Eli Mallon - Streamplace and the future of decentralized video streaming

devtools.fm April 27, 2025
Source

{/ TAB: SHOW NOTES /}

This week we're joined by Eli Mallon, founder of Streamplace, a decentralized video streaming platform. In this episode we discuss the future of decentralized video streaming, the challenges of building a new platform, and how developers can leverage Streamplace's tools. We also go into Atproto, the decentralized social media network that stream place is built on.

Episode sponsored By WorkOS (https://workos.com) and Mailtrap (https://l.rw.rw/devtools_2)

Become a paid subscriber our patreon, spotify, or apple podcasts for the full episode.

{/ LINKS /}

{/ Paste show notes /}

{/ TAB: SECTIONS /}

[00:00:00] Introduction [00:01:46] Understanding Decentralized Video and Live Peer [00:05:06] Ad [00:06:11] The Birth of Streamplace [00:18:53] Ad 2 [00:19:16] Monetization Challenges in Decentralized Platforms [00:28:01] Navigating the Crypto and Video Worlds [00:29:46] Stream Place as a Dev tool [00:47:26] The Future of Social Media and Decentralization [00:52:42] Conclusion and Future Prospects

{/ TAB: TRANSCRIPT /}

Eli: what really appeals to me about the space. They're not looking to just build one more centralized platform that gets acquired and forgotten about over the course of years.

I really want to be building things that could endure for decades or hundreds of years, um, when people are looking back.

[00:00:19] Introduction

Andrew: Hello, welcome to Dev Tools fm. This is a podcast about developer tools and the people who make 'em. I'm Andrew, and this is my co-host Justin.

Justin: Hey everyone, uh, we are really excited to have Eli Mallon on with us. Uh, so Eli, uh, you're building this thing called Stream Place. Uh, we're really excited to talk about that. Uh, but before we dive into talking about that and what you've worked on in the past, would you like to tell our audience a little bit more about yourself?

Eli: Sure. Uh, yeah, my name's Eli Mallon. Um, I have been working in decentralized video in one way or another since 20 16, 20 17 or so. Um, try to get my own startup going on, that kind of thing. Um. The whole time it's been the, the, the plan has been, uh, figure out how to, uh, more or less break Twitch's monopoly of, uh, of certain kinds of, of live streaming.

I love the streamers on that platform, but don't love that platform for, for lots of different ways. Um, so we can get into that, but. I tried to get a startup started for that originally. Um, doing that led me to Live Pier where I was working on, um, uh, decentralized video streaming in the Web3 space for a long time.

And, uh, since last July I've been working full-time on Stream Place, which is live video for the app protocol. So if we are, um, if, uh, if Blue Sky's trying to make Twitter using the app protocol, we are trying to make Twitch using the app protocol is the short explanation of that.

Andrew: Cool.

[00:01:46] Understanding Decentralized Video and Live Peer

Andrew: Before we dive into Stream Place, I wanna like dig into what Live peer is. 'cause like you say, decentralized video, and I'm like, what, what does that actually mean? Like, in practice and like, uh, yeah. So let, let's just start off with that.

Eli: Yeah, so live, here's a few different things, but the core of it is a, uh, decentralized network of orchestrators. So if you are, um, if you're a video streaming platform in particular, if you're what we would call a UGC platform, right? User generated content, something like Twitch, where it's lots of, um, you're not trying to broadcast the World Cup to 2 billion people at once.

What you have are lots and lots of, you know, thousands and thousands of streamers. With medium sized audiences, um, the biggest pain point for you in that situation is going to be transcoding, you know, to, to effectively provide video over the internet. You need to take in, you know, a 10 80 p or 4K source process, it turn it into seven 20 p two 40 P output.

So even on a phone, on a really bad cell connection in a developing country, you can still see a, a really blurry version of the output. Um, you only need to do that once per input stream. But, um, if you are, uh, uh, and so if you're doing the World Cup, you don't care. But if you're, that you need, if you're a, if you're a UG, excuse me, if you're a UGC platform, you need to, uh, do it once per stream and that gets very, very expensive.

Um. When I started out in this, I haven't checked this number in a while, but when I started out in this Google and Amazon were charging about a dollar an hour to process this kind of video, which is silly because consumer grade, Nvidia video cards among other hardware can do this sort of processing and pretty inexpensively, right?

We benchmarked, um. Like a, like a 1660 ti, which is a pretty old like mid-grade Nvidia card at this point. And, uh, saw that it was, um, uh, that could handle like processing about 12 of these streams. I. So the thinking behind the live peer decentralized network of orchestrators is you send out this video rather than to a, a centralized platform.

You send it out just into the world, um, into, uh, uh, different people's computers that can do this sort of processing. They send it back to you and, um, they get paid via cryptocurrency is the, is the, the short version of that and it's effective. Um, again, this is, these numbers are a little bit of out, out of date.

But, um, last time I checked it was, um, compared to about a dollar an hour for Amazon and Google. Um, the live peer network could process it for about 17 cents an hour. So it's like pretty radical cost reduction.

Justin: So how like latency sensitive are the. These like applications, it seems like that'd be like a big thing if you're doing a live stream. I know you have like a little bit of delay between when you're recording and when it's going out anyway, but like what is the like latency sensitivity in this space?

Eli: Yeah, it's, it's, uh, it's pretty good. Um, the, the nice thing about a decentralized network is you can hopefully, um, find an orchestrator that's very near to you, um, geographically, so that helps help save on everything. Um, but uh, yeah, I mean our target is always just like fast was always historically faster than real time, right?

So if you send in or live streaming platform, if you send me one second to video. I need to get it processed in one second or less, or somebody's gonna have a bad time. Um, so, uh, that's, that's the target for it. And it's usually able to hit that. Yeah.

