 40% of something. Feeling like what? Sorry. 40% of full energy. What I think is my full energy. Who knows? 40% low. So. Bentley. Cool. Hey. Yo. How are you doing? Was that to Bentley? I was just checking in a little bit, Bentley, so. Really, and I, Peter. Yeah, I've been sick like, I mean, I don't know, Jerry, how sick you've been, but we've been sick in this household. And it wasn't COVID in our case, but so I'm catching up and getting stuff done. So not much interesting stuff going on. Sorry about that. Yeah, getting sick is not fun. Pete, did you want to check in? I'm doing well. Not going to win. Majelo, how the earth is going on. Yeah. Some cool stuff on math and wiki, chat GPT is cool. Yeah, look at boss. Has there been any activity on the project with Matthew to there? There has been a little bit. Yeah, I'm going to type online 15 in the docs. I don't know that there's a bunch of external stuff. We had some we haven't we haven't moved too much. But the website is now on a reasonably permanent URL, tftmap.massive.wiki. Oh, cool. And hoping to have phase zero done next week sometime, I think. Oh, that looks really good. Yeah, it's yeah. Is that a mock-up that's driven by data or what? It's a mock-up, total mock-up by Matthew. Good work. I guess there's a let me see if there's anything useful on today's Massive Wiki Wednesday notes. Nothing from nothing, nothing on tftmap project. Do the cool stuff, maybe I'll maybe not. Yeah. Do you want to sort of say where the math has gotten to or where it's stuck or what needs to happen next? Yeah, and actually I should maybe talk a little bit about Massive Wiki and since I'm talking. 25 minutes off focus. The robots are with us. Yeah, sorry, I forgot to turn it off. Let me see if I can share my browser. So I have to mention this because it was a big deal, even though it looks like a tiny thing. I hacked the Massive Wiki theme to have this little notice bar, which was a whole big thing. Anyway, and then separately, I've been talking with Jonas and San a little bit about Massive Wiki themes and general and stuff, so I know way more about CSS frameworks than I used to, and it's all sadness and disappointment. No, that's so bad. Is that ribbon a part of the standard default theme or is that something you have to choose and add? It's hacked into, it's not part of the standard theme, although you're welcome to copy it if you really wanted to. The other thing that this made me realize was that this theme is originally from Bulma and the thing that keeps the top bar, which is nice and doesn't scroll, the thing that keeps that from bumping into the content is actually just having this content have the CSS equivalent of break, break, break, break, or paragraph, paragraph, paragraph. It's just a static thing. So this made a difference when you make this thing small. I don't know if this is an example of it. But anyway, this thing used to wrap and then it would actually overrun the content down here. And so this version of this theme has a little bit more space bumping down. Whole thing's terrible. Cool. And then I went to look at a bunch of sites because we're talking about doing fancy things. We have another theme where there's a static sidebar that doesn't scroll while the content scroll. I was looking at Wikipedia and stuff. And there's a bunch of websites that the sidebar scrolls, the top matter scrolls, we're trying too hard is my. Yeah, you also have to be careful with mobile. When you scroll or when you get fancy with? When you get fancy, leaving it scroll is better for mobile. Yeah, the fancy stuff works. Now, this is interfering with positive, but there's a hamburger menu there. The themes that I've been using for Mass of Wiki are mobile friendly, even the one with the sidebar that doesn't scroll. But then, of course, the sidebar takes up too much space. Yeah, I mean, the head or not is going right now. You lose about a fifth of your mobile phone. Yeah. Yeah. On a side note, I've never been able to make the hypothesis bar not render there. I tried. Yeah, I haven't tried that. That's another whole thing. Yeah. I tried the way it's documented, and it's just not working for me. So it's there as well in the algorithm. I will try to fix it as well, and I'm sure I will run into the same thing as you. Or not, and then I'll copy your solution. That would be awesome. Also, if the menu ever gets wide enough, I don't know if you like it switching to a hamburger or a menu button. But all that stuff, I do a lot of. So if you ever need help with CSS. OK, thank you. Making it sticky at the top to where it doesn't. You don't have to pad or it automatically just do some things. I've done between menu bars and hamburgers. I'm getting a little hungry. Let's say that's helpful, Bentley, because I've had a lot of thought, obviously, about. So anyway, the other reason I could talk about that little thing is because I can click here. So if you look at the project plan. Version zero is basically a mockup of stuff. And so I can show you the air table. We have basically Matthew says we're pretty close to phase zero. All we have to do is Pete needs to fill in his contact information and a few more