Rel8 Use Cases (transcript)
Hi, I'm Jerry Michalski, and this is a description of use cases in the Rel8 project, a sense-making story.
The normal situation is there's a bunch of people out there in the world engaged in sense making. They're taking a whole lot of different approaches. There's something called Zettelkasten, which a philosopher named Nicholas Luhmann created; Build a Second Brain, Linking Your Thinking, which are kind of proprietary educational systems; Bullet Journal, which is a way of keeping notes; Roam Research, and other tools that have backlinks in them. All these things are different parts of the community.
In fact, if I go over to my Brain, I can show you I've been mapping a variety of different communities. doing work on personal knowledge graphs, personal knowledge management, personal knowledge networks, enterprise knowledge graphs, distributed knowledge graphs. back in the Navkin diagrams, all sorts of collaborative brainstorming presentation tools, semantic hypergraphs, facial analysis tools. There's a whole bunch of subcategories. none of which are particularly connected. So a lot of these efforts are personal. We don't really have a shared memory, and a shared memory would be very, very useful. So why can't we share what we know better? Well, this is one way of looking at the current situation. There's a bigger picture as well. It starts with disorder rules the land. We are drowning in the information flood. Misinformation, disinformation, malinformation, and spin are rampant. We are being spun and civilization is in decline. Because we're not collaborating. Education is not the way out of the box, and critical thinking is actually an endangered species. So Relate is about rethinking collaboration to try to create a reliable shared memory that makes it easier for each of us to take notes and then figure out what we believe and express our points of view. This presentation is just one path in. It's a way of painting, a word painting, a prototype for a framework through the discussion of some use cases. So let me start with the use cases. Imagine PowerPoint or Keynote, a presentation tool, as a playlist of wiki pages. I'll come back to what I mean by wiki pages in a little bit, but. Imagine it as a playlist. That means that on the left here, PowerPoint is a playlist of slides, slide one, slide two, slide three, slide four. Slide one is imagine PowerPoint as a playlist. is this page. Then the next one is uh imagine a book as a playlist which we'll actually see next but imagine each of those is on one page which when you hit play Then a little bit of the app picks up and says, oh good, I have a playlist which makes a presentation. I'm going to create. I'm applying my favorite theme to that. So it looks like a presentation. blow away all the menus and go full screen and then add left and right arrows so that I can hit uh next next next and go through the presentation Now, that's really interesting. It means that the presentation consists of a grouping of individual files that are just in this namespace. And there's no more attached PowerPoint file. I'm not shipping you a PowerPoint presentation as a file that has a series of slides in it. I'm actually shipping you just, hey, uh the the the current presentation is this set of pages So there's no more wrong versions of the PowerPoint. It also means that each of the pages is reusable. You're no longer hunting through a variety of decks to figure out where's my favorite presentation. my favorite slide uh that explains income inequality uh because you modified it and you you've you've got a whole bunch of different presentations so you end up opening them all and and searching searching searching Here they're actually in the namespace. It also means that you or someone else can use the same page in different pitches altogether. So different decks could include the same page just by reference or a piece of a page. Gets really interesting. And here comes some of the wickiness. Every page has a full version history, so you could scroll forward and backward through every version that was made. And if you want to create your own version of a page, you just fork it off and make your own version and keep your own version control. That makes it really easy to play old versions of the deck. So you could say. Show me this deck the way we played it in November of 2020. . . And bloop, all the pages would find the most recent page as of November 2020, and you hit play, and shabing, you're back, you know, back in the past. It's also easy to bring that deck up to date. So you haven't watched this presentation since November 2020, but all the numbers have changed. Let's say it's a sales forecast. You could then say, hey, play this playlist, but bring all the pages up to current. So any modifications that anybody had made to the pages would be applied and you would then see the latest uh the latest version of it. Slide sorting, the big slide table, is just a view on these pages that are in the namespace. So that's that's the way to rethink PowerPoint or keynote, basically presentation software. Now imagine a book as a playlist of wiki pages. In fact, it's a playlist of chapters So each chapter, one, two, three, four, the first chapter could be Imagine PowerPoint as a playlist, etc. etc. You know how that goes. And when you hit publish. Smart software adds front matter, end matter, gives you the files for an ISPN number, puts, plops that in, sequences the chapters, and formats them the way you want it to with your theme you could commission some cover art and add that in, automatically generate an index and table of contents, and then you could export the whole thing to EPUB or Kindle Direct Publishing, shibing shaboom. Now, if you are a bad writer and you have bad editors, it's going to be a really bad book. But if you're good at that, and if you collaborated on the pages the way wikis let you collaborate, this could be a hella book. And so the pages and the chapters, again, are reusable in different books. Imagine five books publishing concurrently that share chapters but have very different messages. You could intertwingle your book with other people's books. Each page, each chapter has a full version history, as we said before, and it's easy again to update to output up-to-date versions. Now imagine a blog as a collection of tagged wiki pages. That means that blogs, you know, normally blogs are just, hey, here's a new post and bumps it bumps the other posts down because we have like chronological reverse chronological order. And what if blogs were just persistent pages in the wiki namespace enhanced with a hashtag that tells you which blog or blogs that particular post belongs to So at the end of a post page, it would say the wiki blog or the relate blog. And then when you visit the