If you would like to view the full recording of the live State of Livepeer Q3 call, you can find it on our YouTube channel. You can also read the full Quarterly Report by Mihai Grigore and Stephanie Dunbar here.

Participants:

  • Mihai Grigore – Messari Sr. Research Analyst
  • Stephanie Dunbar – Messari Research Analyst
  • Doug Petkanics – Livepeer CEO & Co-founder

Stephanie Dunbar (00:27): Hi everyone, welcome to the first quarterly call for the Messari quarterly reports in Q3. We’re getting started here today with Livepeer. I’m joined by my co-author Mihai Grigore as well as Doug Petkanics from Livepeer. Doug, you want to get us started by introducing yourself and the Livepeer network for us.

Doug Petkanics (00:47): Sure, thanks Stephanie for hosting. Thanks to the Messari team for writing the report and hosting these calls. I’m Doug Petkanics founder of Livepeer where we’re building the video and streaming layer for web3. A decentralized network of node operators that power video streaming and video transcoding to enable developers to add video to their social apps and creator apps. Excited to dig into what we’re doing and talk about Livepeer today.

Stephanie Dunbar (01:17): Great thanks so much for that! I’m also going to pass it over to Mihai if you wouldn’t mind giving us a little intro to yourself and getting us started on the questions.

Mihai Grigore (01:26): Thank you so much Stephanie. My name is Mihai. I’m Senior Analyst with Messari where I cover decentralized data indexing, decentralized live streaming and Livepeer is one of those protocols that we cover on a quarterly basis. And here in front of us we have a sample of the Q3 Livepeer report. So in a nutshell what we do is we are reporting on the key metrics of the protocol over the quarter and for Livepeer specifically we have put together a table with the core metrics with respect to the network usage as well as with respect to the staking activity and to what extent the protocol and its participants have been able to generate revenue over over the quarter. So Doug we have a first question for you here and this relates to the Livepeer usage, so here in this chart we see that this quarter we have an increase in terms of Livepeer usage by 12% would you like to give us an introduction of what this increase means and what minutes transcoded actually signify in terms of Livepeer usage?

Doug Petkanics (02:57): Yeah absolutely. Kind of as I alluded to a little bit of background on Livepeer and who’s using it. It’s developers that are building video applications apps that look like Twitch, apps that look like YouTube. Powering the video streaming and the video encoding to deliver that video to all different users on different device types is a really hard challenge to operate yourself, it’s expensive to scale if you’re using services and the Livepeer network is an open infrastructure network that can power all of that video stuff for those applications so that the developers can focus on their users and their use case and their social features. So this chart that you show here tracks Livepeer’s networks usage in terms of how many unique minutes of video were transcoded by the Livepeer Network quarter over quarter. It’s exciting to see the growth and the growth last quarter. I guess where that’s coming from, it’s coming from developers building these social applications and these creator-centric applications typically connected to the web3 space in some way so think of a social application that lets people build their follower graph on-chain or write their video data to decentralized storage they’re building on blockchain technology and they’re using Livepeer. There’s typically two different ways that a developer might use the network, the easiest way and the majority the way the majority of users use it is by using a hosted service, so a service that runs all of the Livepeer nodes and funds them with cryptocurrency and interacts with Livepeer network for them. The most popular hosted service, in fact like the main one that’s driving usage is livepeer.studio which is built by a company Livepeer Inc. that I’m one of the founders of. The second way that people use it is by running a node and you using this open network directly which is you know a little more technically challenging and complex but really powerful because it delivers that promise of infrastructure portability and the ability to exit from any hosted service. That’s really kind of aligned with the space and you see certain applications experimenting with and beginning to run on the Livepeer node, the Livepeer Media Server which is called Catalyst and I think we might talk a little bit about that during the call.

