Tech Brew Ride Home - (Bonus) BlueSky CEO Jay Graber

Episode Date: February 19, 2024

An interview with Bluesky CEO Jay Graber about the future of social media. Sponsors: FactorMeals.com/ride50 and code ride50 for 50% off Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to another bonus episode of the Tech Meme Ride Home. I'm your host, Brian McCullough, as always. I'm joined by your usual co-host, Chris Messina.

Hello, hello. And, you know, folks, you've heard me do, what, three or four segments about Blue Sky over the last two weeks on the show. So, hey, why not go right to the source? Our guest today is the CEO of Blue Sky. Jay Graber, Jay, thanks for coming on the show. Thanks for having me. So you guys got rid of the wait list, opened up the platform to the world last week.

What have you seen so far in terms of uptick people coming on? How's it going? Yeah, we've got, we almost doubled our user count. I think we're at 3.3 million when we opened up last Tuesday. Now we're at 4.8. So almost 5 million. Pretty good. Things are going well. Did the decision to finally open up, was it based on, you know,

obviously this was something that has a project that's been developing for, I think, three years, maybe more than that. Was it a question of you felt like you were feature complete enough that it was time to open up to everyone? Yeah, I think something actually people don't understand about our timeline, is we were first sort of just an idea within Twitter and didn't actually become a thing until I took it over mid-2020 and then we weren't really a company until we spun out in 2022. So it's been out two years. And for the first part of that, we were building a protocol for Twitter to run on.

And we didn't start building an app until it looked clear that Twitter was never going to adopt this protocol. So then we went ahead and built an app sort of like late 2022. So last year has been mostly been about rolling this out, rolling out the app on top of the protocol that we built the app protocol. And it took some time to build out the app protocol. It's really the rails for a sort of social experience where we tried to combine the ease and efficiency of centralized social with the openness and freedom to build of decentralized social. And it's just taking a while to get things right, but we've scaled up.

We're ready for more scale now. And we've built out a lot that's going to be rolling out in the next few weeks. So next week, actually we're starting federation, data federation. The feeds are already federated in the sense that you can go in and any third party can run their own feed and host their own custom algorithms. And then moderation services are coming later this month.

So federating out moderation to any third party service that wants to build a labler, an annotator, some way of giving input to the network. So those are all the things that we've kind of been working on over the past year and wanted to sort of bring it all out in a period of time, where we could really show people what the full vision is. So it's great to finally have this conversation, Jay.

Like you said, we've sort of been around each other, I think, on blue sky here and there. I've had kind of my own personal history with working on the social web and social networks for a long time. And what I find is so sort of interesting to watch both how you operate in the space and the things that you've been working on for the last several years, starting with like events and then working on like Zcash and working on a bunch of different like, like decentralized and federated technologies, including crypto, is that invariably, it feels like when we're designing social products, it's really balancing a set of tradeoffs. I sort of refer to this set of tradeoffs as like a complexity balloon. In other words, when you sort of squeeze one part of the complexity out of, say, the user

experience on one end, that complexity will invariably go to some other part of the system where you'll have to clean it up or sort of solve for it at some point later in the future. And so as I've read more of the blue sky technical docs and the white papers, it's clear to me that your threat model is a little bit different than what some of the other platforms are that are out there. So, for example, my background and experience is probably closer to what Mastodon has been doing with Activity Pub. Since Activity Streams was the predecessor to Activity Pub, and we did a lot of work to start out with things like Open ID and Oath back in 2007 through 10 or whatever. And so my question to you is, given the time that you've had,

I guess first working with an assumption that Twitter might adopt the AT protocol. You call it at protocol or AT protocol. What's the short form? ATP. It's a ad protocol. Yeah, and sometimes at proto for short. At proto, great. Okay, let's do that.