[00:05:06] Ad

Justin: A big thanks to Work OS for sponsoring this episode of Dev Tools fm. If you're gearing up to sell to bigger customers, there's a moment when you realize it's not just about the features anymore. It's also about security, compliance, IT integrations, stuff you probably really didn't plan for work. OS makes that moment way easier.

It's a complete platform for enterprise features. So think SSO with SAML or OAuth, SEIM, uh, directory Sync Audit Logs, fine Grain auth. You name it, but the best part is how fast it can get you there. Workwise gives you clean, modern APIs and great SDKs across most major languages. Um, and genuinely well written docs.

It feels less like wrestling with an integration and more like flipping a switch. You need an admin portal. They've got one need. Secure storage for credentials that's built in need, domain verification or webhook events. They've got it. Um, so every piece of Work West is modular, polished, and built with developers' mind so you can stay focused on building your product and not getting buried in enterprise requirements.

That sounds good to you. You should check 'em out. Go to work west.com and again, thanks to Work West team for sponsoring us.

[00:06:11] The Birth of Streamplace

Andrew: Cool. So, uh, that's where you did work, but uh, while you were there, the idea for Stream Place came to you. So how, how did Stream Place come out of your work at Live Peer.

Eli: Yeah, so there were, there were a couple different things. Um, my explanation of that is always live peer had an era in there where we sponsored lots and lots of hackathons all over the world. I personally attended hackathons on four different continents, and so I was always there sitting at the live peer booth and people would come up to us and they would say things like, oh, live peer, that's like decentralized twitch or decentralized YouTube.

Right? And. Because it was this processing platform, we had to give what I thought was a fairly uncompelling answer, right? We had to say, well, it's more like infrastructure upon which somebody else could build decentralized twitcher YouTube, and what's your idea for this sort of thing? Uh, and I always wanted to be like, yeah, it's like decentralized twitcher YouTube and here's what a decentralized live stream is and here's what a decentralized VAD is.

Um, but. For a variety of reasons. Live peer itself, like Live Peer Inc. Where I worked, um, didn't, uh, wanna get into that sort of thing. So, um, another thing that happened around that time was, um, the live peer cryptocurrency. Uh, LPT had established a treasury. Um, basically, uh, a percentage of LPT every day goes into this big pool.

You can submit proposals to that treasury. And, um, get awarded, um, seed money essentially for, for different ventures that they thought would be beneficial to the live peer network. So, um, that was sort of the area I wanted to go coincided with that happening. Um, and there was just sort of an opportunity there.

So. I pitched them on the idea of doing a project that was a little bit more focused on social video, figuring out and inventing, you know, what does a decentralized live stream actually mean? What does a decentralized bot actually mean? Um, and they approved it. And, uh, I left the company to start Stream Place in July of last year.

Andrew: Were there any other co uh, like people trying to do this, this same sort of thing of like recreate, uh, other social networks on this decentralized, primitive.

Eli: Yeah, there, there's been a few different projects over the years. Um, I would say, um, there's a, there's a few in the Web3 space, uh, there's, uh, the, the, the biggest one in the activity pub space is called. Well there, well there's Peer Tube, which was sort of their YouTube re-implementation and, um, own cast, which is sort of a, a, you know, host your own, host, your own Twitch stream kind of, kind of thing.

And I, I liked both of those projects, but, um, I would say it's just sort of similar to, um. Uh, blue Skyes Critique of Activity Activity Pub, where it's sort of designed for individual servers and individual people to be, to be exchanging data and not really for like a, like a big world. Like here's, here's all the streams out there in the world right now.

Um, um, that kind of thing. So, um, yeah, for the, for the same reasons we were sort of searching for a different primitive, um, a different way to, to represent all of this, which led us directly to, um, the app protocol and, and Blue Sky and that sort of thing.

Justin: So where, where does, uh, the app protocol come in? So you're, you're building this decentralized social video platform, um, app protocol is like very open. You can build a lot of social products on it. Like how does that play into video?

Eli: Yeah, absolutely. So, um, you know, sort of our struggle a lot of the time was we had sort of video primitives. Um, we, you know, I, I knew how to make a live stream. I knew how to upload a video. Send it to people. Um, but then the gap between that and something like a Twitch platform is not necessarily video knowhow.

It's, uh, what does it mean to follow somebody? How do you get notified when somebody goes live? What are, what are what, what are your friends? Uh. If I'm gonna host another user, you know, I'm done live streaming, I'm gonna host or raid another user. Uh, how, what is that user right? Even, right, like what is, what is that other, what is that other piece?

Um, and so where the app protocol became a very natural fit was in, um, this sort of preexisting social graph. That we could use to bootstrap the network. Um, and, uh, have really clear answers to all of those questions immediately. Um, as well as what I think is a really compelling, decentralized architecture in the, in the PDS fleet.

Um. Uh, you know, the, the reading the app protocol for the first time for me was so interesting in that it was so clearly written by people that had, uh, tried scuttlebutt, tried noster, tried crypto in the Web3 space, tried all of these different things, and been frustrated with all of them, which I had. I had been sort of following all of those communities and, and experienced frustration and they sort of landed at this place where, um.

It's incredibly decentralized. I operate my own PDS, I have custody of my own signing keys. Nobody else can forge my signature and, and, uh, and claim to be Eli on, on, uh, or at ie. Lie on the at protocol. Um, but I. With, uh, with enough compromises in there that you can still have an email password sign up, which is what people are actually looking for, right?

And you don't have to, you don't have the activity pub problem where you need to go figure out which federated server that you're gonna jump into immediately. Um, so in that context, um, I. I thought that was really, really appealing. We weren't a hundred percent committed to it. Um, and then blue sky really blew up after the election last year, um, and ballooned into, I think they're gonna hit 35 million users today, or something like that.