Stephanie Dunbar (05:24): Thanks Doug for the overview and you know in turn with increased usage this quarter we also saw you know increased fee revenue, I believe up 7% in USD terms but that number was quite a bit higher in ETH terms and obviously Livepeer fees are priced in ETH. Would love to know a little bit more about how fees work on Livepeer, Mihai and our colleague Mikey are planning on writing a piece on the Livepeer fee market this quarter because we see that it has a really interesting design so be curious to hear from you how that works, how does one pay for transcoding services on Livepeer and how that whole thing kind of adds up together.

Doug Petkanics (06:07): Yeah good question. If one of these developers runs a Livepeer node and wants to use the Livepeer network to encode video then they pay for that service in ETH right. But how much ETH does it cost to transcode an hour of video or a minute of video on the network, there’s actually no straightforward simple answer to that question because Livepeer is an open marketplace any node operator can charge whatever they want to charge in order to encode video and whoever is using the network and wants to encode video can offer whatever level of payment they want to offer in order to meet their needs and ensure that video gets encoded correctly. So this actually creates an open marketplace, it creates competition and ultimately leads to lower prices for users using this network. So the way the market works is that each node operator sets the price they’re willing to charge, users that want to stream video set the price that they’re willing to pay. They will discover the nearby node operators and test their connectivity and then anyone that is willing to accept the job at the price that the node operator is willing to receive will be considered in the set of node operators to perform that task and the fees will flow to them in the form of ETH. Probabilistic micropayments, which is kind of an interesting mechanism for scaling payments when you’re paying for say two seconds worth of video encoding at a time via blockchain network. This has been powerful and it’s worked nicely but I still think there’s so much potential because a lot of the usage comes through the livepeer.studio product that I was talking about before. The price that that product (livepeer.studio) sets forms a little bit of a coordination point for the node operators to know what they can charge if they’re willing to do the work and I think there’s lots of opportunity as more operators come on the network or as studios start to vary it’s the fees, it’s willing to pay in different regions of the world or for different types of redundancies and whatnot for this market to become more dynamic and more reactive.

Mihai Grigore (08:18): Thank you for this Doug. We’ve seen that revenue from fees is like one part of of the revenue accrued on the Livepeer network the other large chunk of this comes actually from staking and as you will see in this chart we do have a reduction in terms of the revenue from staking, however interestingly we do see in in this lower chart here that the staking participation actually increased this quarter so what are your thoughts with respect to the reduction of revenue from staking rewards versus the staking participation increasing in this quarter?

Doug Petkanics (09:09): Yeah good question. So you know you mentioned the kind of the minutes streamed on the network we’re up to 36 million unique minutes of video transcoded. The ETH fees were up you know 7% in dollar terms and even more in ETH terms but the staking rewards which are paid out to node operators and token holders in Livepeer token are down this quarter. What’s interesting there is they’re down this quarter in US dollar terms because as we’ve seen the whole broader kind of crypto macroeconomic markets decrease in value the Livepeer token value in USD terms has decreased in value and so that leads to the amount of rewards being paid out being worth less in US dollars. However it’s interesting the number of Livepeer token issued to these node operators and delegators in reward actually increased last quarter and what that means is those who are staking and participating are actually kind of earning a bigger collective share of the network than they had previously because those who don’t stake and those who don’t participate are diluted away and so it’s actually kind of a compelling opportunity if you think about it in terms of well the bigger share of stake I have the more opportunity I have to perform future work and earn more fees. That’s actually been a beneficial opportunity in the previous quarter. About to drill in and address your question about this 52% participation rate but before I do any questions or follow-ups on the kind of staking rewards portion?

Mihai Grigore (10:50): Absolutely yeah.