So if the idea as sort of Jack had originally suggested that Twitter should have just been a protocol, now you've built this thing out, now we've seen what happens when the captain of the ship sort of goes rogue, i.e. Elon buys the company. My question is sort of like both looking backwards, how would things have been different, do you think, with Twitter, had it actually been built on at Proto at the time? And then looking prospectively into the future, where do you see this kind of going next in terms of what this turns into? Are we talking all the way back like 10 years ago or back like 2021?

I think, you know, if we just go back to like 2021 when there was the first sense that the, again, like the people who are in charge of driving these ships suddenly, themselves where they're a threat model, where they became the man, so to speak. And that became problematic because of the incentive structures that they were bound within, became, I guess, adversarial to the user base, right? Because advertising started to take over. I mean, this is why Twitter shut down their platform. So I'm just trying to think through some of the logic and thinking that you've brought to Blue Sky as a demonstration, I suppose, of App Proto. And then where this kind of goes next as a corrective for the ways in which the social web has

developed and then devolved. Yeah, so a huge part of our motivating design philosophy is just make social more like the web begin. Like the open web back in the days when you had blogs, RSS feeds, email, XMPP. One of the inventors of XMP is on our board trying to advise us on this path of how to do it, you know, better this time. And I think something really changed when social platforms became the norm for a lot of people's socializing and content creation.

online because all of our identities and then our social graphs got piled onto these large platforms and we couldn't leave. And so the network effects were so strong because we don't have the account portability. You get stuck there. And then even when the social site does things you don't like, most people are fairly beholden to it. And then back when there were more open APIs, developers could come along, build different experiences, help you augment your experience. There was moderation tooling like block party built for Twitter. There were other experiences like tweet deck, you know, but a lot of these things got shut down over time and killed a lot of the innovation in the space. And that has continued on a downward arc over the past few years. And so we have

gone back to fully open APIs, fire hose of open data people can build on, trying to reset user expectations to this is all public. Some of the other attempts in the space like mass done activity pub, developed a culture of being much more sort of like local oriented, here's your local community, let's not build global fire hoses or global. aggregators of data. We leaned into this sort of public conversation nature of what Twitter used to be where it was real-time news streaming from all around the world. And so we tried to build that in a way where you could have the guarantees you get out of decentralization, although in order to build a global platform, you do need some points where you're like pulling all

the data together, right? So we have, we've built, we've kind of turned social apps inside out and turned it into a collection of microservices where this is how they all run on the back end, But here anyone can run the microservices that make up the social app. So anyone can run the relay. Anyone can host their own data. Or we'll be able to soon once we start federating with everyone who's self-posting, right? And so our goal was really to make it so something where you get the guarantees of decentralization, which give you basically locked open APIs, keeps the freedom to build for developers.

It gives users the freedom to choose because now you have these things like you have a tweet deck type of thing, deck blue, built on blue sky, and users can move between those. they're operating off the same protocols and we're not going to be shutting them out right and then creators get to keep their relationship with their audience which i think is really key in an era where people have built their followings on these platforms and then they become beholding to the platforms and unable to leave so many creators have gone back to optimizing around building their email list because email is one of those last open protocols that actually you know you still have a direct connection to your subscribers with and so we would like to have social be an open protocol again

at base and then there's services in the ecosystem, there's other things built on top, but ultimately there's like a set of things where you can build these services on, that stays open, resilient, and able to change because society's needs for a public conversation platform is constantly evolving. And when you bottle like innovation on one company, you can't evolve as fast as you need to, I think. I think it's super interesting just to think about both like the locus of control at the level of the individual. Like you suggested, in the early social web, you know, we, many of us had our own blogs and we would syndicate through RSS. You could follow anybody. You'd have a feed reader and would

sort of bring those things into a type of special formed inbox that will allow you to basically catch up with people no matter which service provider or which domain they happen to be hosted on. And gradually, there was sort of like the mollification, M-A-L-L of the social web where it became a much more commercial space. And I think one of the things that's interesting again to think about these tradeoffs with what Blue Sky, or at least at Proto, is suggesting with the PDS. It's a personal data store. Is that right? Yeah. So essentially, you kind of have like this, you might think about it like a reverse inbox where essentially it's a collection of all the things that you've, you know, posted all the letters that you've sent out to all the op-ed,