things, and we're good to go. So while I've got the screen, let me also talk about Massive Wiki. I've got a new thing called Tech Notes, which is a little bit like an RFC or a PEP from Python. And the first one was about moving to, I think it's in here someplace, moving off of GitHub, I guess. So it looks like Massive Wiki stuff. We're going to move code, and Wiki is largely over to Code Berg. I don't see a good link here, but. And the link to this HackMD, this HackMD happened just like an hour ago. So it's not yet at a permanent location, and this isn't going to stay at this location. So a little bit later in the column, move this to a permanent page, and then put the link to it. But so first thing that was about moving off of GitHub, maybe onto Code Berg, thought a lot about different other things. The next one is about something called Permanent Versions, which is, conceptually to me, it's similar to permalinks and what permalinks did for blogs. The kind of the short thing is a Wiki is supposed to be bubbling and burbling and changing all the time, and so it's always felt weird to me to point into, give a link to a Wiki page. Because if it's a healthy Wiki, it will have been composted away from what it was, and maybe even subsumed into other pages by the time you click on it. So this is a proposal plan to have a process for making a permanent version of the page. And Matthew was really thoughtful with some comments this morning about different ways to do that, and it made me realize that there are use cases where you want to have a more formal document management kind of thing, and you can do that in Massive Wiki. So if you were doing a government Massive Wiki, I think I can close this now, because I can't see the screen very well. If you're doing a government version, you might have versioned pages and links between versions and stuff like that. For more casual or more Wiki use, I'm actually not expecting very fancy heavy-handed version management, which you can kind of get from the Git log. And I'm not really thinking about helping readers much with getting from this version to the next version to the next or the previous version, previous version, previous version. Matthew was saying that Gordon Brandner is doing that subconscious, for instance. So the permanent versions thing, it's going to be like, here's the way it was. And if you want to find out more about this, search around for different versions of this or things like that. And in kind of thinking through that and talking to Matthew, part of what I hope, I guess part of what I hope for Massive Wiki and content in general is that it's going to morph and change and get copy and pasted in sections to all different kinds of places and all different things. And I think that's a beautiful and wonderful thing, even though it's not very database-y right once. And don't repeat yourself. I really like that organic kind of combining and merging and changing and spreading. So that's part of what's under Massive Wiki. And then to compensate for that, to make it so that I feel that way now, I didn't feel that way 30 years ago about information. I would be much more careful about keeping sources and keeping track of what was going where. Nowadays, of course, we have text search and web search. That's pretty good. Very good. So I don't worry too much about content getting loose because you can always ask a bot to go find it. And now watching ChatGPT this week, it's even more so. It's like, hey, ChatGPT, read this page or it's descendants. Check out this page. Can you tell me some of the influences that were part of this page? And in the future, we'll get a much richer and more vibrant description. Yeah, it looks like Pete kind of came up with some of this, himself. Bill added some cool stuff that was pulled from this source and that kind of thing. I think that's, I don't worry too much about very strict version management and attribute management because I think search and AI is just going to get better and better and better and better. Brief question. It's all. I have a question. Go ahead, Peter. I'll remember. Go ahead. So you mentioned the how-to version and those interesting thoughts. But I guess I didn't quite get what this is what you want to target in massive Wiki. Is it going to be linked to a Git commit, an anchor, or something that says load this page in this commit? Or is it going to be pointing time? The latter. OK, makes sense. So essentially, you will pass some sort of like, you want the version closest to a point in time run, like a retry, and the system sort of finds the right point in history. Actually, what I'm expecting to do, and this, the tech note literally is still a little bit in draft mode and I finished that draft this morning. I've been thinking about it for a few days. So this might all change. But what I'm thinking is going to happen is I'm going to use the same publishing process that I do to make a website. But I'm going to make a single HTML file and then actually take a SHA-256, check some of that, and use that as the URL fragment for it as the file name. So I kind of came up with that architecture intuitively. And then talking