Doug Petkanics (10:54): Cool okay, so yeah what you had mentioned is Livepeer network passed an interesting milestone during this past quarter we for the first time in a number of months we surpassed 50% of the tokens on the network being staked. Why does that number matter? It matters because the economics of the protocol automatically adjusts the Livepeer token issuance rate, the inflation rate, depending on whether we’re above or below the 50% target for how many tokens are staked. When we’re over 50% of tokens staked the amount of Livepeer tokens issued each round decreases and when we’re below 50% staked it increases. And the idea there is the network targets about half of the tokens participating. We want a greater incentive and more inflationary rewards issued if we’re below that to encourage people to stake and if we’re above 50% we don’t have to pay out as much in terms of inflation in order to ensure that target, so we can reduce how many tokens are being printed and this is all automatic within the protocol. It’s not an arbitrary decision, it’s controlled by governance actually, what that target 50% rate is and so it’s interesting you noticed that participation rate is rising above 50% even though that means less tokens will be issued each round. You might be wondering why would that be the case? I think that the way that inflation, I think there’s two factors one is that the way that the inflation adjusts is relatively slow. Yes the inflation rate is dropping every single day a little bit because the network is over 50%, it’s 0.00005% per day which equates to about a 2% drop per year and because only half the tokens are staked it’s really like a 4% drop on your APY per year. So you know that’s significant over a year but over a couple of days or weeks or months it’s not incredibly significant you’re looking at earning a you know 22% APY on Livepeer right now versus you know 23%, a month, a month and a half ago or two months ago. So you know it’s a slow reaction which means people wouldn’t all of a sudden immediately unstake just because the rate dropped a little bit. I think the second reason is just a reflection of the broader macro economy where yields are available, what alternative uses of capital look like at the moment. I think that that you know that sort of APY that you get for doing work on the Livepeer network is actually probably more attractive than many other opportunities out there especially in light of the kind of the exposure of some of the weaknesses of the the DeFi system and all the kind of high-yield attention that was in that space. Here you actually have a chance to be increasing your ownership in a network that’s generating you know real fees for a real utility like video encoding and you know an enormous market of video on the internet, something that’s only growing.

Mihai Grigore (14:16): To zoom out a bit from this quarter what we’ve been doing is to actually have a look from the beginning of the year and to visualize how actually the inflation adjustment mechanism works and indeed as you were saying Doug we can see like four different periods within the year. In the first one we had staking participation lower than 50% and here we had an increase in inflation, a gradual one. Then in in the second one we had staking participation just over 50% but that was enough to trigger a decrease in inflation and then for the third one we had again staking participation below 50% which again triggered an increase in inflation and currently we are at the level of 52% staking participation and inflation decreases gradually. So for us it will be interesting to see over the next couple of quarters to what extent the lowering of the inflation will also gradually bring the staking participation back to 50%. We received one question from the larger community on this 50% equilibrium level and we had some engagement on Twitter about about this would you like to quickly elaborate about how actually Livepeers experiments led to this 50% target equilibrium level and whether that’s going to continue for the next next quarters?

Doug Petkanics (16:07): Yeah good question. So you know what’s unique about Livepeer is that instead of targeting a certain inflation rate it targets this participation rate of 50%. That’s the parameter, like why is that, how did the community arrive at that number, why does that number matter? It’s both an arbitrary decision for it to be 50% vs 55% vs 45% right. It’s an experiment it can be moved by governance, but it’s rooted in I would say some pretty pretty basic logic like, you want a high participation rate, you want to be able to reward participants, you want people using the token to stake and do work on the network but you also are realist and recognize there’s alternative uses for the token, like the token unstaked token and its role in kind of providing liquidity for people to be able to enter or exit the Livepeer ecosystem also has a valuable role. The ability for it to be available for creative incentives that aren’t addressed directly by the protocol such as getting tokens in the hands of builders and application developers and people contributing in ways that are not just video encoding work but let’s say on some of the business functions or marketing functions around growing the community and adoption of the project. So that room for liquidity, that room for creative uses, that room for people building on top of Livepeer to make alternative uses of the token, say in governance ecosystems, DeFi ecosystems DAOs etc, is important. So the logic of about half available for kind of security and the protocol half available for those alternative uses was the starting point. So far I think it’s worked out pretty well. I don’t think there’s any obvious takeaways that says it should be dramatically higher, it should be dramatically lower but it is you know one open governance proposal and vote away from being modified should the community wish to make a proposal have a debate or you know evolve the experiment.