you know, sort of centers of the internet. And if your, you know, publisher of these letters goes rogue again. If you want to move, you can do so by utilizing DNS. So sort of going back to this infrastructure that's there, that's a global international service that makes it possible for you to leave. So you guys didn't focus so much on like usernames that are tied to a domain as in the Macedon instance case where I'm like Christmasina at masadon.xyz. On blue sky, I'm just Christmasina.com. Me, which is my domain. I control it. If I don't like the way that you're running the service, I can just point it to a new instance using DNS. And then I've moved.

And in theory, all of my followers don't have to sign up to anything new. They just continue to get my updates. That's roughly how it works, yes? So my question- Yeah, we decided to use the DID standard, which is this proposal that emerged over the last few years for a decentralized identifier. That supports lots of different methods. And then we supported a placeholder method and the did web method, which you do have to have your own domain name for that. But then if you really want control over it yourself, you have this option.

And then over time, more of these methods can get added. Right. Which, I mean, honestly, that's actually what we ended up with with Open ID and Open ID Connect many years ago. It's just because it's already there. My question is or reflects a recent experience that Casey Newton had when he decided to leave Substack. Because there is a conversation in a story that says that, well, if you have your email addresses of your subscribers, you can move from one host to another until the Nazis show up. And obviously, in Casey's case, the Nazis showed up on Substack, and he decided to move off.

So my question is, with this portability, with the idea that you can move your content, especially this is a great reset for the creator economy and how people, I guess, can take whatever they've worked on and the following that they've developed. When monetization becomes a piece of that, and where a substack is the payment processor for subscriptions, how does that work in the AT protocol world? I know that, like, you, as far as I understand it, like, there isn't monetization built into AT protocol yet, and there's some tipping that's been done, but when it comes to actual like subscribers, how do you see that developing if someone is to move from one PDS to another?

Yeah, so there's services in the ecosystem like the app that is, you know, the app is essentially a service. The PDS is a service. All these things are different services. We kind of run a bundle of them. And then we'd be providing that to users, you know, as a bundle like here, we'll host your data, we'll, you know, give you a nice experience through the app. These are all things that we've bundled together.

We've bundled together, you know, moderation, a set of some feeds. And then we give you access to this much broader marketplace, this window and all these different feeds and different moderation providers, right? And so that's a direction that we're moving to sort of be a service provider in the ecosystem. But then if something happened again, like it happened with Twitter, like, you know, leadership changes or the company drastically goes a different way. The goal with this account portability, as you mentioned, is to give people a way to move to a different service provider. And so rather than, oh, no, like the social site went down. Now my whole social identity is dead and they have to start over again. Our goal is to get this to a place where it's the last social identity you'll have to create because there's another app out there that'll be waiting to take our place and then users can seamlessly migrate over.

And this kind of being seamless, you know, migration path might also let things fragment and come back together more gracefully. So people are upset at a company, they leave. And then if they want to try it again, they go back. And it doesn't disrupt all of their content and their relationships. It kind of goes with them. So if we had this, you know, if Twitter had been running on the ad protocol at the time that leadership changed, People could have, you know, taken the whole history of their content, all their posts, all their followers, their username, moved off to another service. And then if, you know, Twitter stabilized, they could move back.