with Matthew, I realized a little bit more about why I was doing it. Matthew said, in the blogging world, especially in the olden days, the old blogosphere, you could go back and edit a blog post to fix a typo or something like that or update a link. But there was an unwritten code, a code of ethics, which was once I've made a blog post, I'm not going to, here I'm saying, well, I'm confused about this thing. You don't go in later and say, I'm so smart about this topic. You don't change the meaning of the post. And you don't even edit it very much. So I was always weird. It was always weird for me using a Wiki as a blog because a Wiki is changing all the time. And that's a good thing, right? So I realized kind of what I want to do is make a point-in-time snapshot of the page in HTML, probably, rather than markdown. And because that'll be more kind of long-term and more useful for more people. And then snapshot the SHA-256 of it. And then that's a way of essentially codifying that code of ethics that we used to have in the blogosphere. And it's not that I really appreciate code of ethics more than I appreciate hard mathematical precise firm things. But the code of ethics existed in the blogosphere. It's not something that we have globally in the web. It's not a common cultural convention that a web page won't change. So in a way that SHA-256 URL is a way of demonstrating to people for all time whether or not they're of our culture or some subsequent culture hundreds of years from now. It's like when Pete says permanent version on a massive Wiki, what it means is this is a point-in-time snapshot. And I can actually confirm that by checking the hash. Interesting. So your finger thing to mention. It reminds me, it sounds like content-based addressing, no? Yeah. What you're going for. It's like I see the direction, you know? It's kind of content-based addressing. It's similar to it. But it's actually for a particular reason. It's for verification, verificability. Right. It's like a proof. Yeah, it's a proof. And it's funny. IPFS, of course, is this content-based addressing. So instead of SHA-256, I'm like maybe I should use an IPFS content ID. It turns out that the promise of IPFS content IDs is a little bit different than I thought. Given a file, it's easy to, and it's hard not to, it's easy to create different content IDs from the same file. It depends on kind of the way you hash it and the way you coalesce it with. IPFS is made as a file system, more or less. So it actually doesn't prefer to store individual files. What it does is puts the file in a data structure called the UNIX FS, UNIX file system. And then depending on chunking and stuff like that, the ID that comes out of that might be different depending on how you chunked it. So it turns out that the IPFS promise is that if you have a content ID, you'll always be able to retrieve the content. But it's not like a hash. It's not, you can't do it backwards. It's interesting. So it's a little bit like content-addressed file system, but it's really for verification and a proof. You said something else interesting, which was, so the, I'm wondering if it was like picking a hash out of the Git log. Another thing in conversation with Matthew is like, I guess in my mind at least, Massive Wiki has always had that idea of being able to point to a version and maybe get sometimes other tools. Git isn't necessarily part of Massive Wiki, versioning is. But it made me realize that for people like Matthew, who aren't geeks, I need to, we need to, Massive Wiki needs to present a better interface to the Git log because Matthew stole like, you know, I'm going to save this version of this file as another file name in the Wiki. And I'm like, yeah, you actually don't need to do that because it's all on Git log. And he's, well, that's cool. But then if there's a thousand comments, how do I search to a thousand commits for the interesting ones? And it's like, well, there are actually tools for that, you know, Git blame or, you know, whatever. So it made me realize that we need to make it more of a first class operation for a non-technical person to say, give me the version history of this file or give me, you know, when did this paragraph get introduced in the Git history? And that's something that I can do because I'm a geek but everybody should be able to do that. Just two key comments about the topic. One is like, and did you really consider this? But like, I didn't quite get exactly how the hash will be used, but the fact that you're hashing the HTML output, I guess that introduces an dependency on the particular rendering engine you're using unless you're, you plan to store the HTML for each version, but you wouldn't be able to like in the future rely on just having the hash to find the right version if you don't have the same engine, you know. That's correct. My intent is actually to render the page and then save the HTML. Yeah, okay, it makes sense. If you save it, that makes sense. So that's where like, I guess, yeah, you could save it. And then the styling of that might break or something like that. If it depends on external CSS file. Exactly, but the content should still be