, this is all running on my API key. And I want to see, I mean, how long it lasts. It's actually not very expensive, mister. And the algorithm is not super popular, as you know. So then I will just bankroll the operation. Great, sugar daddy. Yeah, yeah. And then the idea is will be to have client-side generations for this kind of tool. So when you go to say, instead of the API keys for the services you have, and then you get the UI for those services, and the way it's also completely like client-side and you can choose precisely how to share it. Right, cool. Aram, Michael, questions, comments, answers, suggestions, jokes, recipes, wishes. I think it's cool. I have like less interest in AI-generated text in these spaces, I think, than most people who are participating here do. But I like how you used it and as long as there's a warning on it that says, you know, this is AI-generated text really clearly, I don't think it's a problem. No, thank you, thank you for that. Yeah, I think it makes sense to, I mean, and the idea of the algorithm is human first, of course. It's like a, you know, it's a kind of place where I, I hope the idea is like, maybe there's still a mention of users, actually users on the top, and I want to change it by replacing by people, right? This idea of like, you know, the use of users as a, so yeah, I really wanted to be like very people first. I think as an assistant though, I really like the idea of like the, well, the cyborg approach where you are like, just being clear about what is AI, what is human, and I'm just focusing on like, you know, improving how we collaborate, essentially. But yeah, very fair. So maybe I won't add the auto-save. It's better if the humans always take action and like take ownership maybe of the environment. Yeah, I think that would be better if people like, were forced to review it to some extent, you know? There's just like a lot of potential there for errors to occur. Yeah, completely. Yeah, and once you start, it's like what you're seeing as well, once you start like, having generated content in the corpus, and, but this is where like, the idea, I mean, this is not as optional every hour and since I is on the pragmatic level, but you know, each hour defines a corpus, just because it's a collection of repositories, right? It's an issue, no? So, and the idea has been to like, you know, you know, explore the potential of that. How communities can define corpore, and so on, but yeah. We're learning this. Sorry, go ahead. One question on the AI generated copy of the auto-save is one thing that always slightly annoys me with AI generated content that I'm dealing with. You know, if I promise chat GPT to create something, I've, annoying is too strong word. I'm thinking in a situation like yours, this has been generated, and you have the option to save it or not, is it editable while you're looking at it or does it need to be taken out and pasted into some other context to be edited and brought back in? Because I've often been struck that, you know, this is a good start. I really want to just edit this on the fly right away. Yeah. Yes. Here I know that at least some of the areas I've seen in the, actually why I have this notion of mark text and say like replace it or like, you know, incorporate the edit operations into the generation flow or user journey. Yeah. I mean, here, I mean, of course, here is for now is very, very simple. So, you know, by efficient, you have to copy and paste it. I think ideally, yes, it will have like, when you save, it will open it in edit mode immediately. So then you can start changing it. And, you know, because it's all get back in the end, we show how to stability and then, you know, then we end up with like the shape of tissues, eventually where you are like, you started with something generated and you edited enough. And in the end, I think these conversations lead into the aura concept. You know, like as per Benjamin, I think, you know, yeah, it's what we choose to see, you know, whether it's mostly human or mostly, you know, or mostly I, but yes, I will try to experiment with like the editing aspects and I just really did. Cool. Well, what's on your wish list? Besides that. Flancy, and what's on your wish list? Where do you want to? Oh, my wish list. Where do you want to steer this? So you mean, sorry, today's conversation or the experiment? The experiment. Oh, yeah. I mean, well, going back to the new book aspect. I mean, so on the I direction, it's like the idea is to have like the idea, improve the idea of a router of prompts. Right. So this idea that, you know, add value in having a center, a one place decentralized, not centralized, where like community can say, we can easily try different providers and pick the best and so on. Which is not something usually corporations do. Very willingly to be on the, yeah, yeah, in the comments. And on the direction of like UX. So yeah, going back to new book, I was thinking of like just taking, because I think it has previously on the issue of transclusion, how to transglue and then generate, right, into, I say, a PDF or like a document. So what I was thinking is like, you know, with pushes, we are quite close in the sense that, you know, I could, one incremental step will be to add like a, you know, export to PDF option in a node or any initiation here. And once you have