Um, and so at that point it became very clear to us. It's like, oh, okay. Live video for the AppD protocol. It's like a, like an excellent niche for us to occupy.

Andrew: And it feels like App Protocol has just like these different properties that make making a primitive like that a little bit different. Uh, on more like in the centralized web, only Twitch can do Twitch streaming well, and it's in their best interest to only do it on their website. But since like, uh, I'd say the ethos of app proto is like openness and sharing.

That doesn't really apply. So like you, you have streams. I can integrate your streams onto other platforms or even my own platform. So like, how are you thinking about Stream Place as like a primitive in the ecosystem rather than like a, just a one-off Twitch replacement.

Eli: Yeah, it's, I think that's one of the things we're most excited about is, uh, you know, the. The tagline of the company is, or the, the mission statement of the company is solving video for everybody forever. Um, which is what really appeals to me about the space. They're not looking to just build one more, uh, centralized platform that gets acquired and forgotten about over the course of years.

I really want to be building things that could endure for decades or hundreds of years, um, when people are looking back. So, um. What we've really invented at Stream Place is a novel form of decentralized broadcast. So, um, a live stream can be modeled as I take your stream. I slice it up into one second, MP four files, and I throw those files into your face one per second, right?

That's really most of the time, that's what's happening behind the scenes when you're watching a live stream as you're just getting video file after video file, uh, o over and over and over again. So what we've done is take those one second video segments. We sign them with a, with a user's key. Um, it's a, it's what's called a C two PA signature.

I can get into that. It's from the Coalition for Content Provenance and Authenticity, which was a project from Adobe and the BBC to sort of get us ready for a world full of deep fakes. When anybody can deep fake anything, how are you ever gonna trust the video you're looking at? Um. Uh, part of what they invented over there was sort of this novel signing scheme where there's a cryptographically secure, um, signature, but it's embedded within the MP four file.

So if you run stream place on your computer and look under the hood, actually look at the files, all you're looking at is a series of one second MP four files with these sort of embedded signatures. So super easy to work with, uh, while also being cryptographically secure and having all the sort of properties that you'd need for decentralized broadcast.

So the thinking then is you can post these MP four files to, uh, a stream place node that can propagate all over the internet, and you can download those segments from, uh, you know, sketchiest website in the world, ru and still, uh, be confident that you're looking at the right thing through the magic of cryptographic signing.

Where the at protocol comes in right now is that that signing key gets, uh, posted to your PDS just the same way you make a blue sky post. And so that, uh, that signing key gets associated with your at Protocol identity. Um, and we can sort of draw a line then between the stream LY stream and your, um, you know, your Blue Sky account or your app protocol account, right.

I would like to deepen that integration a little bit. That is, you know, cryptographically secure, but still a relatively shallow integration with the app protocol in that it's just this one key that gets posted there. Um, as we move into doing things like archiving live streams. And, uh, clipping. I want to deepen the integration there.

Um, what I can to do is just take every one second video segment and throw them into your, uh, into your Blue Sky account, into your, at proto PDs, right? Because I would immediately run a foul of of rate limits that's, that's gonna, that's gonna get us banned from all of the hosted PDSs very, very quickly.

So, um, we're gonna have to be. A little bit cleverer than that and come up with, I think, some novel mechanisms for, um, syncing those two things while, um, uh, while not overwhelming the PDS basically, but um, uh, yeah, that's how you sort of draw a line between all the different pieces of the architecture there.

Andrew: Yeah, it would be cool if the PDS did offer that though, since like one of the like table stakes opportunities, blue Sky's going at is like, you own your content, it comes with you, but if there's a big asterisk and it's like, but if you posted a video, it's not actually hosted in your account and it's hosted in a third party and that might go away at some point and you'll not have it.

So it's like, that seems like a, an odd line to be there.

Eli: It's a really interesting thing from a product perspective too, because as a live streamer, you're not necessarily expecting that the moment you live stream something that is archived forever and, and anyone can seek back through it and that sort of thing, right? At a minimum, you might want, um, like a minute delay.

On that, on your live stream being archived in case something embarrassing happens. Or you might want to wait, you might want to do a full four hour live stream and then afterward decide whether or not that gets archived to your account. That sort of thing. These are all the sort of user assumptions that people coming from other streaming platforms will have, um, that we need to accommodate within the framework of, of credible decentralization and, and still figure out how of all of those pieces work.

So, um. Yeah. Yeah. Not, not a, we're not finalized on any particular design there, but from both a technical and user perspective, there's, there's, um, some interesting things to accommodate.

Justin: So are you still using, I mean, are you using live. Here, like a, a version of lip here for transcoding. Okay, cool.

Eli: Yep. Uh, so within the, the stream place node, we have an integration with the live peer gateway so that all of the, um, if you do a 4K live stream through stream place, that will send it to the live peer network processed into 10 80 PI think it's 10 80 P, seven 20 P, 360 p, and one 60 p right now. Um. And, uh, yeah, it all works together really well.

You can watch a, a lower quality rendition of, of everything. Um, one thing we haven't had a chance to do yet, but I, I really want to sort of upgrade the live peer network to use this C two PA signing scheme that I described earlier. Where, um, you know, one of the, one of the recurring things for, for live peer is how do you, how do you trust this, right?

If you sent it just to somebody, how do you, how do you trust that the output is good? And there are some technical responses to that. Um, in that the, you can do sort of a, a partial decode of the output and do some clever mathematics to make sure that the output looks kind of like the input, um, in a, in a way where you could trust it.

But there's also sort of a. A social, uh, element to it where if you can look at that and see, oh, somebody took in my video segment and just sent me back empty black frames that were very cheap for them to produce. Um, if there's an embedded signature in there, you could look and say, okay, well I'm never gonna use this orchestrator ever again, kind of thing.