fine. Yeah, it makes sense, it makes sense. Jerry, so you had a comment. Yeah, it was, you have to scroll quite a ways back in what we've been saying. I typed it partly in the chat, which is Pete, does a shift to CodeBird mean more complexity, less complexity, same, same, just a different host? CodeBird is fairly analogous to GitHub and actually even the interface is very familiar to a GitHub user. So it's, I think it's not as complete as GitHub, but otherwise it's conceptually very similar. So it's basically a copy and paste files into a new set of directories, open an account, et cetera, et cetera. Yeah, so CodeBird is committed to always being open friendly and they actually, they're running on software called GitTee, which has a storied history in itself. GitTee, the GitTee project recently, the maintainers decided to incorporate GitTee limited and commercialize a little bit, start making money off of consulting work. That made the CodeBird people freak out a little bit. So they announced a fork of GitTee, right now it's a soft fork called Forgeo, the yo part is spelled J-O. So that made me concerned a little bit. It made me feel a little bit unhappy, but I talked to it with John Abany, he's like, actually, no, that's a good thing to be. So GitTee itself is a fork of GOGS. So there's a whole fork fork, you know, thing going on. That's a rare, reclusive forkitude. It's kind of like when it's, my wife and I sometimes we're watching Hong Kong movies and, you know, that one of the protagonists is a mistress of somebody and, you know, it's like, you know, oh, and then I'll marry him, you know, when he divorces his wife. And it's like, dude, you know, if he's cheating on his wife, what makes you think, you know, so it's kind of the same thing with the forks, you know, if you fork from somebody else, you know, that doesn't mean that you're not gonna get forked yourself, which is why we say fork. I put a two things up top, Gordon Brander and Paul Roney, which I'm happy to go to, but it's a change of topic. So I'll wait until everybody's done with whatever we're on here. So just to close on the anchoring, so we will discuss this at some point for the hour and like, I didn't bring anything. So there's no version reference or anything yet. I was gonna go with the GIT hash, the commit. And then I thought, and you're going back to the, how to, and this seems like an interesting side quest. I was, I wonder, since somebody has done this, it should be simple, like take the GIT commit, commits, and like produce like, you know, three words or four words, some mnemonic, traditional hash. I think there's probably like general libraries to do that. Yeah. And that will be, you know, anything you could say, oh, it's just like, you know, dice, phone, key, your version. Yeah, that's brilliant. I like that a lot. I'm not sure exactly when I would use it, but I like the idea of it, yeah. Yeah. I mean, I guess it will make the, you know, just use that as anchor and perhaps more like user friendly, but yeah, I don't know. And the other thing is like, I was thinking of pointing time anchoring at some point for this, where like, you know, there should be, there will be a de-reference step, essentially, I hint at, you know, you pass like a timestamp or like a date and then the system finds the right version, essentially for you. And then which sort of like has a nice property that it deals this 3R UI for navigating because you can just change the date and get a different version. I am also aware of at least one very large documentation system based on my lab that supports a link into a particular commit or head. And then you can just like, it has built in like, you know, show latest or show adversion. Yeah. At some point when I'm really writing a massive wiki client rather than relying on other people's, I'm gonna be super interested in that kind of stuff. Helping, you know, helping people navigate, get history without being geeks. Right. Because of course, then I think I would like anything time-based will be, you could have wiki to wiki, inter wiki, inter hour or whatever links that are sort of consistent, you know, like you could browse cross wiki at the point of time. That's actually, yeah. I hadn't even thought about that as kind of a use case, but that's a great example. And yeah, Gordon Brander and some crunches, no? Yes, subconscious exactly. And I kind of wanna know what you're thinking about him and subconscious because I talked with him on Monday when I was at BetaWorks before I realized I was actually sick with COVID. And he like, he's really interested in maybe absorbing a piece of my brain data into NoSphere and figuring out what to do with it. And he was like, you gotta talk, gotta do this. And I'm thinking, is there a way to fold him into more of our collective work here so that we can all learn from that and do something there? So that's just a side note. And if he contacts me, I'll say, hey, there's a call you could join and I think some of us would be interested here in how that works, right? Yes, that sounds great, yeah, cool. And then Paul Roney is the founder of Cosmic and I did a podcast episode with him that I