You know, it would just be like changing between cell phone service providers. Like, if we were locked in and had to change our contact list and our phone number, every time that we changed from AT&T to T-Mobile, you would really be stuck. And so in other parts of life, I think we've started to decide that if this communications infrastructure is important, important enough, people should have the right to really choose between service providers, but I think we're still getting there with social. And part of this push towards protocols instead of platforms is to push for that kind of user choice and user control. It seems like, and I think it might just be like the maturity of the EVE protocol, like

and maybe it'll evolve over the next couple of years, but as more content ends up behind paywalls and the fact that AT protocol presumed that the content is sort of of locked open, I think, is my understanding, like that all of my content that I'm publishing essentially goes to sort of like a public repository of my stuff, that the paywall mechanism isn't defined yet, would be maybe the most like sort of generous way to think about it? Is that right? I would say it's not defined, not it's not impossible. It's just not defined because in an open service where all these different, you know,

feed consumers, all these different, you know, services running are all using the same data, they need to have access to it or at least access the fundamentals. And when you want privacy there, it's either a coordination game among all these services or you're encrypting things. So I think that the most protocol robust way is to encrypt things and then give subscribers the key, right? But it's a little bit more technical complexity there.

And we've started off focusing on the global public conversation because there already were good protocols out there for smaller, more private conversations. There's matrix, there's signal protocol. So we looked at those. Settlebutt, yeah, yeah. And so we're definitely inspired by those. And we made sure to design things so that we have keys associated with accounts and,

you know, sort of a structure where later we can move in that direction. Basically, people's user data is stored in a way that's very inspired by Git. It's a user repository. And then your personal data server is much like GitHub where you can use GitHub or you can self-host. Yeah, I think like what I'm noticing in this conversation is there's this constant tension between kind of monetization and making a living through producing content for the social web versus just being a participant in the social web. And it's much easier to solve for the latter in a way because the monetization stuff requires

you to kind of invent artificial scarcity. Obviously, that was the whole like NFT thing for a moment, where you impose a restricted access to content or you make it scarce such that people are willing to pay for it. And so it is orthogonal to the distribution of that content, but it seems like with a way that at protocol is currently specified. It's like there's not an obvious answer to someone who, let's say, like Ben Thompson is publishing to Stratory and he wants to have paywall articles that if he were to move his PDS, that his subscribers, his paying subscribers would

automatically be able to follow him at his new PDS with everything intact without doing any extra work. So let me pause there. Let's just say that that's a little undefined, which is fine. The question that I was going to ask on the other side is, One of the most compelling and interesting aspects of Blue Sky, of course, is Composable Moderation, which I feel like you've been talking about for a very long time. And so it must, one, feel gratifying to come to this place where you can now show people what

you're talking about. And then, two, my question sort of on the flip side of what I was just asking about when it comes to creators is what you see in terms of opportunities for businesses to become, whether it's like labelers or whether it is like block party or whether it's like a a VPN style or ad block style business model where essentially there's people coming into the network and they're putting together composable feeds or moderated feeds to provide people with a better experience. Do you see a way or how that's going to build out or I guess over the next year or is it still a little too soon to be thinking about those types of opportunities?

Yeah, I think that's a pretty clear path forward for things. I mean, moderation is something that offers value to the ecosystem. It's something that there's, you know, hasn't been enough new approaches tried, I think. And generally, the centralized companies have been the bottlenecks for change. So it's, you know, taken a while to figure out how to explain our approach to moderation. And I think we're still figuring out how to best explain it. But, you know, for a while, people were under the misconception that we didn't moderate because we were decentralized or something. And that's not at all true. We have a moderation team. And basically, essentially what we've done is we've tried to, it's a, it's composable because it's like building blocks that stack together.

And so we've laid a foundation with the Blue Sky app where we're doing moderation of the set of baseline. And the Blue Sky app is like a service that you sign up to use. And then we give you a package of defaults and you can customize from there. And then the app protocol ecosystem is saying there's this other layer beneath even the Blue Sky app that other apps can build on and stack their own building blocks and operate outside of our foundation if they want. But in the Blue Sky app, here's sort of like what we've built and then just like we I think custom feeds going first was the greatest example to show people because people also didn't understand that until they saw it which is well we give you a default you have a feed when you sign up and you have an algorithmic