Um, so yeah, one way I hope we can push the life peer network forward as well.

[00:18:53] Ad 2

Justin: This episode is sponsored by MailTrap, an email platform that developers love. Go for high deliverability, industry best analytics, and live 24 7 support. You can get 20% off for all plans with a promo code dev tools. Check out the details in the description below.

Um, shift the conversation a little bit.

[00:19:16] Monetization Challenges in Decentralized Platforms

Justin: Uh, so centralized platforms are. Just generally easier to monetize, um, know, for, for better or worse, like they're centralized. 'cause that makes a good business. You can have a lot of control over, not just like technical decisions, but like, over the presentation and like where it shows up and how content is shared and like all these things.

And, know, one of the key things in, in what you're saying here, like kind of what you wanna build is like. More agency for people to like, maybe wanna embed your stream on your site and you know, all these other things. But you're integrating with the live peer network for video, trans coding video is, even though you're like doing something that's technically a lot cheaper than the status quo, it's still, you know, not free. It's still like, uh, uh, potentially expensive thing. So as you're going through this journey, how are you thinking about, uh, revenue managing costs? Like what? The shape of the business looks like while still achieving sort of your like in like social and technical goals.

Eli: Yeah. Yeah, it, it's a great question. Um, and it very much an open question for the entirety of the app protocol ecosystem where, uh, there's been a couple. You know, blue Sky's well funded and we saw Grays and Skylight both receive some sort of seed funding. Um, and then us, of course from the Treasury. But, uh, to my knowledge, that's the end of the list.

Not a lot of people are really, are really figuring out how to, how to get paid doing this yet. And of course that's particularly important to us building live video infrastructure, which is. Arguably the most expensive thing you can, you can possibly do. Um, Twitch got to where it is by, uh, being, you know, Amazon's the company that's willing to just shovel money into the money pit until they've bled out all of their competitors.

And that's where you and that really happened, including like Microsoft getting out of the game. They had Mixer, which was supposed to be their option. Facebook had Facebook gaming. All of these things, uh, have sort of. Fallen by the wayside because Amazon just was willing to, to go at it for longer. So, uh, we're very interested in tackling that problem head on, uh, so as to not just run into an immediate financial wall.

Um, two answers there. Uh, one is, uh. I was, uh, live peer, recently sponsored the first Atmosphere Conference, which was the first at Protocol Conference in Seattle. And, uh, I, micropayments was probably the word, most common word that I, that I heard there. Uh, everybody agrees on, uh, the, AT protocol needing some form of, of micropayments for supporting creators and for supporting platforms as a streaming platform.

That's of course, like especially core to us, right? Because that's what people, you know, subscriptions and tips and. And bits and all of these sorts of things are, are very core primitives that our, that our users expect. Um, I, I would love for Stream Place not to be the ones that invent that, uh, alone. Uh, it's a, it's something that a lot of us in the ecosystem need.

I would love for that to be a collaborative effort that. Um, we all sort of come together and, and, uh, figure out what that's going to look like. Um, I will say, given my background, um, the conversation I had a lot at the conference was, uh, someone would say to me, hat protocol needs micropayments and here's my idea for it.

And they would say a bunch of words. Then I would say, that sounds like a cryptocurrency. And they would say, oh, no, no, no, no, no. This is, here's, here's the reasons why. It's not. So, uh, there is a, there's a tension there in using these sort of cryptographic primitives to represent payments, uh, but not going full blockchain, which is sort of unpopular in that community.

So I don't, I don't have the resolution there, but I'm sort of, uh, uh, watching all of that with interest.

Andrew: Yeah, it's de it's definitely interesting how like at Proto itself is a decentralized thing and just by that alone makes it almost like a cousin to cryptocurrency and like a lot of things that get me excited about at proto like. Owning my own account. That's very much like having a wallet where people can put things.

They just happen not to be coins right now. So like this, this friction point in the community where it's like crypto bad, 'cause we all think crypto bad, but like crypto has some, some benefits that could be really useful in a decentralized social network. So like. I, I know you said you didn't have an answer, but like, how, how do you think it should work?

Eli: Well, I think it should work as a, as a funny way to, to, to think about that. You said it's like having a wallet. I would go one step further. It is having a wallet like your, your PDS has a signing key associated with it, which is an SECP 2 5 6 K one key. Which is also known as an Ethereum key or a Bitcoin key.

Right? That's, that's the thing actually securing all of your, your data in there. We don't think of it like that. Um, but, uh, it's, it's, uh, there's, there's a, you, you, you can make the case that we're about 95% of, of the way there at protocols, not a. Blockchain is the thing that the, the creators like to point out.

That is, uh, it doesn't have the sort of necessary mathematical structures to make it a blockchain. Whether that is a sufficient condition to, to be or not be a cryptocurrency is, uh, a little more, a little more up in the air. Um. How do I think it should work? Uh, my prediction is that the, the first project that successfully integrates those two things, uh, sort of cryptocurrency payments with the yet protocol will be both extremely.

Unpopular and very, very successful. Um, I don't necessarily think I want to be the one to do it, but I do think that it is probably inevitable. Um, and, uh, uh, we should, uh, figure out when it, rather than trying to just complain about it, we should figure out, okay, when this happens, what, what's our reaction to it?

And, and, and, uh, and how would we like it to look? That sort of thing. Um. I, I, I don't know. I don't, I don't really have an, an advocacy for that going forward. Um, I've sort of got my own sort of the video thing going on. I don't necessarily take it there. I I, I will observe that, like, I think the Blue Sky team really gets it in that, um, the, the term they've been using for what people don't like is hyper financialization, right?