really, really like. He's very thoughtful. He's French. And he really loves computer history. So one of the things he's kind of spinning out is the possibility of a podcast or a community or a something to talk about. For example, some of the early visionaries like JCR, Licklider and the Weneaver Bush's Memex and all those kinds of things. What was the promise? What got fulfilled? What didn't get fulfilled? And how does that bridge to new tools and what kind of honoring and offering newbies and on ramp to the older visions of computing and then making it relevant for today. And we had a really nice conversation over this and I think, you know, he seems really energized about it. And I know that we have several humans who are really like Chris Aldrich loves the history stuff and some of us know a lot about it. So I just wanted to put that in front of the group as well. That sounds very exciting. Cool. Any other relevant topics for where we are? It's on the topic of like importing your reign, which is something I've been like wanting to look at it a lot for a while. I just want to share that we do have now an aura for the future of the link up, which is just like right now it's being served as a mirror of, well, it's just like the same content as the reference aura, but I can empty it and I put whatever there. And yeah, so I guess I just wanted to share that. So it's A4TL for now, but then we can use any other. Yeah, and like this is running containers the new version of the setup. So it's easy to run many hours. And yeah, in general, it goes to be a composition of any number of data repositories or sources. And I guess tying back into the NOS here, of course, the NOS here developing is like technically a bit advanced, like completely different architecture, IPFS based and so on. But I would expect that the general problem of take an input in some format and produce some, you know, like it transforms it through a series of steps until it can be ingested into a NOS here. It will have like a common path, right? Up to some point, to some perhaps intermediate exchange format that is common to many platforms. Like, so it could be, you know, go to MasiWiki, of course, you could go into the Phylloshedini aura or OGM or your own aura. So I guess I'm interested in trying to keep it in mind and I don't know if that means we try to keep in mind like a common exchange format, like, you know, or at least we share, for example, like the, so the idea behind the current architecture, which has like aura breach as a separate component is that essentially, you know, if we have to pay the tax of, I don't know, like run a process so we can get data out of the rain, and then we do that once, right? We have actual days that just does that and produce like a common exchange format, one of those. Just sort of like, I mean, I would love to keep this in mind, not only because, you know, it could benefit many projects from the group, but also because like, I want to avoid repeating, you know, the common pitfall of open source and so on, which is like, we all end up in reinventing the wheel too much and paying like double or triple or in full, you know? Totally agree. One thing that came to mind while talking to Paul and telling him about Gordon Brander, wanting to suck in some portion of my brain was that, and this might be interesting in bridging to your Agra, was that we could take a couple of degrees away from the link I just shared in the chat, which is basically the link to those vision documents, like personal dynamic media by LMK and computer lib dream machines by Ted Nelson and Engelbart's demo and so forth and use that as a starting piece of, a starting web of content to play with, to do with importing and then to tell stories around, because that would sort of serve three or four different purposes in different ways. And that might be a nice, just might be a nice topic to choose as a focal point. So just to think about it. So I wonder, we've got a project called Meme Brain and we've got, which includes maybe Bentley and me and Mark and Tuan, Carol. We've got basically all of Jerry's brain in a Postgres database where each row is a node, I would imagine. So we already have that. I also kind of wonder, there's an odd thing about Jerry's brain. Jerry's brain is almost... These one. Yeah, that too. Yeah, there's two. The way Jerry uses the brain, he only has node titles. He doesn't have node bodies, which is good in some ways and bad in some ways. But anyway, I, you know, representing, thinking about representing in a standardized format, a massive wiki page, even the sections of a massive wiki page don't map great to, you know, nodes out of the brain which don't map great probably to Rome nodes which don't map great to, you know. So one of the challenges, I think, that our space has been dealing with for 30 years or whatever is everybody who's representing information represents a different shape of information. So a Miro node is not the same as the brain node. And so what we've been doing is, or sometimes we do lossy translations, but they're very lossy. I don't know how to resolve that kind of. Yeah, I mean, I think this is where like