May 14 - Building In Public ft. Jeff Tang (Athens Research)

On Building in Public today featuring Jeff Tang, who's building Athens Research. Welcome, Jeff.

Hi, David. Hi, Misty.

So Athens Research. So I feel like this is a name that I first heard about on Twitter because there's kind of a— there's a little bit of a— what's the fair way to call it? A little bit of a beef With kind of the Rome cult. Is that, is that an accurate way to describe it, or?

Best story. Wow. Obsessed with the story. You should definitely tell it.

Yeah.

Yeah. Yeah. I mean, some of, there's some beef. It's also kind of playful because I don't know. I think we get along at the end of the day.

Yeah.

Actually, I'm in Utah right now and I was hanging out at Connor's room. Oh,

amazing.

Amazing. And then it may come over again later this week. So

is this actually kayfabe?

Yeah, there's a little bit of kayfabe, but there's a little bit of reality to it too. It's fun that way. So yeah, I mean, the origin story is I was getting really deep into Roam and before that, deep into Twitter, or sorry, deep into Notion. But eventually figured out that Roam was like really great for my needs. Saw this humongous vision behind it and got really into the Roam cult on Twitter. Eventually the founder Connor, or one of the co-founders Connor, he asked me to interview and ultimately I didn't have enough Clojure experience, which is the language that they used.

All right.

So decided to start making an open source alternative as a way to learn Clojure, and it kind of just got some steam behind it. Eventually just continued working on it, and I was working on a different startup idea, very, very like early stage, like a few months in, no traction. And I was like, well, I should probably just pivot because this has way more interest. Continue working on it for, you know, 6 months, 9 months. Tons of contributors started helping out. Eventually some open source founders and investors were like, you're at the perfect stage to go to YC. Let me refer you. And I applied. I thought I bombed the interview with like Michael Seibel and all of them. And then they called me later that day and said, like, come through. That's awesome. We just finished the winter batch about a month ago. Raised a seed round, and now we're growing the team and hiring.

That's cool. And so how did— this is kind of an open source project that you were doing kind of as a side project while you were working on another startup idea. Were you already deep in kind of open source communities, and do you have a lot of people like kind of following your work, or was this kind of one of your first forays into starting a new open source project?

It's certainly the most open source success I've ever had. I had previously worked at a company called Strada Labs, and they were in the Ripple XRP ecosystem. And we were developing open source software, but it was more for infrastructure, and it was pretty directly related to infrastructure and very attached to, like, the company behind it. Hmm. And before that, I mean, I had pushed a bunch of projects with, like, you know, 1 or 2 stars by my friends.

Right.

But nothing serious. That being said, I've always been a longtime user of Vim and Emacs and Linux. I've done the majority of my programming career with these tools. But I was never just, like, one of the big open source indie developers that, like, have big presences that we all know of, like Dan Mob of React, something like that.

And so did you begin this as an open source project imagining that maybe, you know, it would have a commercial application with it, or was it really just a hack project to learn Clojure with no, you know, no vision beyond that?

Well, you know, I didn't have expectations going into it. Like, when I started it, I didn't have expectations like, oh, I'm gonna be done with YC in a year and raise a seed round. And that was definitely not the case. However, as I said, I saw a huge vision behind this, like some of the concepts really pioneered by Roam and thought, well, this is a huge space.

Yeah.

It's really bigger than anyone else sees except maybe the actual people working at Roam, but I see something huge here. And as it turned out, open source made a lot of sense in a lot of ways for this kind of product.

Yeah.

And

so what is that kind of, you know, when you get— you're deep in this ecosystem. We've dabbled with the tools, we know of their existence, we have accounts, but I wouldn't say that we're deep in probably the way you are. You're like living it and you've been in it for years. So what is it that— how should we think about The vision of what can be achieved here?

So I boil the vision into 2 parts, generally speaking. One is something more personal, and I call that like a second brain. Tiago Forte really made this building a second brain thing, and I think the second brain is its own thing now. And then the other part is more collaborative, it's more social, it's more multiplayer, and I would say that's more of building a second internet. And there's like, basically there's 2 concepts there. One is like the second brain where it's like the original vision of the Memex coined by Bush, which is like this like second computing machine. And sometimes people call it, you know, personal assistant. It's also the vision of a lot of AIs and like personal assistant, like Siri, Alexa, all those kinds of things, GPT-3. And maybe even like the knowledge graphs that we're building now. at Athens, which isn't like using AI directly, but it's building this map of your mind.

Right.

And then the second is, yeah, the second internet, which is more like, okay, what does it look like if we were to connect our computers? Like we had computers at the beginning, it became a lot more interesting once we had internet and we could like go to each other's websites, MySpaces and whatnot. And then a second internet in terms of actually connecting applications together with bidirectional links. Because we've had these single-directional hyperlinks since the advent of the web. What if some of those were bidirectional, like 1%, 2% of them, right? Mm-hmm. That's how we navigate the internet. Every single internet user is a user of hyperlinks. What if some of them were bidirectional between applications? So, like, within applications, you could say that Athens is like this intranet. We're linking everything together. What about between different instances of Athens? So, like, me and David could share notes.

Yeah.

On

building in public. That'd be so cool, right? And then building and then connecting between different types of applications. So we've kind of got this like note-taking app, this research tool as this first case study, but that's really just the first killer app potentially of these knowledge graphs and bidirectional links. What if we had chat with bidirectional links that had this emergent structure as well? Or if we rewrote Twitter with stronger linking primitives, um, and then we could connect our knowledge graph for Athens to our social media, to our calendar, to our chat applications. Those are kind of like the two things. Yeah.