What they don't want is, um, people, what people think of as, as blockchain stuff, mathematical structures notwithstanding, right? They don't want. You know, NFTs embedded in posts and people pushing their, their tokens on the, on the network and, and that sort of thing. And um, and I think that's, that's perfectly understandable and, and something to go for.

But, um, it's a very, uh, very interesting and fine line to walk there, I think.

Andrew: Yeah, definitely. Because if we follow the definitions, we just set up a post is just an NFT, like that's all it is.

Eli: right, right, right. And it doesn't, uh, I, I like to point out that the, the required fields of an NFT, according to the EIP 7 7 21 standard are, uh, uh. Title and image. Uh, so if you've, uh, ever put those two things together and, and signed them in a decentralized way, you've, you've, uh, you've officially done a cryptocurrency and, and are condemned, right?

So, uh, yeah. Have to, have to figure out that part.

Justin: Yeah, I mean I think this is like an interesting thing 'cause it's like one of the areas where may actually make a lot of sense for it to be a cryptocurrency and just like, know, you know, we, we have to sort of acknowledge the. Cultural baggage that comes with this. And there, there's like a, there's like a big knee jerk reaction against it, and I think it's not completely unwarranted, right?

Because there is a lot of like scams and people just like, you know, pushing out cryptocurrencies that are just like themselves, only the coin or whatever. That is the only like value of the thing that they're providing. Whereas like this is like. Um, just a mechanism to make transacting easier. It's like you would have to have that bridge between, you know, what is the real world finances and how do things get on this network.

But also there's like, you know, if there's a conversion, uh, there's like economics to this where it can go up in value and down in value and like how do you, uh, you know. Manage this reliably and safely and like give people confidence in the system that's non-technical, you know? And so there are like all those like really gnarly human problems that would still have to be figured out, I think.

[00:28:01] Navigating the Crypto and Video Worlds

Eli: Yeah, it's, it's been really interesting for me where. You know, everybody at Live Pier either had a crypto background or a video background, and I was, I was sort of a video guy. I had experience with scuttlebutt, but not really with, with cryptocurrencies going into it. So it's been a really interesting transition to me going from, I, I was sort of went into crypto with a, with a help to healthy skepticism and, and never, uh, got, got super into the, the DGen culture or anything like that on that side of things.

And then I turn around and I'm in the room with the app protocol, and I'm the only one there that's been funded by a cryptocurrency, treasury. And all of a sudden I go, oh wait, I'm, I'm the crypto advocate in this room Now that's really interesting. But I've sort of found myself a, a centrist between those communities, um, in some ways.

Um, I, I don't think, uh, it's a problem. Uh, I, I, I, I, I guess I reiterate that I, I don't think it's. Yeah, I, I think we need to get ready for when it happens rather than, uh, uh, trying and maybe, maybe, I guess the accelerationist argument would be to try to make it happen. So it happens under, under ways we control, not something I'm super interested in doing, kind of got this video thing going on.

But I do think, uh, uh, it, uh. It's probably too close to the underlying, and we're literally using Ethereum libraries to parse at protocol primitives and do signing and that sort of thing, right? Like there's a, it's a distinction without a difference under a certain level of, of, uh, you know, once you peel back the technical layers a little bit.

So, um, uh, yeah. Interested to see how that plays out. For sure.

[00:29:46] Stream Place as a Dev tool

Andrew: Yeah, so getting back to video things, we're on a Developer tools podcast and uh, I think it would be nice to talk about how like developers can use Stream Place as a tool. I was also at the at Atmosphere conference and a conversation that really interested in me and I actually like. Thought of the idea before.

I heard you guys talk about it, but you teaming up with Skylight Social to be like the live video aspect of their TikTok clone. So like can you go into that a little bit and maybe once you're done with that, how that might translate to other people building other platforms and even building their own websites?

Eli: Yeah, absolutely. Um, first off, uh, shout out to Skylight second biggest app in the, in the app protocol ecosystem. Now. It's 150,000, 200,000 users. Something, something like that. Now, um, Tory and Reed, also Seattle locals like me, so that's been, that's been awesome, uh, uh, working with them and, and, uh, working to get Stream place, uh, into there.

But, um, yeah, like you, like you got at, so, so, uh, we're working now to embed stream place streams within the Skylight social app. But that is as a prototype of, uh, just shipping the Stream Place primitives, um, as JavaScript libraries that can then get embedded within React and React native apps. Um, so that's on the, on the client side.

Um, we're looking to make that very easily embeddable and make that just an obvious choice for any at Protocol Ecosystem app that wants to have live video. Let's make it the obvious choice, right? Stream place. Obviously here's MPM, install, stream place and get started, that kind of thing. Um, on the infrastructural side.

Um, so something I've been very, very committed to since the beginning of Stream Place, uh, from a developer tooling perspective is that it is a single file binary. Uh, it's not microservices, it's not a Docker container, though we do ship a Docker distribution. It is one, uh, one single file node. If you wanna make changes to it, you can clone our repo, change some code, run, make, and you have that one single file, statically linked node with no dependencies.

That works on Windows, Mac, and Linux. And that's something I got wrong. Like I I, I, I'm tempted to say we got wrong, but that's something I got wrong specifically early on in live Peer studios development is not for any particular reason, but. This tech industry sort of runs on microservices, right? You, um, and I get it, if you're Microsoft and you have to coordinate a thousand teams globally, you're gonna want yourself some microservices.

But coming from an open source, um, and, and decentralized, primitive sort of perspective. Um, you don't want, we, we had Live Peer Life, peer studio's. Whole stack was open source, but nobody other than us could run it. 'cause it was like six different services that all got deployed to a Kubernetes cluster and communicated with each other and, and had different dependencies and all of that.