wiki blog or the relate blog, it says Hey, in this namespace, find me all the pages that have this hashtag on them, and then sort them in chronological order and add in matter to each post so they look like blog entries. Done. So you can start imagining that a page might actually appear in a presentation as a piece of a book or as a blog or something else. Now You're thinking, yeah, but a presentation has to have big text and not too many words and a blog post is different. Absolutely correct. And we'll kind of get to get to that in a little while. But but conceptually, there's not that much difference between these different things And what we're starting to do is pivot around the data rather than jam the data into a series of strange abstract structures that then just confuse us and get lost. So again, the posts in blogs are reusable. You can cross-post to a variety of blogs. The blogs could become seeds for book chapters. So you could roll up five or six blog entries and turn them into a chapter, for example. They can become seeds for other people's stories and each page has full version history. Imagine your notes. Your simple notes is a collection of tagged wiki pages. And some people are doing this. My friend Bill Seitz has a Sitesweb and a variety of other things where he's been taking notes for as long as I've been using that brain thing I just showed you But tagged wiki pages means that each note winds up becoming a freestanding nugget of information that's addressable and shareable. And thinking of it as a nugget gets really, really interesting. So when you take notes, you start to separate the stream into nuggets. You add tags, so you have some metadata, your AI might add some metadata, and then you post all these notes publicly, and then now and then you don't post something publicly. You keep something private because it was said to you in confidence or you don't feel confident in sharing it at the time. But uh that that's sort of how that starts working. Now, again, notes are reusable. They're the seeds for discussions. They can become other people's notes or include other people's notes. You have full version histories. And then we can begin to offer one another suggestions. If you're familiar with GitHub's pull requests, sort of fork and pull, you could do this with each individual node. Some of these notes become shared pearls, I'll call them, which is a nugget that's actually really special, where a bunch of people have agreed that this nugget is really good and speaks for me, even though it was written by another person or another small group of people I'm not even going to bother creating this nugget because that one is really excellent. And this gesture of pointing in toward similar sorts of nuggets. I think of it as kind of crystallization is really, really important because as we start we we don't really want everybody having every thing in the world described in their own individual way, which other people are just never going to visit. We kind of want clusters to develop and these crystallizations to happen so that we have streams of context and shared meaning. And this is where collective intelligence really starts to bubble out of the system. So more on that later as we actually sort of figure more of it out. So what do I mean by wiki pages and nuggets? Well, it's it's documents that are versioned, that have group editability, that become a collective asset. A nugget is a linkable reusable link. And then nugget size matters, something that we'll touch on just a little bit in a moment. Some of the magic behind the curtain is that these documents are in something like markdown format, which is a really simple version of hypertext markup language. They might be on a shared platform like GitHub, so they are shared and versioned by the platform and separated from the app. That means that they live in the new commons, the new information commons. they would have metadata either in the markdown format in something like YAML or whatever or connected to and and uh linked to the raw data. And then there'd be a proliferation of backlinks. Backlinks would be a really simple way that these things all have ties that connect them through the web. So Everything I've just described kind of looks backward. I'm just emulating PowerPoint books, weblogs, and so forth. in a completely different way, requiring no magic. I don't require AGI, you know, uh and a little bit of code would actually make these things kind of uh stand up. We can we could prototype this pretty quickly But it flips our conceptual model of documents and messages and how these things and ideas, how these things all relate to each other. Awesome. Now things get kind of interesting. So for example, what can machine learning do in this in this kind of environment? Well all kinds of interesting things. So while you're busy harvesting information and taking notes, it can greatly increase the speed with which you can do anything you can do. It can give you enormous breadth to search or find or link across all sorts of domains. It can suggest connections to you and make connections much easier than you would if you were doing it manually, which is what I've been doing, feeding my brain for 24 years. It could have a variety of different roles. It could be a conversational aid when you're in conversation with other people, sort of listening and offering suggestions and taking notes and getting things done. It could be a virtual assistant to you, off searching for things when when you don't have time to do that. It could be a wise chat bot that says something like, um, hey Jerry, I noticed that you're in the middle of kind of a dilemma in a kind in a group conversation. MIDA suggests some patterns out of the wise democracy pattern language to implement. That sounds interesting. Machine learning could also enhance metadata, generate abstracts and summaries, reduce duplication, all of that. There's a bunch of big open issues, including, hey, how do we write APIs for machine learning so that machine learning can play a part in all of our work here? And really importantly, how do we design machine learning so that it doesn't sort of kill us or sell off our data or do whatever else. How do we keep machine learning on the good side? That's machine learning here. There's an essay I need to finish writing called Data is the New Soil, where we separate apps from the data so that we build a new commons and different people using different apps. Can improve the data. And then when they come in, when someone else comes in from a different app, they benefit from the improved data. So think of this data layer as fertile soil. And if you know anything about regenerative agriculture, If you pay attention to soil fertility, a lot of the rest of the system sort of uh sorts itself out. So