Yeah. Wait, can we get into that a little bit? Because I saw a tweet that you'd written a while ago being like, should we reimagine what Slack or WhatsApp are with bidirectional links, or should we go this other direction, which maybe you're starting with now? So what were you thinking about then and how did you decide what to start with?

That tweet was the inaugural tweet because like I'd worked on Athens for like a week.

Oh, wow.

To learn Clojure, but I'd also worked on a Figma mockup of a chat application with bidirectional links and like a message channel. And so I tweeted everyone, I was like, do you want open source Roam or do you want Slack, like a Slack replacement? Yeah. I just think people can really imagine what the slot with the chat application. Well, except for a few people. Yeah. And stuff with Open Source Realm. But like now for something like chat, I guess the question was like, what would that look like? Yeah. Well, so the way I see it, like chat is really similar to note-taking and so is like social media in a sense, like Twitter at least. Like a chat is just like a bunch of notes between people. A tweet is just like a note to the world. And for chat at least, I mean, the same problem that Athens is solving, which is that these notes don't fit into like a file system, like a strict discrete tree.

Yeah.

Actually, the best ideas nest, like they definitely span across multiple trees. They're like in multiple categories. Like right now we're talking about tech, we're talking about building in public, we're talking about like knowledge graphs. And in the same way we're using Slack or Discord even. You can put one chat in one channel and it's like the general channel or it's like the engineering channel, but like, this is definitely touching design. It's also touching product. It's touching, it's touching marketing. And so if there was a way that you could actually see like all these different channels at once, but the same message in all of them, that would allow kind of the structure to emerge and not have to figure out putting one thing in one place, like information silos, as it were. Mm-hmm. Yeah.

Yeah,

so

the general structure then, it sounds like it's structured on this personal knowledge, this Memex kind of daily journal modality, and the kind of network side of it is further afield, or is it sort of just a glide path that people go on and get to the network side quickly today, or how does sort of a kind of an experience with With the service work today?

For Athens?

Yeah.

Yeah. So right now we have some people coming in from other like alternatives like Roam or Obsidian. There's a lot of great options right now in the PKM space. We also have some people coming in cold, like using Apple Notes and just like realizing that it doesn't work for them anymore. Some people coming in from pencil and paper. Honestly, I use all of the, all the above. Use multiple tools, by the way. But yeah, I think that right now the really like the people that I'm thinking value the most, get the most value out of Athens are people doing a lot of text-heavy research, like consumption and production. So those are really like researchers, maybe in academia doing maybe a high-tech research at a company like OpenAI. Maybe, yeah, people doing biotech research, something like that. Because basically it's really great for like literature review where you're like pulling in a bunch of information from a lot of sources and then you're referencing them, right? So really what this knowledge graph enables like 100 times more easily than a typical note-taking application is the ability to make references, which is like the foundation of what research is. Like, every time you start doing research, the first thing you do is a literature review. You pull in all these references.

Yep.

Try to gather like the current claims. And then from that, you can, you know, move forward to a claim. And we can do that with bidirectional links very easily. And then we have block references, which is like the most atomic form of reference where like I can point to a specific line of text in my graph database and reference it from somewhere else. And if you think about like academia research, that is The first thing you do is referencing. And then again, like, in just going to the tech application, it's like, how did Google eventually figure out the best search algorithm out of all the dozens of search engines? They weighted annotations, which they got from doing their PhDs at like Stanford. And that's what we're really emphasizing. It's those backlinks. That's, that's what makes something interesting.

Do you— I was saying, uh, to DK earlier that you should screen share if you want, like, if you have any favorite examples of like cool things people are—

Oh yeah, can we look at it?

On Athens just to like get a peek into some of the mind maps that are being built right now?

Well, I don't want to use mine right now. Let's see.

Are there any kind of public ones that we could browse around?

I'm actually— let's see. Well, we have our demo online. Actually, so I can show that.

Yeah.

Yours must be quite incredible by now.

It's quite massive at this point, a couple thousand nodes. But yeah, this is our GitHub Pages and this is an empty database. There's nothing in it. You can see this empty graph. If we were to load some test data though, this pulls some data in from— there we go— one of my original public Roam graphs that I just imported into Athens.

Oh wow.

And yeah, this is the graph visualization. It's not particularly useful yet in my opinion, like especially once there's tons and tons of nodes. You have to have better filtering capabilities, better search. But symbolically, this is what's going on. And I could just start typing, like, today I learned Clojure. And then you could see all the previous pages and links that I have referenced in the past in my Athens. So these are all articles. Or, you know, just maybe just a proper noun like Clojurean, someone who writes Clojure. And I could autocomplete that. Then I could click on that link and see all the notes I took on this blog tutorial on Clojure. I could see, oh, today's date, I actually just referenced this. See the reference. And then I could see other pages. This is a blog post I wrote when I was one month into Clojure explaining why you should learn Clojure. And then I could expand this context and be like, okay, this is actually like where I wrote that and continue to zoom out and see that upper-level context.

Yeah.

So, that's the bidirectional links. And again, these are the block references where I'm referencing a specific line of text. So, I could say from this book, Closure from the Ground Up, or blog. An amazing intro to Clojure. Clojure isn't just function-oriented in its syntax. It can be object-oriented, blah, blah, blah. All this stuff. And if I clicked on this reference, then this is, like, really huge. Probably just make this smaller. But you can see that I could— I can see where all the references to this specific block actually occurred from. Right. And that's really powerful, right? Because it's no longer paste. It's no longer just a quote. removed from its context, it actually shows you where that context comes from and all those places that it's referencing. So really creating this second, more granular internet. Yeah,

right. And when you think about these— this kind of second internet, are— I guess as an open source project, I presume there's Like, uh, we're right now on, on a URL where you have like a hosted version of the open source project. Is that right?