feed we give you discover it's an algorithmic feed but then you can browse 25 000 other algorithm feeds and then you know substitute out the you know for the feeds that you want like maybe you just want to see cat pictures there's like five cat picture feeds you know and so people can see how much you can start to like tailor things and be creative within the realm of the curation, which is surfacing information to you out of all the data flowing through the network. Moderation is the other side of the same coin. It's suppressing data, hiding it away, not promoting it, right? So, curation, all these feeds like, you know, dredge stuff up for you to see, and then moderation tucks it away and suppresses it. And so the baseline we've said in Blue Sky app defines,

like, what are the boundaries of like what is the stuff we're even going to show? And then on top of that, you get to customize with all these labelers that are coming soon, which is going to work much like custom feeds. And then outside of that whole foundation, like in parallel, you can run your own services and you can have infrastructure autonomy. So if you really just want to index part of the network and have a small community that doesn't tap into global fire hose, sets its own rules, does its own thing, you can do that and it's protocol compliant, although whether or not it talks to the rest of the instances is really a matter of like, you know, governance and social consensus. I guess like, and that's where it becomes really interesting, right? Because again, you have sort of on the one end,

the global town square. And then on the other hand, you have like the individual that's like, you know, publishing more or less like blog posts or misses, you know, whatever into their PDS. Then like, I think what's interesting about the Macedon approach is that these are more like, almost like townships or like smaller sort of, you know, groups of people coming together to decide what their values might be and then coordinating around that. So I guess my question there is like when I think about, I don't know, one to two years and, you know, in the future, and I think about people maybe using Blue Sky, if they're generating their own version of composable moderation,

I know one of the things that's very important to you is like user experience and ease of use. What is the process by which someone at an individual level comes up with their own composed moderation beads? It's easy enough to imagine just adding accounts to a list, like a save list, but composable moderation can actually go much, be much more complex. It's literally, it can be an algorithm.

be an algorithm, but that sounds like either writing a bunch of like reg X's or something even more powerful with like, you know, machine learning. What is the level of complexity that a user starting in the social web in 2027 should be expected to know how to do in order to get a good experience out of this new, more diverse social web? So this is really exciting. I think there's so many possibilities here. And again, it's, I think seeing the path towards user adoption is possible when you look at what's happened with custom feeds. And so at first you had to be a developer to consume the fire hose to produce a custom feed. And then someone came along and built Sky feeds and a few other things, which are a custom feed builders that give you

a graphical user interface like a control panel where you can go in. And without knowing how to code, you can start building custom feeds that do things based off lists, based off hashtags, based off words, based off regular expressions, based off machine learning. And these tools are getting better and better and creating more options for people who want to be creative, have an idea for a feed, but don't know how to code.

And I was talking to a friend the other day and with the state of machine learning, you're probably going to be able to, in 2027, do something like, you know, tell a service, I want a feed that shows me this or for moderation that hides this, like hide all these topics and you just maybe just tell it in words, it transcribes it, builds an algorithm based off that, and then you tweak it until you get what you want. That's probably going to be possible in a few years. And so having interfaces where you might have to not even have to type, you just talk to your computer and then you like get feed that you want. That would be something I think would be very cool and

it would be possible to build. So same for moderation. I think, you know, when we first roll this out, you're going to have to be at least stable. We have, we're going to be open sourcing some software that we've built to sort of manage and tag things and label things. So you won't have to be a developer to build your own software, you'll just have to run it. And then the next step will be probably people will be adding on to that open source code. They'll be building their own software to do things in their own way. And then probably third party things will emerge like Sky Feeds, but for moderation, that let users who don't know how to code set their own moderation criteria.

And then over time, you know, that will get even more complex and even more refined. Got it. Okay. Two more questions. Since you mentioned it, one of the things that, of course, I've gone back and forth with Paul at Blue Skagit about is hashtags. Blue sky currently does not support hashtags. And yet you just mentioned that they can be used for creating these moderated feats, or yes, composed defeats. So you can also deflect and not answer my question. But I'm just curious, like, what is the thinking there?