And I tried to bring it together and ship it out in a project called Life Here in a Box. And that just never, never went anywhere. Um, and so this is sort of my, uh. Not pro monolith approach necessarily, but this has been sort of my advocacy, um, is that, uh, for decentralized networks like this, you really need to have the easiest possible development experience.

So it's a single file. You can, you know, you can be just a, just a sort of power user, very sort of enthusiast about this stuff. If you just know how to make a VPS upload the file, run the node. Um, and you can, you can get started really, really easily. So, um, those, those are a sort of two approaches. Um, but, but still get the benefit.

This is something I really learned from working with Blue Sky and the app protocol ecosystem. The coolest thing about that is it's all production data, right? As soon as you start developing your app, you connect to the fire hose. And here's like five megabits of blue sky posts and likes and follows and everything that are flowing right through you.

And you can really sort of build in production. It's not this sort of sandboxed uh, you know, fake data kind of thing where you're only building things in your, in your own little toolkit there. So that's the other piece is you run your own stream place node, um, configure it to replicate streams with other stream place nodes and.

You can, you can work directly with the data there, right? Something I haven't, I haven't done this a lot, but, um, uh, something I do on my own. So I do development streams for Stream Place almost every day. And I love to, now that we're a little. Further along, I can download my actual segments of my live stream into my dev node and change how they're processed and, and, and push them back out.

Right. Very sort of, very sort of meta approach to, to doing development. So, um, that's, uh, yeah, that's, that's another piece of it is, is the building and production and, and making it as easy as possible to people to get apps that are, that are, uh, uh, working with the, the actual data.

Justin: One interesting question that I always like to ask people when they're building on app proto is like, so app proto has this notion of a lexicon. It's like you can sort of define your own schema for your own data type, and then Blue Sky has their. Lexicon. So if you use theirs, it shows up as content on Blue Sky and if you use your own, just like you kind of gotta figure out to display it or whatever. And I'm curious like what is your balance for that? It's like how much do you think about using like the regular app proto Lexicon, like just to have things show up like on the Blue Sky site versus like, eh, we wanna have like custom data for our platforms or whatever, and we want to use our own lexicon for that. What does that

Eli: Yeah, great question. Really active discussion in the community right now about that sort of thing. Um, I can give a really concrete example of how that, uh, has, has sort of played out for us, the very first version. Uh, so if you're a live streaming platform, you need chat. Uh, that's, uh. We actually got pretty much zero users until we had chat.

'cause that was sort of a, a, a necessary condition to having the audience interaction that, that makes people actually wanna do this sort of thing. I. So the first version of chat we shipped was, uh, blue sky replies. So basically when you go live, and this part's still true, you when you go live, it makes a blue sky post and it's got your thumbnail there and, and, uh, and your, your sort of stream title.

Um, and a, and a link so people can click and watch your stream, play stream. And so the first version of chat that we shipped, when you typed a a chat reply in there, um, it, it posted a reply to that post on Blue Sky. Um. Which was, uh, uh, interesting and funny. Uh, it would lead to, so afterward you'd have the, after a long live stream, you've had this sort of blue sky post that's got 700 replies, but none of them are threaded.

So it's all just sort of like throwing random messages out here. And, and, uh, people are like, people are like, what, what is this? How, how, how? I don't, I don't understand what's happening here. So, uh, yeah, I got got some feedback on that. Um. Ended up moving it to our own lexicon. So instead of being a. Uh, app B, sky Feed Post, they became place stream chat messages.

Um, and so now we don't have that problem anymore, but it really hurts visibility, right? One of the things that was nice about the silly old system is that people saw all these weird, random, non threaded replies and they're like, what's going on here? What, what, what is this? And then they would go, click on the stream and go find out what it is.

So it was actually a little unpopular among, uh, some folks for us to make that change. Um, and so, uh, this has been, yeah, one, one area where, um, it would love, it would be great if there's a middle ground, right? Right now it feels very all or nothing where either we're making blue sky posts, which works, but is sort of this semantic mismatch or you're making.

Stream place chat messages, which are, uh, you know, we can customize to our needs, but they don't right now show up anywhere else until other people build integrations. So I would love if there was sort of a, a middle ground in there. Um, one thing I had a conversation with a couple folks on the Blue Sky team about would be, um, you know, blue Sky's probably, but you know, obviously, uh, when you're, when you're doing what I do, people come up to you a lot and they say, oh, I understand at, at critical live streams.

So are they gonna be in the Blue Sky app? And I say maybe like I not really up to me. Right. Um, I would love for that to be true someday. Um, obviously they have moderation concerns 'cause we exist sort of, um, unlike something like skylight videos, which are also blue sky posts and are sort of under the protective ages of their moderation infrastructure.

Um, we operate outside of that and have to have to do our own. So very legitimately, they've got concerns about. Um, making sure that stuff gets moderated well and that sort of thing. But something I had a conversation about, which would be an interesting middle ground for us, would be something like a presence, API, um, which, which is to say, um, or, or it's, I think Discord calls these activities.

Um, but basically it's like, I'm live right now and that means there's a, there's a red circle around my avatar and you can click that and see where, where I'm live, and then go to a different app or a different website to actually watch the live stream. So something like that would be. I think a really interesting middle ground where they don't have to be responsible for hosting the content, but they still sort of throw a bone to the other platforms and, and post them in that, in that same direction.

You know, you could obviously do something like a story in the same way. Um, I would imagine, which is another interesting category of thing where, uh, a story on almost every platform is a temporary piece of content, right? Where it's like a, like an expiring video. That will exist for eight hours and then go away, that kind of thing.

So, um, some interesting, interesting places there. I, I, I, I tend to think, um, I. I, I don't like the idea of ever really having different social graphs on the app protocol. I understand that the argument is that your, your LinkedIn friends and your Instagram friends might be a different group of people, but I, I don't really, I.