Stephanie Dunbar (18:16): Thanks for all the granularity on that Doug. I think it’s good you know for the community to understand this mechanism, it’s you know a little bit kind of complicated to understand and also looking at the stability that it only goes you know up and down a little bit you know based on on the participation rate changing, I think is a really good thing to hear about when considering if one wants to stake or not. So appreciate all the details. Switching gears a little bit here I wanted to talk about some of the developments going on in the background with the core team who’ve been building out some great tools for the community to also use. You mentioned earlier, Livepeer Catalyst, which is basically an upgrade to the broadcaster node and an alternative to using livepeer.studio to connect with the network. Would love to hear a little bit more about the improvements, what Catalyst brings to the table and how the community is responding to this new upgrade.

Doug Petkanics (19:12): Yeah so this project called Livepeer Catalyst, super exciting. It’s very early but you could think of it as the Livepeer node that lets you use the Livepeer network from the perspective of the demand side, from the perspective of the developer that wants to act to build a video application. The reason it’s exciting is because the early Livepeer node was built on this open source piece of software we called Livepeer Media Servers, it’s very basic video media server that lets you do basic things with video. But it was you know super Livepeer aware so it knew the whole Livepeer protocol and you had it issue payments and discover nodes etc so basically the starting point building block. So Catalyst is really like a whole other level of powerful. It’s built upon both the Livepeer node software combined with a really powerful video media server called MistServer, an open source video media server that’s been in development for 12 years, that speaks every video protocol it’s excellent at being a building block for CDNs, it’s excellent at being a building block for scalable social applications and media applications and embedded devices. It’s used all over the video ecosystem already and so now combining this server with the Livepeer aware node into Catalyst creates this new concept that exists for the first time that we’re calling a Decentralized Media Server. I know I’m nerding out on video a little bit here as I tend to do but typically the builder of a video application could either deploy a media server and run it themselves as deployable software or they could use a hosted service, right like a SaaS service. What a Decentralized Media Server lets you do, is it lets you deploy and customize your own software which is really powerful as you’re building and scaling your application but access an open infrastructure network for things that need to scale like compute, transcoding, content delivery etc. So you have this really powerful combination that’s not just integrated with SaaS services where you need to sign up for accounts and have business relationships and whatnot but you actually have all the power of a custom customizable deployable software combined with infinitely scalable open infrastructure network and that’s one reason we’re really excited about that Catalyst. It’s open source. It’s available you can see the developer docs and get started building on it. I still think there’s a lot of work to do to fully productize it and create the right interfaces, make it really easy to use but you see applications building directly on it already and it’s a killer building block and it’s what livepeer.studio is built on as well that powers you know much of the streaming on the network.

Stephanie Dunbar (22:20): Awesome! And you know we appreciate when you nerd out on the video stuff, so it helps us understand it better and I feel like you know there’s a lot to decentralize like there’s so many different parts of the stack. It’s not as simple as just an app streaming and the transcoding network there’s just so much in between and you know something else you you guys are working on is a Decentralized Media Server API as I understand. Which is kind of like so you have Catalyst in livepeer.studio which are the nodes that are interacting with the transcoding network and then on the other side of that when you have an end user, like a decentralized YouTube and people who are watching the videos, you have to have something in between right to kind of like send to the broadcasting node. So would love to hear about your work on making that a more seamless process and when one can expect the standard Decentralized Media Server (DMS) API to be available for broadcasters.