And is there any possibility that internally, Blue Sky would linkify hashtags or is just that's not going to happen? It actually is on its way. And yeah, and we kind of want to wanted to also give it some time to like see what would happen with custom feeds. Like that was the main thing I wanted to get out first to show people, you know, what was new. And people started organically once again using hashtags in feeds as a signifier. So hashtags actually do exist in the sense that you can build feeds based off hashtags.

And that's the way a lot of people tag things into feeds through hashtags and emojis. But linkifying them as a logical next step. And I think, you know, it's really cool to us, the history of how these things emerged back when Twitter was an open ecosystem. And you know, like you could just invent the hashtag. That's awesome. And then the company would, you know, adopted. And it was sort of like, you know, follow where the user base goes and where the culture goes. Well, I mean, the company resisted it.

So, you know, and then eventually, you know, Twitter acquired some eyes and some other apps. And that's how it got in. So it's always, it's always the Trojan horse. Yeah. Yeah. Yeah. But I love that it was like kind of, you know, it came from, you know, people just being creative. And I think there's already a, you know, a pull request open.

Because the app is open source. People have made a request to linkify hashtags. And we actually have one that's further long. Yeah, yeah. You know, timelines are always in flux at an org that's like fast-moving and juggling all these things. But I do know that it's on its way. Okay, great. So my last question then is just about the types of social apps that you are anticipating,

or perhaps that you hope to see the community build on the app protocol. I know that it's possible to build things like Reddit or to build blogging software. So I think one of the challenges, of course, with app protocol is people see it, they see blue sky, and then they think, okay, this is like Twitter, pre-Elon, great. That's the end of the story. But in fact, as you've said, this is meant to be a technology that enables a thousand different flowers or app flowers to bloom. So of all the thousand app flowers, which are the ones that you're most optimistically

looking forward to, let's say, in the rest of the year? Well, I know that images are easy to support the things within the sort of content forms that we've already done. It's possible to support video and stuff too, but actually the challenge there is more, you know, just moderation and making sure that users have a good experience. It's not so much technical limitation of like supporting it at the protocol level. It's the social coordination, you know, making sure that we're like getting a good experience out of it.

So I think, yeah, I'd be excited to see some things of merge in parallel as as we open up Federation, we're also enabling third-party lexicons, which are sort of like specifications of how you go in and define new protocol like record types. I think it will be a period of experimentation as people play around with those. And it's always a juggle to sort of like guide, you know, protocol governance in a direction where everything comes together in a cohesive way and makes sense. And so we're talking about, you know, how do we go through that process and make sure that when all these different apps bloom, it's not just chaotic and you're confusing to users.

But we're excited to see a lot of merge. And so, yeah, different kind of conversation types, different record types, different kinds of apps, long form as well as short form, I think, would be very exciting to see. And then the experimentation with radically different types, like early on, we had somebody build a user interface where you walked through your timeline in a third three-dimensional VR space. And there was like an AI, you know, anime character that summarized what was on on your timeline for you and like follow you around.

I love the idea that people can just be like using the social app. Emitting or is like these little 3D characters like little chat bots like every time, you know, the J-g-a-bot like you know tweets or posts or skeets or whatever you want to call it. Like it's like a 3D character that you're like in a room with. Okay, that's cool. Totally. The first vision pro app like this guy. Yeah, I want to see radically new experiences like that. Exactly. Yeah.

Yeah. And just like see things emerge that like do things outside the box of how social is being done because suddenly innovation is like unbounded. Like we also had a sky spaces experimentation, which is like a clubhouse thing. And because we designed everything to be composable, you don't have to use the whole data model. You can just use the identity thing. And we just put out a proposal for an Oaf spec for like how you can like log into these

things. And the goal would be for your app protocol identity to be something you can log in across all sorts of new apps. And then what we've done is we've built up, you know, this first user base on this first sort of app. But then the next app that comes along will be able to draw users from the same user base and they get to keep their relationships. Even if you're not operating over the same data, maybe you can jump into the Sky Spaces app and talk to all of your friends that you made on blue sky.