I think one of the biggest selling points of, of the ad protocol is that you have this sort of portable social graph that that works across a lot of different apps, and you don't have to refind your friends when you move from blue sky to skylight to stream place to, to what have you. So, um, uh, yeah, I, I, uh.

There's different schools of thought on this, and I, I guess I would tend toward, uh, like, like a, like interconnection. Maximalism would be the, would be the term I guess, just as much as possible. I would love if we were all using the same lexicons and had interoperable data and lots of apps. Feature data from lots of different lexicons that were designed by lots of different teams.

Um, and, uh, yeah, hope, hope we can push that direction.

Andrew: Yeah, I'm, I'm definitely one of the interconnected maximalist too. It's like we, we currently live in a world of Wal. Gardens and like trying to imagine the world not like that is is hard. We've never really seen that. Like you could say maybe Web 1.0, but Web 1.0 is like so bare bones that it barely even resembles the modern world.

So like, kind of like the. The walled garden aesthetic, but with interconnected data, it just seems like it could be so fun. Like Blue Sky itself. It is trying to be a Twitter replacement, and Twitter is trying to be the everything app. I think Blue Sky, I. Oddly enough has a better opportunity to be in Everything app.

Since all of the data is already open, there's nothing to say that they won't, uh, start displaying your streams or, uh, posts from white wind or, uh, images from flashes all in one app. It, I'm very interested just to see where all the innovation goes in that space.

Eli: Absolutely. And, and the, the reality of it is if you're, if you're decentralizing successfully, you don't want, um, you, you want lots of apps that work with the same set of data and you want there to be lots of apps, right? So part of our pitch. To big Twitch streamers when we're sort of ready to start going after them is don't switch from Twitch to Stream Place.

Switch from Twitch to your own app that has its own presence in the app store and play store, and you have a direct relationship with your fans. But because it's powered by Stream Place infrastructure, you still have sort of all the creature comforts you would expect from a rich live streaming platform, right?

You have chat, you can host other streamers. You can even decide to sort of promote. You've got your own sort of squad and they all show up within your app, and people go to your app to watch their streams, but then they launch their own app and it, it jumps over to there. Um, I. I think that's a a, i, I don't know.

I think that's super promising. Big Twitch streamers are sort of bus medium sized businesses under themselves, right? With the, with the revenue they bring in and the, the sort of audiences they have. The hard part, uh, and this was also a huge topic at Atmosphere Conf, is how do we explain any of this stuff to average users?

Right? So like, it's, it's so hard because it's, you start with, okay. It's, uh, blue Sky is is Twitter, but it's j instead of Elon and, and we're all happy. Okay. Well, actually it's built on these things called PDSs Personal Data Servers, and. That's where all of your data lives. Uh, and now there's other apps in there, uh, and it's so impossible to like, because people think they get it.

'cause they've seen sign in with Google and sign in with Facebook before, so they're like, oh, they go to Stream Place and it's like, oh, it's sign in with Blue Sky. I understand what that means. And it's like, no, actually there's something more interesting and awesome happening here, which is that it's not just, you're using it as an OAuth mechanism.

And I've already used the word OAuth, so I've lost everybody. But it's not just that, but it's that you are. Interacting. It's the, it's this different app interacting with your same set of data and, uh, and, and can build lots of interesting things in that way. And, uh, there's nothing like that really. Uh, Web3 people get it because like all your data's on chain and different apps interact with that.

That's fine. But your, your average user. There's nothing to compare it to really. Um, and so I, I don't, I think it's a huge open problem is how to, how to educate people and, and sort of sell them on how that works.

Andrew: There are lots of thorns there. And even with just login, which is kind of just like a little mind-boggling, like even as a developer understanding how login is different and what that means for you is, is a hard thing to get around 'cause like. I, I, I work at Gray's and we've had our own OAuth issues and uh, I was like, why, why are we having issues with OAuth?

It's just like a thing that everything does, but it's because like in a normal situation, you have a centralized server. You go OAuth with them In the PDS world. There is no centralized server. It's wherever your PDS is. So you're logging into your PDS and there's this whole weird flow of like, okay, I'm logging into Graze.

Wait, nope, I'm actually logging into this PDS over here, and it kicks me back to Graze. So even explaining that difference succinctly to a non-technical user is a challenge. Literally everybody in the ecosystem is facing right now.

Eli: A hundred percent. And, and even among folks that want to, you know, help to, you know, inarguably things are pretty centralized around blue Sky right now among the, um. The, the PDS fleet, the relay and the app view. Um, and even among advocates, you hear them say, oh, we really need to stand up our own app views and stuff like that.

But that's, that's nothing in the abstract, right? Unless you're going to fork Blue Sky and make your own app that then connects to your own app view and make sort of different editorial decisions than the Blue Sky app you makes. Yeah. That that's something, but just running your own. App View does, does nothing by itself, right?

Stream Place is an app view, but it's not a Blue Sky app view. You can't point the Blue Sky app at Stream Place and get meaningful data back from it and that sort of thing. So, um, yeah. Yeah, I mean, just like even among the community, uh, getting them to, to understand everything and then even there's a step lower than that, that like even I am learning new things about the protocol all the time.

I'm pretty much as in it as anybody could be at this point. But it's still learning like. An interesting distinction about the at protocol from, um, stuff like Web3 and, and scuttlebutt where I used to, uh, work around is in those and, and, and no Stir actually, um, is in those systems. You sign a message, uh, and that sort of then exists in the abstract, and I could show you this message and prove to you that it's signed with a certain Ethereum key or scuttlebutt key or what have you in at Proto you sign.