that, you know, essentially rendering all the pushes one after the other will essentially yield transclusion. So essentially you could, to build a new book, you can write anywhere in a new repository. And then you essentially have to say where you want to send it to. So for example, you write a chapter and you think, well, this could be propelled in these and these new books if they existed. So then instead of having to go to each new book and create it or like, you know, and add the link to the place you were coming from, you only need to do one action, which is to say push, push, push. So it sort of like flips it on its head. So you're basically publishing into a variety of books, some content that would be relevant, which is intriguing. I mean, it then needs to be integrated into the books unless it's just the list item. Right. If it's at least item, this will work really well for pattern languages. Yeah. So this is why I guess my default proposal would be like the one of the new books we work on, maybe the first one we work on, I don't know, is a pattern language. For example, a pattern language, right? Because it's same shape as a list of patterns. And you know, of course, like the traditional pattern languages are still order and some of them like a pattern language, they go from large scale to very bottom. Right. Exactly. There's more. Yep. And then it does say, but that's an operation that can apply to the list. So you know, like, and also a lot of pattern languages are more or less, they don't have a notion of size. So then essentially that will be, you know, even relate or, you know, whichever repositories we're using, we say in the out of the link, for example, we say, you know, be right things. And then we are like, this is in Disney or in Disney book, just by saying for example, like book title, like here, right? You say book title between, so that's a, between you link it. And then essentially what you write there, we will just like be, I don't think we can switch out the distance. And, you know, I can work on the skinning aspects of how it renders and so on. But I think we could get to a generator that is even too agnostic, you know, the idea without a lot of work. Yeah, and in my experience, I use pushes like these. Well, now it's essentially I'm making it like, so everything pushes. If you link it, you'll see it sort of like like a very strong reference. But, you know, that can be tuned essentially. Yeah, so that's just the idea, I guess on the new book's direction. Cool. Love it. Any questions, thoughts, anyone else? Let me know if you want any features, essentially, or maybe we can discuss the, well, I'm happy to discuss a new book, like next steps or if we want to discuss that, but I don't know. Happy to. Michael, did you want to go to your topic? Sure. So I'm thinking a lot lately about and this is something I've brought up in. And your tone, your gain is really low somehow. You're not very loud. Let me see if I can do anything. Yeah. I mean, we can hear you. It's just that we're all leaning in a little bit. All right. Pasadena, do you better? I'll just try and talk louder. Cool. I'm working on a project about kind of database thing at its broadest. It involves helping people to inventory and deal with the artifacts in their lives, the physical objects that they've got. And, as Jerry, as you know, it grows out of my archival propensity. Cool. The need that I've seen is you have people, particularly, you know, aging baby boomers and older who have a lot of stuff, have a lot of stories, have sometimes, you know, attachments to things. Sometimes, you know, you have the dynamic between a parent and child where, you know, the child thinks the parent is a hoarder or some such thing. But, you know, sometimes you just have people who like are fascinated by stuff and books and, you know, things and have accumulated a lot of them. And the thing I think the archivaly minded all wished for is that, oh, I wish I had a complete inventory of everything and it was easy to find and I could also know the value of the stuff that I have and be able to part with it without the pressure of like, oh, I'm putting it on sale right now on eBay and I have to take the highest good that somebody's got opposed to sitting there in my inventory and it's either available for sale or it's not available for sale. And, you know, if somebody makes me an offer on it in 10 months, that's fine. I'm not urgently trying to sell it. And it also speaks to the idea of people who are downsizing and, you know, might want to have all their stuff inventory taken away, you know, put the warehouse sold off. Now, the more interesting wrinkle about it is the knowledge that is attached to these things and how people feel about what they know about the objects they became related. Whether they, oh, I think I would just seem to have that. Michael, I went and I just went and just, you know, typed in makemeanoffer.com and it's very funny because it looks like it's a domain that somebody's bought but you could buy it from them. That's funny. And it would be a really interesting service to have and then you could have a different aspect to it, which is the home inventory thing. And I've got a bunch of things I'd love to pass to you about that, but the nexus of these things is really interesting. That's interesting. Yeah. Yeah, there