Yes. So, this is just the demo. We don't have like a hosted version yet. And we're—

Okay.

Working on the self-hosted version for people that want to self-host their data. But yeah, I think both of those are both on the roadmap. We want both of them, of course.

And so in the world of like a self-hosted version, I'm just trying to get my head around how all the linking works. Like if you have a self-hosted version of Athens and I have a separate self-hosted version, we're keeping it on different domains, different servers, do they— they're able to share these links?

Yeah. So, you know, first, before we like go after this big, hairy, audacious, like, vehicle and vision. I think that we can kind of like reel it in a little bit and just try to see like what happens at smaller scales. So like Second Internet is like millions of people using this.

Okay.

Let's like, let's see first how it'll work with like 10 people or 100 people.

Yeah.

And yeah, in that case, if we were all working at the same company on the same team at Highlighter, then it would be, you know, kind of not too different from how permissions are done on the Google Docs or Notion. now of just like sharing. Now at a scale of like, oh, these are like potentially interacting with strangers, then yeah, I mean, Athens would either have to build some gigantic social network internally or we could just use some of the emerging public blockchain technologies so that we could actually like share data without having anyone mediated. But those are both further away.

Sure. So it sounds like the— is the first use case you mentioned kind of like The Highlighter team, we could use Athens and have our own self-hosted one, and we would be building like an intranet of private knowledge to our team. Is that the core?

Yeah. I mean, first it's the individual, and then you're like, hey guys, this is so awesome, we should start using this. And then you're like, fine. And you're like, okay, fine, the company is now using it. Right. Yeah. And then eventually, hopefully, I mean, again, then there's like a big leap, probably have to cross some big chasm where it's like, Okay.

Yeah.

I, as an individual, want to share some notes with David about building in public. But one step at a time, you know?

Yeah. Right. And that almost to me seems like, I mean, I agree it's far out, so I don't mean to get too deep into that, but it's just like such a fascinating idea. I wonder how you imagine that future may look like. Is this, is there a new internet protocol that would emerge to manage this bidirectional linking?

Yeah, so it would either be a new protocol or just using some of the protocols that we have right now. So I actually just spoke to a team building a blockchain called Flurry, and it's also using some similar graph database technologies, some ideas a little bit different, but what it allows you to do is embed permissions into the actual blocks of the blockchain. And that would allow like you and me to share data and it's like totally there, you know, on the blockchain, but only you and I would be able to see it and only I would be able to read and write to your— to the data or the blocks that you actually share with me. Mm-hmm. So, you know, it could be a protocol that we build, but there's so many exciting protocols being built right now that, you know, I'm not an expert on that. Hopefully someone can explain to me like Our team did today.

Yeah, so kind of like a permissioning within— like, it— I mean, to me, if I, if I can kind of try to— I'm still trying to digest and get my head around this, but I'm thinking of like an anchor href pointer to a website is kind of how we do this today in HTML. And in an Athens context, there's some internal mechanism for using those braces and connecting an idea to a page that doesn't yet exist, then you can fill out the page, right? So then that page has something like an href link back to the source. Because even without the permissioning stuff, I'm just trying to imagine like this— there's a lot going on here, right? It could go in a lot of different directions. So I'm wondering how this new internet might look.

Yeah, you can go in a lot of different directions. And I've thought about them before. Like, oh, what if it's just an anchor tag that says, like, bidirectional equals true or something? Yeah. I mean, I don't know. Like, it's— people have thought a lot about this actually, though. Like, bidirectional links were originally considered for the first spec of the web by Tim Berners-Lee. But ultimately, it didn't work out. And, you know, it's likewise in the blogosphere for, like, WordPress sites or, like, Blogger, they also have plugins that actually allow some sort of bidirectionality when you're referencing another Blogger's blog. Right. So, like, there's some prior literature that we could look at. And, you know, there's so many ways to go about

it.

Not thinking too much about it yet, though.

Got it. Yeah. A little further out. So it sounds like some of the core early use cases for this, you mentioned, like, maybe the Highlighter team would have its own Athens Research instance. And then I think you also mentioned that like a traditional researcher, maybe like scientific researcher, somebody writing a paper at an institution, like an educational institution, is that kind of where most of the understanding of this process already exists? So they kind of, you're just like fitting into a workflow that they already understand and have demand for? Yeah,

I mean, I think that the, research community, like it resonates with them because they do get that you want to preserve that original reference and not replace it. And they're also just processing a huge amount of information. So that's number 2. And then number 3, it's like a lot of this research is quite exploratory and it's really cutting edge. So you don't actually know the structure of the information before you discover it. And so this is why I don't recommend like someone use, you know, us over Notion. For most of the time, you're just like creating this internal wiki. It's a bunch of stable information and it's just, you gotta just like have it there. Whereas when you're doing, you know, deep research on a topic, you're really just like, first off, it's best to just start off with like this daily log of like what you learn. And then you create like, you wrap it around and like, today I learned about machine learning. And then like a week later, you're like, okay, read another paper on machine learning. And like this, like whatever, some simulations. So eventually then you gather those references and you have this— it's not even really organized, but it's just a set essentially of references. And then like from that, you're like trying to make pattern match and eventually come to some sort of more organized structure. But, you know, with really exploratory research, you don't have that before. It's very emergent. And that's why I think it resonates more with the research community for of those reasons.

Yeah. Also, for you personally, like, have you had any kind of like second brain eureka moments that you've just been kind of doing your thing, taking notes, and then the fact that these references have come up to unexpected things have shaped your thinking in a cool way? Like, can you think of— I don't know.