Doug Petkanics (23:14): Yeah so this standard DMS API sounds like a really technical concept with a lot of acronyms right. What’s really powerful about this it’s the notion that anyone building on Livepeer whether you’re building against a hosted service like Studio or you’re running your own node, you’re using the same interface, the same API in order to interact with the video streaming capabilities and that’s really powerful for a uniquely web3 reason, which is it allows for something we refer to as infrastructure portability. It’s the notion that you won’t get locked into a single vendor, you won’t get stuck starting building on AWS and be unable to migrate your services to your own decentralized infrastructure, your own nodes or even another provider. I think anyone who implements this, the standard DMS API would make their infrastructure portable and give users that guarantee that hey you’re not being locked in you’re not being, yes we’re in an agreement of how we exchange value for for providing this service, but you have the control and you have the freedom to build how you want what you want and exit if you if you want to or need to. And that’s very different in the web3 space versus what you see in the traditional web2 world. So this is a powerful concept, still a lot of work to make progress on and finalize this. I can’t give like a release date but I think you’ll see in kind of a shorter term milestone is, much like inspired by the Ethereum RPC API, this is the same API that the Ethereum nodes use, Infura and Alchemy use which are popular hosted services for accessing nodes. They’re all implementing the same API that’s just defined in a wiki document that lives on GitHub, right. And it’s been iterated on over the years, everyone contributes to the updates and the next version and I think you’ll see a similar pattern followed here where you know there’s a wiki document put up where the community can give input, can shape what this needs to look like based off of how the APIs already worked largely and then it can become more of an open building block for others to build on as well.

Mihai Grigore (25:36): You refer to Livepeer as providing infinitely scalable infrastructure. Tell us a little bit about the potential capabilities that go beyond the current implementation of Livepeer that actually use GPU computing power. What are those at the forefront of development for Livepeer and where do you see this going over the next quarters and years?

Doug Petkanics (26:10): Great! Yeah I always get excited when I talk about this stuff and I also get myself in trouble when I talk about this stuff because I’m excited by it but you know it’s not necessarily live on the Livepeer network yet. At its core the Livepeer network is this big network of GPUs, you know Graphics Processing Units. Tens of thousands of them distributed around the world and these happen to have chips on them that are great for doing video transcoding but these GPUs are also used for more general purpose compute particularly when it comes to AI based tasks, machine learning based tasks things like rendering tasks with graphics and so even though all the Livepeer software specializes in how to make use of the video capabilities, there’s no reason that the Livepeer network can’t expand to perform additional types of tasks. I think there’s areas that the research team has worked on done prototypes on or even used in beta by some of the users that are AI tasks, compute tasks that are related to video, we’ve talked about some of these in the past, for example when your video is being transcoded by the network those GPUs can also apply an AI inference algorithm to try and detect whether this video is likely to be adult content, violent content, soccer content (which is a prime source of abuse believe it or not), copyrighted content right and if it detects that it can make that information available to the application developer so they can do with that info what they will. Do they want to flag it? Do they want to route it to user review? Do they want to have it be displayed in a separate audit section of their site, for example. It’s really valuable to have you know AI doing that as a first pass instead of humans having to do it especially for live streams right and so that stuff is further along it applies to video. There’s other tasks like kind of sub, like closed caption generates speech or video and audio to text. So if we’re having this live stream can we automatically generate captions. Great open source libraries like Whisper which can be run within Livepeer software in order to do that and you know that stuff’s the prototype phase and certainly any users that want to leverage that would be great design partners to work with to motivate how we go beyond just testing and building it into the Livepeer network to meet those needs. Then I think there’s a category of these types of computations which are interesting even beyond you know live stream and on-demand video. You may have followed the generative art, generative video, generative creative space through projects like DALL-E 2 and Stable Diffusion which is a great open source library. This is where people can give a prompt and these AI’s can create amazing artwork based off of that prompt, for use in many different use cases. And there’s so much experimentation going on in that world but the challenge is that anyone can create a prompt and yeah Stable Diffusion is open source but you need a large cluster of GPUs if you’re going to be rendering these images very quickly and returning them to users especially if a lot of users are doing this at the same time and you know Livepeer network is a big cluster of GPUs and our node operators are already experimenting with, hey like what interface do we expose, how do we run these things side by side, how could we perform Stable Diffusion rendering and so that’s really one of the you know exciting things going on within the community. I’d love to see you know whether through a grant or community driven project or just a concerted effort you know someone to take that from experiment to actual capability on the Livepeer network.