used to be a function on, I think they got rid of it as they got further into bed with the real estate industry, but there used to be an aspect on Zillow where you could say, I'm not moving my money and my house isn't for sale, but if you made me this offer, I would move. Right. And so, you know, it was an outlandish price, but some people were more serious about it than others, but it's a great dynamic to create. I think that's a great idea for a site, actually. Like, for example, like, you know, what people could like actually ask for a lot of money for the personal information, which people like corporations are like amassing already. And they could say, actually, this is worth like 10 million for me. Right, right. Or, you know, or let Elon Musk have dinner with you. Like 10 million. Yeah, sorry. This is how demand curves are built. Yeah, yeah. Now, then a price is out there. Right, it's true. It's like, you know, it's like that famous, is it George Bernard Shaw? I think it's a George Bernard Shaw line. Something like that. It's like, madam, would you sleep with me for one million pounds? Oh, yeah. And... We're martyring. Well, I suppose I would. How about, you know, 10 pounds? What do you take me for? Well, we're going to establish that. We're just talking about the price. Yeah, yeah. I like it. Although I think we need a new phrasing because it feels a bit misogynistic. Yes, I agree. I agree. But, you know, but I mean, to triangulate prices on just about anything, to have, yeah, to have that, that the passive marketplace is what I think is the key thing I'm trying to create here. Yeah. And also the aspect that is not a marketplace, but a place of interest that you go not shopping, but just because you're interested in a different, you know, in a particular area of stuff. And so it like, it kind of flips the notion of boarding on its head, where, you know, people can say, oh, you know, I'm creating something valuable. And the fact that I've collected this thing and can tell you something about it, and can have its digital twin be available to you is me doing something of value. And maybe that's enough for me. And maybe I can partner with the physical object now that I've created this digital document. So I think there's a lot of potential for people who are getting rid of stuff to do it without that exasperation of, oh, but what if it's valuable? Or, oh, but, you know, this, you know, I have the sentimental attachment to this. Oh, I know the story and the story will be gone. Anyway, so I'm like, you know, working on business plans around this. You could get VC funding for this. I think could totally be doable, yeah. I mean, I have an allergic reaction to VC funding, but. Good, good. That's a very good allergic reaction. Yeah. Self-preserving allergic reaction. But I mean, my allergic reaction to VC funding is mostly around where, you know, I've been dealing with it in relation to media and attention-mongering, where I think it's a really awful force that has created bad things for us in terms of information flow. But with goods, maybe not, you know, maybe not as bad. I hear ducks, what is that? Yeah, anyway, I guess, you know, I'm interested to throw it out here just for people's reactions to it and relationships to things and relationships to their parents' things and particularly for taking relationships to objects out of the world of, oh, this is something that you deal with in a crisis at the end of life, like in a state sale to, this is something that as a healthy process of leaving my own legacy and, you know, benefiting the world with my physical and intellectual collection, I engage in throughout my life and, you know, put my curation out there. Interesting. So a couple of different things. One is like the bookcase right behind you, Michael. I've long wanted, and a couple of years ago I was trying to play with, you should be able to take a picture of the spines of a series of books. Yes, exactly. And have them all go straight into delicious monster or library thing or whatever. I want to say exactly the same thing. We have a video, ideally. That should be a done thing. Don't care if it's stills or a video, but even just panning across would be great. Boom. Think talk to Wushans. And then second, like now with barcodes, between barcodes and object IDs and all that kind of stuff, like you could really absorb a lot of things very quickly and it would be kind of fun to do. Yeah, there is, I mean, I'm forgetting the name of the, Collectors with a Z has an app that I've used for some of my books. I definitely use it for my comics where with the books, it just scans the barcode or you can type at the end. Yep. And it populates the thing. Awesome. Then in 2018, I had an idea that I put the URL for in the chat, the joy line. I'll give you the TLDR on the idea because I was working with an Australian insurance company and I realized that insurance operates below the joy line, which means there's some line of happiness or satisfaction that you're at right now. And for insurance, you're paying a company money, you don't wanna pay out for an incident, to avoid an incident that you hope never happens. And if that incident happens, their job is to bring you back up to that previous line of satisfaction. And that's all. And you know that there's probably fine print somewhere that says they're gonna