So I think the biggest, the biggest aha moment for, for me, and I think for a lot of people that use some of these tools like Athens is when you're like, you found something super cool. So like, let's say like, oh man, I just found out about Highlighter. They're like talking about building in public and I heard that before. And so like I start typing like bracket bracket and start typing building in public and then I see that it's already there.

Mm-hmm.

But I had forgotten that I had written about public before.

Yeah.

And I click on it, autocomplete, click through it and I'm like, oh shoot, there's like, I took all these notes on building in public. Like, here's some Twitter people that write about it. These are like some of my favorite founders that write about it. And then you just have a huge amount of like serendipity. It all comes back to you. Like, that's, that's like my most like eureka, highest eureka moment when I use like Athens. Yeah. It's like forgetting and then remembering it and then diving back into that context and like connecting the two. So the eureka moments, like it's kind of based on forgetting. That's why it's kind of a hard trick. Because you have to use it for a while enough to forget it. But then it really, like, literally, it allows you to connect the dots. That's the whole point of, like, joining stuff, literally to connect the dots.

Yeah, totally. I also wonder about any, like, eureka moments in terms of, like, on a more meta level, how you're building this in the first place. So just, I mean, thinking about the benefits of open source to this particular project and things that have happened that wouldn't have been possible if you were organized in other ways? And also, how are you thinking about incentivizing members of the community and rewarding people who are doing a lot of work? We had Nadia Egbal on earlier with her awesome book on open source. So I'd love to hear your thoughts on all of that and how that's been changing.

Yeah, one of my favorite metaphors in computer science is, I think, called Conway's Law. And it's basically the Technology that you make is a reflection of like the organization you have. And I think that basically like Athens is super great because again, it allows for this emergent structure to arrive. And that's exactly how I feel about how our open source project has evolved over time. Just like started, I didn't think about it too much. If you wanna realize, I spent like 2 minutes thinking, coming up with the name Athens Research. Like I really just didn't think about it at all. And then people just arrived, like structure started emerging. It started like growing to the life of its own. And even now, like we're talking about this vision, I'm like telling you straight up, like, I don't know what it's gonna really be yet. This, 'cause this is an open-ended project. It's an open-ended platform tool. Yeah. So yeah, a lot of things emerged. And like you said, one of them was how do we get contributors incentivized? How do we, you know, distribute value? in a fair way. And one of the tools that we've used to pretty good success, I think, is Gitcoin bounties. So Gitcoin bounties are these tools, it's like really used by a lot of crypto projects, a lot of Ethereum projects, where they want to incentivize people to contribute. And, you know, it's like you don't have to be a crypto project to use them, you just have to be open source, although I do think most crypto projects use them. So what Yeah, now we have a bunch of issues. We paid out thousands of dollars to contributors for their contributions. And we're leaning even more heavily into that as a recruiting tool. So yeah, I mean, it's so funny. People are always saying how the interview process is so messed up. Like, I can't believe you have to do all these contrived LeetCode problems and whiteboard. And I'm just like, open source has solved this. You just contribute. Yeah. We actually work on a real-world problem that's not contrived and we actually a sense of our work styles and chemistry. And yeah, it's been such a great process. It's been super fun. And the best part is just the community and the contributors.

Yeah.

So how does the team internal at Athens Research— I guess I think you're structured as a company, right? Like a corporate entity. So there's some people who actually work at the entity and then some people who are open source contributors. So do people— are you telling us that people usually start in kind of an open source mode and then become kind of a full-time kind of teammate on the commercial entity? Is that—

Well, yeah. I mean, that's how I want it to be. Right now I am the only full-time employee. I have community developers currently under contract. And then we have a full-time engineer joining us in June. But everyone has been a contributor first, you know. Mm-hmm. And hopefully we can get more people full-time as well. But yeah, how we build is exactly like everything that you see. So 99% of our communications are in public GitHub issues, discussions, and Discord channels. Like, I actually— like, so if you go to our Discord, we've got, I think, 50 or so active.

It's amazing. Yeah, I've been poking around.

And, you know, every single Sunday we've got community calls. We've done that every single week this year, I think. And 11:00 AM Pacific, like, super fun. We've got like dozens of people joining us. Now we've started to do daily standups at morning at like 8:30 AM Pacific, and they're just open. Anyone joins. Anyone joins, it's so much fun. Like we've got like five, ten people in a daily standup. It's derailed sometimes, but it's super cool because then we have contributors like feeling like they're part of this whole thing and they are and they're learning. A lot of people just lurk and that's totally fine too because they're just like, "I'm learning, I'm learning, I'm learning." It's so much fun. And yeah, like we also have this like another great feature of Discord that Slack doesn't really have is video.

Oh no, you listen.

I think we lost you for a moment.

Oh, you're back.

All right, you said you were telling us a great feature of Discord.

Another great feature of Discord is video channels, uh, where you can just like hop in and like everyone can see someone's in here. So basically every single day we just have people in the coworking channel hanging out, and then people will join in, they'll like share a screen, they'll like share like interesting relevant blogs. And literally every single day it's completely— I mean, I don't listen to the Clubhouse anymore But it replaced Clubhouse when I was in Dora Lab because people would just hang out there like it was an open space, like a WeWork, and it was just so much fun. That still happens every single day.

Yeah. What are those Sunday calls like? What kinds of things have come up and what's been interesting about them to you?