Stephanie Dunbar (30:04): Thanks Doug. And just I guess for some of our listeners that might not be familiar with the design so you have you know Livepeer node operators which are known as Orchestrators and they’re responsible for effectively transcoding pools, so you have GPUs that are also are transcoders, so the Orchestrators are the people in the network that stake and based on their stake, their performance and stake distributed towards them they route jobs to the different transcoders on the network. So I want to talk a little bit more about that mechanism because also this quarter we saw that an upgrade to the Livepeer Explorer in Q2 really helped some of the smaller Orchestrators gain profitability because it was easier for delegators to route stake to more effective nodes as opposed to just the ones with higher stakes. Wanted to hear from you more of a breakdown as to what determines if an Orchestrator and therefore their transcoders are assigned work, there’s a fee market on one side but there’s also the probability of being selected, so they’re stake weight, but I believe there’s other factors that you know add into this. Would you mind breaking it down for us a little bit?

Doug Petkanics (31:23): Yeah good, good stuff. So a few things to unpack in there. This notion of the Orchestrators, these are the nodes that you see when you visit the Livepeer Explorer and you see how much work they’ve performed as a proxy of the fees they’ve been earning in the last 30 or 90 days and you also see how much stake they have. How much stake they’ve staked and how much people have delegated towards them, you asked about how do these node operators win and retain work on the network and as you alluded to it’s a combination of a few things. So you know in the purest design of Livepeer and in the kind of game theory optimized outcome the amount of work that you perform on the network is proportional to how much stake you put down on the network. So if Stephanie has staked twice as much LPT as Mihai then Stephanie would perform twice as much work, video encoding work, on the network and earn twice as much fees. The amount of stake on your node is a representation of the security that your node is providing when it does that work because you have more at stake and more to lose if you deliberately miss encode the video and cheat and harm the network and so therefore it would be natural that a user would prefer to use a node with more security, where there’s more at risk, than a node with with less security. But when the rubber hits the road, stake can’t be the only determinant of who gets work because, hey, we’re live streaming video here, if someone doesn’t encode that video reliably in under two seconds your users are gonna not be have a good experience and so the stake weight is basically the kind of the first input into how the nodes get selected. The first and most important is just what nodes are near me, what nodes can I connect to that are near the source of the video, that would be able to receive this video quickly and return it to me reliably, kind of as a function of their bandwidth. So the broadcast nodes generate a set of node candidates based on who’s nearby them and performing well and then from there they take stake weight into account as a security input (that I had mentioned) and then they have redundancy configured so they may be working with multiple nodes at one time as sort of a backup and a primary depending on how much they’re willing to pay and how reliable they need the stream to be, right. And then their performance in real time on encoding the video is also a huge input so you know if I select Stephanie’s node and she has twice the stake weight it doesn’t mean that I’m going to send two segments of video to her and then one to Mihai. It means that I started working with Stephanie and I as long as Stephanie is encoding that video and sending it back quick enough I’m going to continue to work with Stephanie and it just means that Stephanie may be selected twice as often for that initial test as Mihai. But then if my stream happens to be really long, you’ll get a lot of work, if mine is really short, you’ll do the work and the job will end and you don’t know that coming in. So you do see some variation where nodes with lower stake who are actually really performant doing a great job retaining the work for streams actually end up earning a lot of fees and a lot a lot more work. So I know I’m talking a while but the last thing I’ll say here is why game theoretically it would make sense that ultimately the stake will land in proportion to how much work is being performed. So in that scenario let’s say you have twice the stake than Mihai, but you’re not performing well and Mihai is performing well with half the stake, he’ll start to earn more fees and what that means is the APY or the return that a delegator will get by staking towards Mihai actually will increase because the fees that he’s sharing back will be split up amongst less stakers. So our dashboard shows the APY that the node operators have returned in the last 90 days. If Mihai’s is higher people will start to stake towards him, they’ll provide more security on his node because he’s performing better, he’ll get more work as a result it’ll be less attractive to stake towards Stephanie and you’ll see that stake begin to to shift. If everyone was totally rational and everyone was paying attention to the network every day and shifted their stake around accordingly the game theoretic outcome is that you would see stake shifted towards nodes in proportion to how much work they are reliably performing on the network, how much capacity they have and the network would be secured accordingly. There’s some other inputs like their commission and everything but that’s the kind of design experiment. Seems to be working pretty well, I’m sure there’s room for improvement through governance and updates, but so far so good.