this is a metaphoric way to think about data. Shared data, not as the new oil, which is something we need to extract, protect, and sell off to other people to manipulate us with, but rather as the shared asset that helps us gain insights and figure out what's going on in the world. Different take on all of this. It's interesting to take notes, it's interesting maybe to design arguments and to build evidence, but what if the same environment that you were using to do all this were your storytelling environment? And here I talked a little bit about PowerPoint and Keynote, about presentation software, which is a form of storytelling. But imagine other kinds of storytelling more toward Prezi. which is an app that until recently was my favorite way to do storytelling because it allowed me to to draw things and write texts and then create links, basically create a path through that that was visually memorable and compelling. and resonant with whatever the message was that I was trying to say. So imagine stories kind of like that, but that link back to their sources. So you could sort of say, oh, okay, stop, stop, stop. That was a really interesting point. What's the evidence? Who's being quoted? And you could go directly to all those things because they'd just be connected out in this web. Again, the stories would be reusable. They might contain conversations. So there might be a topic, how do we turn, how do we keep machine learning to be good and not evil? That certainly is a topic for the ages that we'll have lots of different interesting discussions in different forums. Why can't those forums be easily at hand wherever that topic lives? And then the stories could support many different media forms. You could output them into videos that you post to YouTube. You could create other artifacts that you navigate through and work in day-to-day. You could have drawings that refer back to them. uh you could put them in three dimensions or augmented reality or something like that. And then and and this is like a really important piece of this whole enterprise for me. But so it's sort of surprising it's this late in this description. But what happens when two people or more people who use different tools and really love different tools and have done a great job of curating ideas meet and talk? How do they share their ideas? How do they connect to one another's ideas? How do they build something together that keeps each of them from duplicating everybody else's work, but allows them each to preserve their own individual preferences? and superpowers on doing things. So what does idea sex look like in this kind of an uh this kind of a world? There are a jillion design questions that spill out of this. How do you design for composability? How do you design these nuggets and all the different widgets and tools so that kind of like the Unix operating system, which is just a big bag of scripts and tricks? that you call up in sequence, you sort of pipe this command to that command to this command to that command and bingo, something gets done. How do we create composability in the design of this sort of environment? How do we convince vendors to design toward this composability? What are the right levels of zoom? From labels to summaries to full blocks of text to some sorts of graphic representations of whole chunks of ideas. What are the levels of zoom? How do we switch from one app's particular way of seeing things, let's say the brain, to a different app's way of seeing things, let's say Kumu or GraphFiz or something else? And how do we switch to the more appropriate one depending on the task at hand? How do we use this environment together to set up better questions and to set up experiments that we can go track and follow over time? How do we fold in visuals and graphics so that the visuals hold links back to the documents and are not just drawing some pretty pictures that are JPEGs, but rather are active parts of the entire discussion and thinking process And then who's going to do this? Who loves this? Who's already addicted to something? Who would play in the larger pond together? These are all sort of the beginnings of design questions for this. There's a bunch of other open questions like how does knowledge crystallize? How do we preserve personalization and preferences? How does this data, is the new soil, get decentralized in a trustworthy way that obeys privacy and permissioning? How do we motivate people to work in public so that what they know is actually shared out as opposed to just private notes? How do we motivate vendors to write toward RHEL 8 specs, standards, APIs? How do we design those protocols and APIs? And then how do we organize ourselves to get any of this done? So those are some of the big open issues here. Why the heck undertake this? If this works, it's going to sort of it's going to boomerang back into a lot of important fields like education, journalism, science, politics, and governance that might help improve critical thinking around the world. We are prisoners of very, very old media formats like the book. and the blog, frankly, and uh PowerPoint and all that. This is a way of exploding all those things and blending them In a way that permits different people to express their opinions as fully as they can with the tools that best suit them, including borrowing a whole bunch of things that they did not create, but other people they trust created. So they can point by reference to a whole series of artifacts and say, look, my point of view is composed of this batch of ideas plus this batch of ideas plus this other thing. And now I can show up and we can talk together and have basically idea sex, if that's not too crazy a way to look at it. Also, we're being kind of shoved toward uh some conceptions of the metaverse that, from my perspective, really suck. In particular, Mark Zuckerberg's vision of a metaverse that that looks like, you know, a big 3D store you walk around in wearing goggles, the way you can buy the the scarf that the lady just walked by with, and that that's some prime directive. important thing to do as opposed to solving problems that matter in the world together. And then the Web3 crypto world in which smart contracts achieve a perfect society with NFTs and crypto being the trustless platform atop which all of this works. I think there's some ponies and some pearls inside of these in these different uh ideas, but really um Wouldn't you rather be trying to solve the world's problems together than buying that cute scarf that walked by or uh buying and flipping an ugly ass NFT? I mean really what like what what do you think is going to be better for humanity? So there's some compelling reasons for for doing this. Thank you for for um coming along for this ride. I'd love to improve this model and and see what uh