Yeah, again, they're still emerging and some of them are pretty focused and Like, for instance, I'll talk about, okay, we just— I just came up with this write-up for how I'm thinking about the Bitcoin bounties, or we just made our fundraising announcement. Like, this is kind of an overview of what we're going to do next and who our investors were. And sometimes they're complete— completely unstructured hangouts. Yeah. Where we just like hang out and we're like talking about tools for thought and like the future of knowledge management. And those are like the most fun ones for me. Yeah, it's just a bunch of people talking about things they care about, just like an interest group. And really like my favorite part of the week. Like, I wouldn't give— I mean, at first maybe there are times like, oh man, I have to do this like Sunday morning, do I really want to do this? But I realized it's really like the best part of the week because everyone's here because they want to be here.

Yeah.

And do people oftentimes come with like questions or feature requests, or are people coming kind of because they want to participate or they want to hear from you, or kind of some mix of that?

Yeah, I mean, people throw in a feature request here or there, but, you know, we've got pretty good— we've got a lot of places on Discord and GitHub to file that. It's more just to check in and see how the community is doing. And, you know, it's really just a great intro to the community because the Discord— Discords are very overwhelming, and ours is It's extremely overwhelming if you're, like, new to it. And even the GitHub, you know, so many issues now. Just have way too many GitHub notifications. So it's really just a way to, like, make it really human. And, you know, with the open source and the increasing distributed world, like, open source has always been really great at the async distributed stuff. Yeah. But I think sometimes it goes too extreme in that. And lose some of the human touch without any of the synchronous stuff. And so the daily standups that are just open to the community, the weekly community calls, really cool.

I'm just popping up.

They're really awesome.

Amazing. Yeah, and I was also struck by— oh, here's the Hangout channel. Okay, so wait, I can't do this right now while I'm on this, but it's really cool. But I was also struck by just, I mean, it seems like you're just coding in public sometimes. Doing all of these kind of experiments with building in public, what kind of stuff have you been trying? Or are there things that you're thinking about trying that you haven't yet? Like, how do you think about more like public online building and stuff?

Yeah, so that's a huge fun part of it. I haven't— yeah, like 2, 3 weeks ago, I just did like a 4 or 5 hour session with some contributors and The designer who like designed it, we just like quickly iterated on the design and a bunch of people joined and then we put the recording out. So yeah, I would love to do more live streams for code. For a lot of like writing and like strategy stuff, like it's really easy to put out there and get quick feedback. Yeah, I don't know, but like probably just Everything that we've done, which is to continue doing everything we've done, which is, yeah, the live code, coworking, like community, like completely open to the public standups, community calls. We'd love to do more media as well. Like it'd be cool to have a podcast, like you guys have one. Yeah. And, you know, a newsletter that we're more consistent of, like more consistent Twitter updates for product releases and updates. Probably just adding, yeah, more of those channels and being more polished. So the 4-hour or 5-hour code stream I did could probably be condensed and edited by a professional to be 1 or 2 hours, make it really concise. And then just its own product for anyone who wants to learn Clojure.

Yeah.

Oh yeah. And then improving documentation a ton, which is really bad right now so that contributors can get their first PR in as easily as possible. And so that new employees can onboard, you know, without much friction. So, like, really going from contributor to employee, really it shouldn't be any— there should— it shouldn't be an event, really. It should just be the same thing. So putting all the docs basically the same publicly, whether you're at Athens, like, as part of the team literally or just a contributor.

Yeah.

You mentioned sort of writing the strategy docs and it actually makes me wonder, I guess I would assume that these probably are written on Athens. Is that—

Well, no. So our web presence, like that kind of stuff is still not great, but we were on Notion for a while. We still were migrating off of it, but I think we're gonna— right now we have a Ghost, which is like open source Substack, you could say, hosted. And then Yeah, just our GitHub discussions. Like, GitHub discussions is like really awesome for basically having like a forum, Reddit kind of like Quora style, right, discussion place that it's a lot more easy to track the conversations than Discord, right? Um, but yeah, we probably just stick to GitHub discussions, uh, get like our, our handbook, which is like Markdown based, and eventually I think it would be You know, people are always like, oh, aren't all the docs gonna be just on Athens? And like, yes, we can definitely have that public. Technically we still use like a public Roam for like just quick meeting notes. Hmm. But there's also like the right tool for the job, I think. And, you know, we can use Notion and like Roam and like Athens as a blog and as a project management tool and as a note-taking app and as a journaling thing. And like everything and they could just sync. But sometimes it just— like, I also like using just like the right tool for the job.

Sure.

Yeah, for an open source project where you have a ton of contributors, having GitHub as like a pull request review process and seeing commit history is like pretty still great at that.

Yeah.

How does permissioning on a document in Athens work? Is it If I'm the creator of the thing, I'm the only creator, or can everybody on my, you know, kind of like if we had a Highlighter team instance, anybody could chime in and comment or collaborate with me on that document? Or what's sort of the right way to get a mental model of what, what, how participation works?

Well, I mean, basically it would just be like a shared Google Doc or a shared Notion, and like everyone has the same permissions right now. We will build more complex permissions at the page level or the block level. But the most rudimentary is we all just get into the same Athens instance and we see who's working on what at that point in time. Like, oh, they're on this block, they're on that page. Eventually, we want to build the more complex permissions.

Yeah, we've been— we had— I don't know if you've seen Chris Piak's writing, but he's used Google Docs to get a bunch of feedback on some writing that he's doing that's kind of in an in-between state. It's not fully solidified and it's kind of not too far afield. He thinks it's kind of, you know, trying to solidify an idea. So he publishes a Google Doc and then posts it out to the internet and then has people kind of commenting and participating on that. But this format of kind of a participatory kind of collaboratively edited document feels like you know, it, you know, Google Docs is not necessarily the right home, but it's maybe one of the better ones, you know, that's widely understood and people have some context on it. I wonder if that— does that feel like the kind of use case that, you know, is already happening or would be early on Athens, or is that kind of a little bit more complex than where we are today?