Mihai Grigore (36:46): Thank you so much for this Doug. I think understanding these dynamics of the supply are very important and perhaps moving towards demand and where this is going to continue to come from. I remember our conversation from the Messari’s Mainnet conference in September and you also wrote a blog post on the web3 killer apps. Could you tell us a little bit about what you mean by that, and what kind of apps do you have in mind being the next killer apps in this space?

Doug Petkanics (37:28): Yeah like I’m really excited for the next six months because you know the promise of Ethereum. The promise of these blockchain technologies was that we could actually make positive impacts for the world beyond just financial speculative use cases, right. Financial freedom is great. I’m all for the creation of empowering alternative digital first financial systems but these technologies also have the ability to impact things like media distribution and knowledge exchange and the global economy and you know social media is actually a huge area for disruption. I think we’ve felt so much pain around just a few centralized social media platforms in the last few years being the arbiters of truth and the distributors of information whether that information was true or not etc. Sort of these open networks, open technologies and blockchain based primitives are really powerful tools to actually disrupt some of those social media apps and I think we we have not yet seen the killer web3 social media app emerge yet but I’m like so excited by what I’m seeing built right now, particularly in terms of early experiments with social applications. Early experience with creator-centric applications, I’m actually calling this whole category Creator- to-Community type apps because it’s really about creators creating content and distributing it to their communities, that their fans are responding to the content or engaging with it. I think one thing that’s clear to me is that these killer web3 social apps are not just going to be clones of the social media apps we use today. It’s not just going to be Twitter but built on web3 tech or Instagram but built on web3 tech. I think it’s going to make use of the killer disruptive primitives that blockchains have enabled so far which have really been like instant global crowdfunding, you know permissionless global open financial system, digital asset ownership, collective coordination. Yeah ICOs, DeFi, NFTs & DAOs essentially. I think the killer web3 social apps are going to leverage and include these tools within exciting social and media based products. So I’ve seen a bunch of really inspiring apps building on Livepeer lately the402 and Bonfire are two great creator platforms for streaming and for distributing video based content combined with NFTs. Similar to this now called The Playground we’re doing an event with them tonight about creators having space for their own communities and dropping events content into them. Huddle01 the app that we’re using for this call right now is an amazing kind of conferencing application, which is sort of the first real incarnation of the metaverse outside of gaming, I think. This is the digital world that we’re inhabiting. I was talking with a gaming company this morning, they have this great soccer kind of manager style simulation game where you can actually own the teams and the assets digitally on-chain. It’s seeing a lot of growth and they’re excited to introduce streaming as a form of growth. So use cases like that are exciting and you have apps like Beem that are kind of serving the filmmaker community and kind of the higher end content community but allowing them to release their content early through crowdfunding NFT drops and whatnot. So you know just some inspirational examples that are being built on and around Livepeer that I’m really excited about.