The root of the merkel tree of your PDS repo. So you only ever sign something that rather than signing one piece of data, you essentially sign all of your data all of the time. Um, and the, the distinction there is really important because it means you can delete stuff, right? You, you delete one of your blue sky posts and then your new signature has this, and then that blue sky post doesn't really exist in the abstract as something that could get passed around or, or indexed anymore.

So I think that's like a really. Profound, um, architectural decision that even I didn't know that until like a couple months ago when I was, I was poking around at, at different sinking, um, heuristics and, and that sort of thing, right? So, uh, yeah, there's just a, there's a lot there.

[00:47:26] The Future of Social Media and Decentralization

Justin: So one of the things that we like to, uh, ask all of our guests, uh, is usually like a forward facing question. And, and I feel like you are working kind of on the leading edge of a lot of these technologies. I mean, uh, I, I feel like decentralized movement in general has always been ahead of the curve and, and like technical thinking and figuring out like how do we make things more safe and more sustainable and et cetera.

So some ways, even kind of living in the future, but like. you are working on, uh, stream Place now and thinking about like what's next, what do you think the, like social media, social platforms, digital world looks like in the next few years? How is it gonna change? You know, what are you, what do you think are gonna be the big developments? Um, yeah.

Eli: Yeah, absolutely. Um, I, I, you know. I really appreciate Elon's buying and trashing of, of Twitter for, I think sort of giving the best possible example of social media and social networking is something that is too important to trust to any centralized company. Uh, I, you know, credit Blue Sky with coming up with the term billionaire proof for describing their own.

Um, infrastructure, and I think that's really where things are going to be heading forward. Um, I can tell you from a, um, it's been interesting here. Here's, here's one way to frame that question that's interesting for developers, I think, is that, uh, when I sort of came up, you know, I went to. I, I graduated college in 2013.

That was sort of coming out of the, the era of Google as this sort of all, everybody's like, oh, their, their campus is so cool and it's this like amazing place to work on, on all this sort of technology that sort of led to this. What I think of in my head is sort of the golden era of the startup. Early Y Combinator stuff where it was, oh, this is, this is really, these are teams changing the world and they've got these cool cultures and all that.

That's, that's where I really wanna work. That's awesome. Um, and then fast forward a little bit further, and you see a lot of those startups getting bought out by huge conglomerates. You see a lot of them like Vine. Doesn't exist anymore. Got bought out by Twitter and then just like thrown away, that sort of thing.

And so I think people have grown a little wiser about that sort of thing. Um, and so when we're pitching Stream Place to people, obviously it's sort of this, this funny, grandiose tagline solving video for everybody forever. But I do really think that if you're a young developer coming up right now, you're sick of working on, you don't wanna just work on something that's gonna get acquired and thrown away in five years, right?

That's not. Interesting to you. If you're coming out of this and you're, you're idealistic and ambitious, you wanna work on problems that will continue to be relevant forever and, uh, and, and solutions that I, I really do hope that a hundred years from now people can go look at my stream, place streams and use the same parts of the signing format and validate that this was in fact one of my streams a hundred years from now.

That sort of thing. So I'm really, um. Yeah, it starts with like the developer will where people just wanna push things in that direction. Um, and then coupled with this is like the Web3 background, but you know, it doesn't need to mean working in open source doesn't need to mean you're a single developer working by yourself and, and not getting paid.

Right. There's a lot of ways just sort of start to. And Blue Sky's obviously got an enormously successful open source app. Um, and, and there's a lot of ways to start to figure out how to fund that sort of thing. And, and fund, you know, blue Sky and Skylight are both, um, formed as public benefit corporations.

Like, like Kickstarter is right, where it's not just operating for profit forever, it's, it's this sort of thing. So I think there's gonna be a lot of, uh. I, I guess I'll just put it as like, I wouldn't want to be just like a ruthless, cutthroat for profit startup trying to hire the best and brightest developers.

'cause I don't think that's where the interest is going to be. And, and I think the, the, uh, really talented folks are gonna wanna work in the open, that kind of thing as to where we head from here. Um. What the, what is the, what is the end result of that, of that world look like? Um, I think we're seeing sort of the, the beginnings of it right now.

Um, with the, AT protocol. In some ways I hope it will be better, but very boring. There's gonna be people that they download their one. Obviously it's the toughest people to convert are the people that downloaded Facebook once in their life, and that's the internet and social media to them. And that's gonna be the hardest to move people past.

And there's always gonna be folks like that. But, um, hopefully they can be, um, hopefully that doesn't need to be monopolistic and limiting anymore. Right. Hopefully that means it's, they're using this one app, but lots of other apps can, can interact with them. Lots of other apps can, uh. Continue to use the same primitives and, and they can get access to sort of the full richness of, of the, uh, of the entire ecosystem rather than, rather than using just one app.

So yeah, that's the direction I'm trying to push things in at least.

[00:52:42] Conclusion and Future Prospects

Andrew: Well, it's an exciting future and you are a part of that exciting future. So thanks for coming on and talking about app proto live peer stream place, and all the other interesting video technologies you've worked on. So thanks again for coming on.

Eli: Yep. Thanks for having me. And, and of course, shout out to, uh, uh, I'm live streaming development of Stream Place every single day on Stream Place. We've got the app live in the iOS, uh, in the App Store. In the Play Store. So go ahead and. Go ahead and grab that. And if this is something you're interested in contributing to, we've, uh, just got everything moved over to, to GitHub and our, uh, just got sort of our first round of external contributions coming in.

So, uh, really hoping to pick up the, the velocity pick, uh, you know, polish off all the, all the little bugs that I don't have time to take on and that sort of thing. So, um, yeah, excited for the future.

Justin: Great. Thanks Eli. I'm really excited to see what you'll build.

Eli: Thanks very much.

Discussion in the ATmosphere

Loading comments...