Yeah, I mean, there's like so many ways to slice and dice collaboration. So there's like GitHub, which is more of like, I mean, really like the best for something atomic. And Google Docs is kind of like, it gets crazy after 100, I think.

Yeah.

And then there's more of like the free-for-all, which is like social media on Reddit or Twitter. It's pretty chaotic as well. And then there's like, yeah, I mean, so I guess with all those, like, we'll still see which one people want. Yeah, I don't really know how it's going to look.

Yeah, I'm also curious. This might be like a naive question, but how should— how do you think people, especially people new to the ecosystem, should think about Athens in relation to Roam and kind of like what's a better fit for what use case right now and just like the different ways both might evolve because of the structural differences in how you guys are building stuff?

Yeah, I mean, I would say that it's a really good question and a lot of people ask it. To me, like, Roam really did 3 things very well. had bidirectional links and block references, and they put it on an outliner, which is like the bullet. A lot of the other features, I don't— I think they're, you know, maybe not as useful to everyone. Like, a lot of them are very prototype-y, and I don't think that, you know, they really, like, move the needle as much as those first 3. So To me, like, we have that base, we've got those 3, and then there's like a lot of other features that, you know, if they were in Athens, I think they would make Athens worse, and I think they make Rome worse in some ways. So after that, there's a lot of possibility, there's a lot of, you know, opportunity and to do things differently. And, you know, if they really need those features, then they can always use Rome, but For now, like, we're still improving the core, so a lot of it doesn't even really matter. Just like collaboration, things like that, backend. Not even a huge product difference necessarily. Mm-hmm. But as it is now, it's like 80% of the value of Roam, like 20% of the features, and you get to see the code. You can change it, you can contribute to it, and eventually, yeah, like coming up very soon, you'll be able to self-host it and use it with other people. So A lot of people don't really use Roam collaboratively yet, even if you can, because some of the collaboration UI is not there yet. And like, yeah, we think there's a huge opportunity there for collaboration as well that is not being served by really anyone in the space. Yeah. There's a lot in the space that they call themselves like PKM. Again, that's like you're already limiting yourself to the personal brain. And really, I mean, the most I learn is With other people right now, right? Sharing knowledge, not just supporting it.

I wonder when you want to move to more collaborative use cases, I wonder if there are specific use cases that already come to mind that seem to have some heat or energy around them. Like, I think maybe Mishti had shared in chat about, like, you know, like the Bible being a book that a lot of people in the world are familiar with and have, you know, opinions on and want to discuss. And so I wonder if there's certain texts like that that would help kind of showcase the very best of a collaborative experience, or if you think that's going to happen more just with the emergent PKM documents that are currently there.

Yeah, um, I think the biggest success story, uh, actually has been like the Roam Book Club. I think they're like on their 3rd or 4th iteration, and basically what it is is Some guy, I forget his name right now, but he has a public Roam graph and then people basically take notes on books together and they reference each other's notes. And yeah, that's probably like the biggest success story for like wide-scale collaboration because I think 1,000+ people do it. And yeah, I mean, it actually feels like— reminded me of Genius actually, like back in the day.

Genius.

Mm-hmm. Like annotating the world's text. So yeah, I think that's a very interesting use case, but I don't know if the current UI will support that because like that's just not gonna work for more and more people. Once you have a huge number of people, well, it becomes read-heavy. Social media is is read-heavy, right? So 10 times people read than write. And, you know, for Athens, which is like this tool, it's really write-heavy. It's like focused on input. Yeah. And so the question is like, okay, once we have 20,000 people, 100,000 people, a million people, maybe the UI actually looks different. It's really write-heavy. It's like focused on input.

Yeah.

And so the question is like, okay, once we have 20,000 people, 100,000 people, a million people, maybe the UI actually looks different. It's not a note-taking app, but more of like the forum or the social media.

Mm-hmm.

Like, people have that. And there have been companies, the founder of like a really, really huge online community, like 20K people of like, you know, these like high-caliber like financiers, executives, like C-suite people. And they're saying, hey, we've got this like really, really intense, really smart community, and we want to use something like Athens to build up this global hive mind of people like that who can share these ideas. And coming up with these compelling claims. I don't know if an outliner, like a note-taking app, can support that.

Yeah.

Right. But can the same ideas translate so that it's essentially like a more rigorous Reddit or like a Twitter that is based more on merit than popularity, right? So that the best claims, the most compelling counterintuitive claims rise to the top rather than just like whoever is the most dramatic or the biggest provocateur.

Yeah. It seems like—

I didn't think about—

Meld right now, I feel like. I don't know if, Jeff, you know the early highlighter story, but I mean, I feel like a year ago, DK and I were talking so much about the Commonplace Book, and it was basically like a book highlighting tool for collaborative reading and talking so much about Roam and Like, what are ways that people could read and learn better together from the same texts? And then we eventually started hosting these book clubs on Zoom and seeing there was so much energy around video, kind of like what you're describing with your Sunday calls. It feels like where if you could make these their own assets and like start discussing pieces of video. So we went in that direction, but I don't know, I feel like we talk about this theme so much. It's—

Yeah.

Giving so much energy right now.

Yeah, it's such a big idea. Like, Genius had this huge ambition. A lot of the— I mean, I think a lot of the annotation extensions, like Hypothesis and WorldBrain/Memex, it's like a very similar compelling idea. It's just the execution, I guess, is really, really difficult. Yeah. And it's kind of like just a challenge of the space of dealing with knowledge, which is So abstract and so personal in a lot of ways.

And how have you— oh, I was just gonna ask a question about just like dealing with the kind of politics of being so open source. And like, I don't know, how have you— what have been tricky situations that have come up? And how have you kind of approached dealing with interpersonal, you know, like community dynamics in a way that makes everyone feel kind of included, but balances interests correctly and everything like that.