find the reason not to make you that happy, but their job is just to bring you back up to that joy line. And the idea was, April and I were both advisors to a little company called Trove that was about micro ensuring things like your downhill mountain bike or your skis or whatever else, and then paying insurance only when you were using these things. That was their play. It ended up, the company ended up pivoting and then getting sort of bought. But that company had a reason to have an inventory of some of your items. And so I started thinking, what would it be like if they knew a lot of your things? And one of my propositions to them was, hey, a message could come from Trove that says, hey, I notice we know because you've let us know, but we know that you haven't ridden your downhill bike in two years. We've discovered that there's a couple of downhill bike clubs you could join that are near you. Would you like an intro or just a link? Would you like to put your bike into the sharing economy? We've already pre-populated a post and we can put it directly into one of the sharing economy sites. Would you like to sell it? We've pre-populated a post. You could put it on Craigslist or somewhere else. Or here's a couple of stores that take trade-ins that are near you. And here's roughly what your bike is probably worth right now. And we know that if you sold this, you'd be paying us less insurance, but we think that you having less stuff is better for you and your life. Anyway, that was a whole narrative that I came up with around this joy line idea. But for me, reframing insurance as being able to operate above the joy line was really an interesting thought experiment. Yeah, that's cool. So that's my thoughts. I like the notion of integration here because I do believe that... So the question is how valuable it will be to people, Michael, the data, having the data that they will get through a service or be able to manage through a service and have it be integrated with different providers, like in the commons, ideally, of course, but in general, like in the economy, like in Jerry's example, this idea of cross-posting and so on, and finding opportunities. That's very cool. I also thought, Jerry, also the book aspect, in particular, I think the book aspect is sort of like, with Goodreads being a shitified, I think it's the current term, by Amazon. And of course, we have Bookworm and all these failure-based tools that are coming. For example, like getting a prototype app that just posts like book collections from photo to failure, so from photo to like the Goodreads alternatives. That could be a very nice first thing to build here. I think Michael, it's essentially like starting with books. I will plus one that from Jerry, I think. Say it again, Clontin, say it about books. Yeah, I will start the cataloging and sharing with books, maybe records, I guess, what's going to be there. I hear a thousand of those too. Right, exactly, yeah. And there, yes, I think there is like, just because you will get consumers for the data, consumers also like a term which can be, but just in technical terms, right? It will be the first people interesting, like knowing about collections, it will be like maybe people in the social network about books that already exist. So the same, I guess, for records, I guess it will be like discogs or I don't know, like what is the current thing. But this will be like integration with these services will be a good feature for this network, I guess I'm saying. And the other thing is like, if you want to prototype something for these, I don't know what your plans were, but like my idea always, I'm always thinking about the agora in the background as you may know. You could totally imagine like the prototype as an agora, it could be like, you know, I don't know if you have a name for the project, but you don't like, personally, like... It's called the Artifactory right now. Artifactory. Artifact, I'll type it in. I like that actually, Artifactory. So, you know, I will easily bring up Artifactory.org, for example, because I have like, I can bring up multiple algorithms there, but just as a, if it helps you prototype, and then you could imagine like a user having a good repository with like descriptions of what they have, and that could be rendered as like, you know, as a part of that website, if that is useful at any point, just an open offer. It's interesting how CDDB and IMDB and all those things had a pun flash early on, and then a lot of them got just, you know, eaten by larger entities and corrupted or screwed up in different ways. Like the commons of information doesn't live very long. It's pretty sad. Another thing on my wish list. A lot of that has to do with economic model and DC funding. Yeah. Exactly. So, you know, trying to think of how you make this work economically where it's self-sustaining and still, you know, still maintains some kind of integrity for the users. Exactly. Oh, I was just having a thought that it's gone. Thanks for the links to Bookworm. I think that's what distracted me. Cool. Are you guys familiar with StoryCorps? Yes. Yeah. There's a StoryCorps piece to this, too. Oh, cool. I was just wondering that, like, there's a, the president of the Tenement Museum in New York is somebody of that who had, see if I can find a link to this thing. Talk amongst yourselves. I'm