And so on and so on. And it all builds on itself until you get a much more vibrant social ecosystem. I think it. Two real quick final questions from me. Curious for your take on the controversy that's happening on threads right now, vis-a-vis apparently, like, you know, demoting political threads and things like that. that there's the, obviously, the, the censorship angle or like the certain content is not, but also like a lot of people that maybe jumped over from X slash Twitter because this was the place where they were, their creators in the political space and they want to do things

and now they feel like the rug has been pulled out for none of them. I'm just curious your take on that controversy. Yeah, I mean, I think this is one of the types of, many types of problems that emerge when you have one algorithm run by one company. It's sort of a black box. The company can do whatever they want. And users don't really have a choice. And the goal of building an algorithmic choice at the start with Blue Sky was so that you can always choose what kind of feed you're going to get.

You can control your scroll. You know, you can pick to have a, pick a very political social experience where you're describing to all the, you know, politics feeds, following all the trending topics there. Or you can filter it out entirely. And we just want the tooling to get very. better and better so people can control their experience. And so two people eventually using the same app, actually I think it's already quite like this.

Two people using the same blue sky app could, one would be having a very cozy, quiet experience, no politics, just seeing their friends posts and maybe like pictures of moss and cats. And then somebody else could be following, you know, trending topics, Super Bowl discourse, Paul and Mossy Cats. Whatever is going on.

Like who doesn't love Mossy Cats? Exactly. So there's some of my favorite feeds, just admittedly. But I actually really like being able to swipe between both Like, I go over to follow, you know, trending topics on Discover and for you and these like mindful moss and then chaotic politics and you just, yeah, right. When I want a mindful moment, I go over to the moss feed and it's usually just like posts of moss with like three likes that people have taken on their hikes.

And I go in and I like all of them, you know, and then people are like, how did the CEO find my post? I have 12 followers and I took a picture on a hike. And I'm like, well, there's all these niche feeds that surface content in like weird ways. And finally, finally, you mentioned AI obliquely and talking about like, tell the algorithm, this is what I want to see, that sort of thing or whatever. It's very early for AI in the context of social media and things like that. But what are you seeing in terms of how AI could transform what social even means?

How are you all thinking about this as a potential paradigm shift for social media itself? I think it's something where we're in a period of huge flux and change, not just the fact that there's this transition of governance essentially when we're trying to talk about, like, how do we open up social platforms again, make them decentralized, but also we have new actors on social networks, which are, you know, AI agents, eight bots like you know all sorts of new sort of chaotic vectors of change being introduced and i think it's very hard to evolve platforms to keep up with it part of the hope of having an open ecosystem is that we can evolve faster because anyone can come along knowing the state of the

art and just plug in and start building something around the interfaces that matter like creation moderation new clients and also potentially there's ability for innovation to be accelerated because you can write code not just algorithms but maybe entire clients, you know, entire, you know, new paradigms of doing things faster and faster with the sort of like machine learning augmentation that you're going to get. And so you could see, you know, potentially a new explosion of innovation in the space if we don't close it down, where you're suddenly able to create all sorts of different ways of interfacing with your social data. The data is just like a sort of, you know, shared public conversation space that were applying all these different lenses to,

all these different views, not just algorithms, but like different apps, different, you know, moderation filters. And then my hope is that we can harness the collective intelligence of all the developers and users and people out there who really want to have a space online to connect and communicate and they want it to work for them. And the struggle is always getting the centralized companies to do it in a way that works for them. And in an open source environment, they can just take things into their own hands and start moving changes before the centralized companies catch up. Jay, I've sat on the show a couple times now. I think y'all are doing the most interesting stuff right now in social. So thanks for coming on the show to tell us about

that. Thank you.