Yeah, I mean, I really worried about that stuff at the beginning, like governance. And I read a bunch of governance open source. I wrote this code of conduct. I wrote a governance document. None of it actually turned out to be a problem. Never had to kick anyone, never had to whatever. I think it will It's always kind of there, but we haven't had any huge issues flare up and we'll see how it is going forward. I imagine that there will be more potential challenges there, but right now it's like, okay, well, we don't even make any money, so let's just try to stick together and stay here so we can continue working on this. And I think that's a good forcing function, I guess, to be like, okay, like, let's just try to become sustainable and do that. And then, you know, that's really the focus. Maybe once there's like more money involved, it like becomes very complicated. But yeah, right now it's like, look, we're just trying to survive. So let's just say startup mentality together, which is like the awesome part about startups, like try to try

Yeah.

How do you think about the concept of sustainability? I imagine, you know, if you're raising money, there's some idea that there's a commercial entity that can build a business here. And do you have sort of a hypothesis around how you might serve people in a commercial environment, or is that kind of going to be figured out along the way?

Yeah, we have people paying us already. And to me, again, like you asked at the very beginning, it's like, oh, did you expect that there would be like this where you are today? And the answer is like, no, I didn't expect I was gonna come this far, but I always knew that there's a huge vision and the ability to create value was just like a non-issue. It's like, obviously this is so compelling. Mm-hmm. It's gonna make so many people's lives better and more productive and more creative and all these things. So to me, like, the value creation is going to be very, very easy. And, you know, if we just have to nail like some parts of the business model, which again, like, the simpler the better, and we're going to capture more than enough.

Yeah, that makes total sense. I'm also kind of curious just like who your inspirations have been. I mean, it seems like you have such a wide-ranging mind and you're surrounded by people who are not only interested in building this, but probably using it for such interesting research of their own. What kind of, I don't know, either people or books or ideas from history have influenced you the most as you're building this or today?

I don't know. I don't have a top 5, 10 list, I guess, of people I'm interested in. But really, I just try to absorb as as much information as you can from as many sources as possible. And, you know, there's a lot of classics like all the programs, essays, and, you know, doing YC, basically just getting wisdom from my peer founders and also my mentors like, you know, Michael Seibel and like, uh, and all the speakers that came, of course. Like, and yeah, like now getting into the space, having a lot of advisors and investors who have done the exact thing, like literally built very successful open source companies, like advisors who worked at Linux Foundation, Mozilla Foundation, and Ethereum Foundation, big commercial open source projects. It's like now it's more— now I have like a lot of resources to help me. So yeah, I don't know if there's a specific person in mind. Yeah, really everyone.

That's cool though. Yeah, I'm sure there are lots of interesting learnings that come out of those foundations, kind of their trajectories that are probably really useful. And also, I feel like a lot of people probably— I don't know if there's an open source— I don't know, is there an aggregation layer of these kinds of builders of open source communities that you've been tapping into, or is it more like one-off relationships?

Yeah, I mean, open source as a community, The category is definitely emerging. You know, in the YC group, we have a YC open source Slack and there's like 100+ founders there

and

like probably 50+ or 50+, yeah, open source companies that have just gone through YC. And, you know, we have like a monthly office hours there, you know, constant chatter through the Slack. So yeah, I mean, I don't know if there's like a straight up playbook yet. But yeah, I think a lot of the patterns are starting to emerge and you just gotta like collate them and figure it out. Things change pretty fast anyways. So yeah.

Yeah.

How was the— do you feel like the YC experience as an open source founder is substantially different than kind of other styles, or what have you learned in that process?

Yeah, I mean, there's definitely a good number of YC companies that have been open source and that number just continues to increase every year. That being said, when I was at YC, I felt a little misunderstood for having this community. And it wasn't necessarily because we were open source, more just because I didn't know what I was doing. I didn't know what we were optimizing for. I didn't know who our ideal customer is. Still figuring these things out. But overall, like, actually, I think about it, like, I don't, you know, go to a specific person when I'm, like, looking for help. It's just like, it's kind of based— I just ask a bunch of people specific questions and—

Yeah.

And I don't even maybe ask for that much help on, like, how we do the community development stuff because,

you

know, there is a lot of stuff that, you know, we are standing on the shoulders of giants for. But at the same time, like, I feel like a lot of Athens's rise has been pretty unique. And there aren't that many companies or projects, I think, that have gotten like this amount of momentum at this early of their life. And so that's why— and then the community takes on a life of its its own really. And again, there isn't like a— at least I haven't found a ton of models for it.

Are there like a few early things that kind of surprised you that they just, they were fortuitous and happened, right person, right time, right audience, that kind of really helped kickstart the whole movement?

Yeah, I mean, I'm surprised that people contributed, you know, like a month into it, I was like, oh wow, people People are helping. And they just continue to help. And, like, people that were, you know, 10, 15+ years of, like, engineering or Clojure experience or design experience being like, yo, like, what can I do? That's awesome. And that's really, really, like, whoa, like, crazy. Because it's like this person's the CTO of a Clojure startup. And, like, way better than I am at this. And they're asking me what they should do. And then that across all the other domains, like design, product management, copywriting, all these things that people are just trying to help out. And that's really just why we've been able to make as much progress as we have in this amount of time. It really could not have been possible without open source and just collective help and action.

Right. If you had kind of, based on the experience and the type of success you've had, and you certainly can't replay these things again, but if you met somebody who was just starting out an open source project today, do you feel like you have any kind of words of advice or wisdom that might help guide them on the path?