just looking for a link to it. So, in the meantime, how are you and I, if you want to share what you're working on or? Oh, sorry, me? Yeah. Working on, I showed sort of last time we met a demo of like, essentially a player, a media player and website configuration that lets you build static web pages like normal using a static site, but then have them act in a single page application style way where you can navigate through the static pages but retain the player playing without interruption through HTMX. Oh, HTMX. I wanted to hear the try. Yeah, yeah. This is my, me trying HTMX. Hey, good. Yeah, the idea being like, if I can figure out how to do this well, then I can have a player that can be like, and a process that can be applied to a variety of other sites easily and allow this type of functionality to happen where the player persists as you navigate through the site. I work with a bunch of like podcast people and my partner, she does podcasts as well. And this is like a huge unsolved problem for them because like to do it, normally you need to have like a big react app that needs to be set up in an SPI way. And then you need to have like a whole bunch of build process stuff. So it ranks on SEO and so forth and so on and so on. But here, like once I get this process working the way I want, you can just have your normal website and then layer HTMX on top of it and layer this custom HTML element of the player on top of it and have it work. So this week, I think I've gotten the player doing everything I wanted to do, which is like being able to navigate through things and have the player work like a player. And the test project for this is building some API queries to pull, if you're familiar with, this is my jam or from 61 or last FM obsessions, right over a variety of platforms. I've been doing picking one song every day to week period as like my obsession for that day to week period. And it's been split up amongst all these different platforms. So a while back, I did the work of moving them all into my Spotify playlists. Of course that locks it all in Spotify. So I built the process to crawl my Spotify playlists that I used to archive all of these old obsession lists. Nice. Hold them into JSON. And then I'm going to write, let's see, I was, I think I have it here, write it into separate markdown files for each track and then I'll have them all in this website. And I'm starting with YouTube as like the way that I'm pulling the, like the way that the player is going to work is you just play YouTube, but like the player element is a custom HTML element. So it can initiate and manage any number of players, right? As long as there's some way that I can monitor the API and understand when a play has stopped and ended. So the goal is to then also have the player expand not just to YouTube where this is very easy and a lot of people have done it, not the meeting the player part, but interacting with the YouTube API and also have a SoundCloud and other players that are out there. Yeah. I like that. I always wanted like playlists as a service like in the sense of, in the sense of like fully integrated at a higher level. So a playlist with YouTube and Spotify tracks. Exactly. And I'm also pulling the Spotify thing as well so you can have it on Spotify. And I'll have the whole bunch of them sort of archived for me, which would be nice. It's very cool. And also, once you have a cross-platform playlist, you can have services that say like, do you want to actually consolidate your playlist on one platform? And essentially offer that as a service which actually could solve for many new providers it could solve the issue of the empty page of customers to put it some way. You know what I was saying, you know. Do you want to import your existing playlist, et cetera? Right. It's getting to a commons first provider. There was a really brilliant thing that Apple did when they first launched, it wasn't Apple music back then. It was iTunes or something where they said, if you have LPs or concepts or just bad recordings of something, anything that you can make manifest digitally to us, we will give you back a cleaned up version in our service, which really just meant that they were just tapping their inventory. They weren't really like, I mean, it made it sound like they were like, you know, doing something to your recordings, but they weren't. But I thought it was just really smart about instantly populating your collection with the stuff that you already have of higher quality. Yeah. I think like that's, yeah, there's a lot of potential once you get the data out to do interesting things. And the other reason that I want it is like, there are a bunch of like last FM or there are a bunch of Spotify, there are a bunch of things I could do with Spotify data that Spotify does not natively support that as somebody who makes a lot of playlists that I would like to be able to fiddle with, like being able to pull off a track and see which playlists I have it in, which you think would be like a real obvious feature, but there's a lot of things on Spotify. And sort of the incentives for like specifically focusing on Spotify playlists here is once I get this first version of it done where I'm manually sort of doing these queries, then I can run it across all of my Spotify playlists, pull them all down to a website. And