Stephanie Dunbar (41:06): Yeah it feels like every time we chat there’s just so many more. I haven’t heard of the majority of these names yet, so I’m looking forward to checking them out. It definitely is exciting to see this ecosystem just continue to grow.

Doug Petkanics (41:23): They’re gonna break out soon and you’re gonna hear, everyone’s gonna hear about them soon, right. I think web3 social is going to have it’s Crypto Kitties moment or it’s Axie Infinity moment, within the next six months and we just want to make sure that Livepeer’s helping enable the video within those killer apps.

Stephanie Dunbar (41:40): Yeah we definitely agree with you on that. A colleague of ours, Dustin, recently wrote a report I think earlier this week talking about how there’s a connection usually between what’s being built at hackathons and then a few months later what you see in terms of projects like really growing and gaining market share. And you know in the past it has been like DeFi, NFTs and most recently it’s been web3 social. So I think also thanks to Livepeer grants but also you know Lens Protocols helping out with that there’s a lot of interest. So I do agree with you, you know six months I can certainly see a lot more attention and tons more really interesting things to check out. Anyway sorry shifting gears a little bit the last question we have for you is about something else that’s being built at with Livepeer and it’s the verifiable video streams. As I understand it’s a new on-chain integration to help with verifying video and connecting that to Ethereum. I’d love to hear from you a little bit more about this really interesting new capability.

Doug Petkanics (42:45): Yeah this topic gets technical fast, but let’s maybe just start with where users would feel pain around this right now. So in web1 we saw kind of text and information being like the primary media source and then all of a sudden images became a popular media source with the advent of things like Flickr and then sure enough video became the predominant source of media on the internet with YouTube, right. I think in kind of web3 if you look at NFTs as the representation of media initially a lot of them have been heavily image based, but it’s inevitable that we’re going to see more experimentation with what it means for you know video based assets to be probably owned and distributed and have royalties around them and all the sorts of experiments that you see in this exploding NFT space. So video, like an image is kind of just a file a single small file and it’s easy to verify that the image that you downloaded is the original image that a creator created that is associated with the NFT. IPFS lets you do that pretty easily through the content addressable hashes associated with every image. Video is a different beast. If you want video to play well over the internet you actually have to transcode it into many different formats and bitrates, you have to distribute it via CDNs and have it be cached, you have to distribute it in little pieces instead of all in one file. So a challenge right now is if someone views a video NFT on a marketplace that NFT marketplace had to create their own copy of the video, they had to have it encoded they have to cache it and distribute it differently and so the file that a user is viewing or a file, more importantly a file that a user is buying in terms of an NFT, it’s very hard for them to even know that they’re getting the content that the creator signed and originated that’s associated with this. Especially if you want to be able to deliver and play it back quickly and so you know Livepeer is doing a lot of work to figure out what does it mean to be able to verify this video is correct, especially as it moves across NFT marketplaces, they all have their own copies, they all have their own encodings and whatnot to verify what you’re viewing, that you’re seeing the content as distributed by the original as created by the original creator. And you actually have that sense of content ownership which is so core to web3. And yeah Livepeer has a verifiable video project that we’re excited to be making good progress on and pleased to deliver this impact for the world.

Mihai Grigore (45:30): With that we’d like to thank you so much Doug, for the great conversation and for our audience at Messari we continue to do quarterly reports on Livepeer. So we will follow Livepeer also closely  in Q4 and if you’d like to follow our work please go to Messari.io or on Twitter. Doug where can people follow Livepeer?

Doug Petkanics (46:03): Yeah visit us at livepeer.org the home of the project. If you’re building and want to kind of use our gateway visit livepeer.studio the best place to get started building your video app and we’re on Twitter, so look for us there. Join our Discord, join our community, a very welcoming community that’s been around for six plus years. So you know a great place to kind of get your intro or go deep into web3.

Mihai Grigore (46:30): Stephanie and Doug, thank you so much and looking forward to talking to you next quarter.