the rubber hits the role. I think that's a question, I don't know, yeah. And I mean, I guess, yeah, lossy, I think represent will be like lossy user friendly representations or lossy, you know, lossy useful representations and thought of, but maybe inaccessible like a lossless. But yes, how do you file? How do you file like say, like an image that doesn't have like an old mapping? Perhaps that means you need like a some level of indirection for like, at least, you know, at least you have a resource ID always. I guess, even if it's just a random. And then you can say, well, for anything that doesn't have a direct mapping go to the like backup database for mapping or yeah, but I will be very interested in that. I mean, I'm not interested in a lot of details. I'm actually like, no detail like Oriental, I know it's obvious, but like this seems very, very like, it needs to be solved. Essentially, there's no other way. And like starting with like, so I guess, you know, like we have Jerry's reign as like in one, in one, one side of the ring to put this on weight and like, and the sort of systems. And I think that's a very nice base case to start with. Yeah. Yeah. But I will personally be interested in any such, any reigns from this group to put it some way. Yeah, me too. Also to explore that diversity and for example, like the more general thing I'm, I think I'm most interested is like, how do like Jerry's nodes map to Peter's and Benley's? Yeah. Even how do we find if there is no mapping, right? Or like, you know, how to define that relatedness or that metric essentially of distance? Yeah. The, it's a tricky problem. And I'm not sure I share your optimism about being able to solve it. And I think the solution is, I don't know, so I guess my solution would be, I wish there were an open source version of Jerry's brain and then, you know, it and massive wiki could talk back and forth on link together, but you probably wouldn't, you know, you wouldn't try to translate one into the other. Dang, I distracted myself. Jerry's brain is a great map, but it's, I guess it would be super easy. It's super easy to represent Jerry's brain as a massive wiki, but it's a very degenerate massive wiki. He's a data term, not a... Man. Things I use. Well, I mean, what it winds up looking like is a series of markdown files that have a title, two thirds of which have a URL, 98% of which contain no text of any kind. Right? And all of which have some YAML or some other metadata that says, up, down, left, this is what you'll locate near me. But that's it. I mean, it's a degenerate wiki and massive wiki in the sense of it contains no content that you would normally refer to that way. So there's, I guess there's, under the quest of these things should interoperate, there is actually a simple way to make them interoperate. It's very simple to convert Jerry's brain into a massive wiki and I've done little ones. It's trivial, but it's not very meaningful either. So I can say that, oh yeah, I've represented parts of Jerry's brain and massive wiki. I could represent all of it trivially. But that is not the same as, I've done a useful mapping of it. But it's meaningful in the sense that you can have plugins to obsidian that give you mapping capacities that aren't native to obsidian that then let you see relationships or navigate through the connections. And exporting my brain content would give us the basis for picking up and then building a new fresh map that mirrors how the brain does stuff. And as you've heard 50 times from our conversations, I just need the simplest of things, up, down, left and some kind of interaction model and we're off and running. It's easy to, it's a little bit like a trap door function. It's easy to move Jerry's brain content over to a massive wiki. But then if you ever want it back in the brain, you can't do any massive wiki things with it. You're totally right that you could use massive wiki or obsidian or whatever, you can use tools to do different things with the data once it's in a format that a mapper or whatever. So that's true. But I guess another observation that we've had playing around with Jerry's brain is until it's easy to do one way. It's a trap door kind of translation, but it's a much harder problem to do it bi-directionally. And annotate it in massive wiki and then bring it back into the brain in any reasonable way. The brain is complicated also. It's not just the data format, it's also the fact that it's more closed. But it's both things. I guess I was also interested in, I think we can have different goals here. And I know Jerry, you have to also let it. So if we want to, for example, Jerry's brain in the context of like a wider knowledge graph, just could that be adding a lot of value in that context? And perhaps that's the first thing. And also it would be like, so something like having a search engine which like is upranking Jerry's brain all the time. That's interesting. Even if you don't solve the other way. So my apologies, I'm gonna shut down the recording you guys can all stay, but I've got to stop the recording so it saves it to my system. If one of you wants to keep recording, that'll work as well.