Yeah, a few people have asked me, like, hey, I'm thinking about doing an open source project, open sourcing my project. Kind of like what I tell them is like, just don't worry about it, about open source or not. Just like try to figure out what you are really bringing to the world. And then if it makes sense to open source it, then do it. Otherwise, like if you do it, it doesn't really matter one way or the other because like we all have this, just like these GitHub repos with zero stars and it's like, Yeah, it's open source. No one cares, right? It's like my .files, my vim config. It's like no one cares. So, yeah, I mean, if it really makes sense for the kind of project you're doing, again, it makes sense for us. We're building this kind of knowledge thing that's very personal to you or your company's or your organization's knowledge. Makes sense. Or if it's something that's very configurable, very great for developers, yeah. Yeah. Go ahead, open source it day one, and it'll make sense to those people. But, you know, more so than the open source specifically, you know, what is the value? What is— what do you want to work on? And open source can be part of that. And if it makes sense for what you're trying to build, then it can be a huge multiplier. But don't just, like, open source it because it's And the second one, you shouldn't just work on AI and blockchain because they're like— Right. Yeah,

yeah, yeah.

Are there any of the activities that you've done, like I think you mentioned like these, are they kind of like town hall community gatherings on Sundays? And I think some of the Discord kind of activities you're doing, are there some that have sort of surprised you to the upside that you might say, hey, if you're doing a new thing, you want to get a bunch of people together, you know, try a little bit of this or a little bit of that? Like the live coding may be another one.

Yeah, I mean, let's see, like how do you have these big Hangouts, like with a community? Is that the question?

Yeah, I guess like, you know, I'm just trying to imagine that, you know, so we're making a video and we're gonna create some highlights we're gonna share out. And so imagine that, you know, somebody is starting a new open source project or even a new commercial entity, and they're thinking about like, okay, I kind of have this core thing that I want to build for people, and I think that it's useful to me, or I think it's useful to other people, but I'm trying to figure out a few maybe unlocks. And I want to learn from Jeff what worked for Athens. Like, was the live coding and streaming a big part of something that really helped get a kind of visibility that may not be obvious to somebody building something new, which is maybe it feels risky to do live coding. And so maybe if you had advice for some of these activities that feel risky but are actually not that risky and turn out to be really productive, is there any advice that you might share?

Yeah, I mean, I want to think, like, if there was one thing that was, you know, 80/20. Mm-hmm. I feel like we just did a lot of things. Basically built a very, very welcoming community, especially to, like, new people to Clojure. Maybe that is one of the things that Like, basically early on, and you actually saw it in the Athens demo I showed you, like, why you should learn Clojure, my first month as Clojurean. What I realized getting into it was like, okay, no one knows Clojure. It's such a niche Lisp language. So the biggest bottleneck to getting contributions is actually going to be people knowing Clojure.

Hmm.

And because of that, I— and a number of contributors, like, we pioneered this thing called Closure Fam. Some people call it Closure Farm or Closure Friends. It's funny. Yeah, it was basically like this handbook to 4 or 5 other Closure resources on, like, how to go 0 to 100 on Closure and make your first contribution to Athens. And, you know, the curriculum isn't that great. To be honest, like, my fault. And we can improve it. But really, it was the values of learning in public and collaborating and teaching and learning from each other. Mm-hmm. And so, yeah, like, basically posting updates on how many, like, Clojure problems we solved that day. Like, we've got a Discord channel called Learn in Public. And actually, we have the standup channel, which is where we drop our async I think like Standup Notes. That used to be called Build in Public, actually. Come on. Yeah. And like, Clojure, like part of it says like learn in public. And it was like, okay, post every single day for 35 days what you did that day to do Clojure. So like, okay, you could do tweets, you could do a blog post. Some people did that. You could do a Discord message. And some people even did like coding livestreams with a few people.

Oh, cool.

So really just, I think, building the Maybe, yeah, again, maybe it was the values of like learning and collaboration and then like doing it in public, you know, being, you know, vulnerable perhaps of like looking dumb and like looking like a complete noob, like making that okay. I think fostered a pretty unique vibe to the community.

Yeah.

And yeah, you can kind of tell. I mean, when I go to other internet communities, like what the vibe is.

Yeah.

It's really hard to— it's really important to do. And yeah, it's very, very important.

Yeah.

Yeah.

That's great. I know we're already over time, so I don't want to keep you longer than we already allocated for time on your calendar. But I wanted to thank you for joining us today. This was a lot of fun. I feel like I come away with, you know, even new perspective on how we think about what we might do to help bring people together for Highlighter, because I think it is, you know, there's a lot of those kind of common themes in early-stage tools that I think you've sort of been through and had kind of a great deal of success and very organic kind of discovery and adoption. So thanks for sharing your story. We appreciate you spending the time with us.

Yeah, really inspired by what you're doing to build in public already. And I mean, we're hosting this the show, but I feel like there's so many things that I would love to try, just livestreaming and doing these group calls and things. And if you ever want to try playing around with, obviously, Highlighter to do any of this for video creation, you're more than welcome to.

Yeah, I was thinking, like, oh, we could use something like Highlighter for, like, the coding streams and our community calls.

Yeah, that'd be fun. We'd love to support you if that's something that you ever want to play with. We can We can be your ops team, we can be your on-air foil, we can interview you with your community, whatever you'd like. We like making videos all the time.

Great, yeah, well, we'll just— I'll send the contributing doc over.

Amazing.

Thanks, Jeff. Thanks for spending some time.

Thanks so much.

Have a great day, bye.

Take care.

I'll send the contributing doc over.

Amazing.

Thanks, Jeff. Thanks for spending some time.

Thanks so much. Have a great day, bye.

Take care.