Originally aired on January 27 @ 2:00 PM - 2:30 PM EDT
Join Wesley Evans (Product Manager, Cloudflare), Nick Sullivan (Head of Research, Cloudflare) as they discuss the maturing Web3 market and the future of the distributed web with guests Dete Shirley (Chief Technology Officer, Dapper Labs) and Dietrich Ayala (IPFS Ecosystem Lead, Protocol Labs).
Read the blog post:
Find all of our Birthday Week announcements and CFTV segments at the
Birthday Week hub Cool. And it's noon. Thanks for joining everybody and welcome to Cloudflare TV. My name's Wesley and I'm the PM for the research team at Cloudflare and also the distributed web. And welcome to our segment today. Our guests for our little round table are Nick Sullivan, the director of research here at Cloudflare, Dieter Shirley, the CTO over at Dapper Labs. And we have Dietrich with us from Protocol Labs, who's an IPFS growth engineer over there. I'd love to have our panel introduce themselves more in depth. So Nick, why don't we start with you? Sure. Hello folks. I'm Nick Sullivan, lead the research team here at Cloudflare. Really excited to get this conversation going. We've been experimenting and building products in the distributed web space for a while. And we have two really great guests that were willing to join us today for Cloudflare's birthday. I know it's super exciting. Dietrich, why don't we go to you next? Yeah, sure. Hey everybody. My name is Dietrich Ayala. I'm with Protocol Labs. I've been here a couple of years working on IPFS and related things. I was at Mozilla for 13 years before that. So working on Firefox Core, Firefox OS Project, and DevRel. Now working on browser collaborations and other types of things here at Protocol Labs. Awesome. And we also have Dieter Shirley with us. Diet, why don't you introduce yourself for us? Hey there, everyone. I'm Dieter Shirley, also go online as Diet. I was the author of ERC721, a co-creator of CryptoKitties. I'm now working as the CTO of Dapr Labs on NBA Top Shot and the Flow blockchain. And so obviously, all things Web3 and NFT are a big thing that we're tackling here at Dapr Labs. Awesome. Thanks. Why don't we go around the horn a little bit? Because I know we're talking about Web3 today. And we're talking about the future of Web3 and the maturing of the market. I think all of us have had big announcements in the last couple of weeks. And I'd love to hear more about what everyone's working on. Cloudflare today announced that we're moving our distributed web gateway projects, which we have actually been working on for about four years now, from a research product, public alpha, into a private beta as we get ready to actually start scaling them as full enterprise products that will be across all tiers of service. Diet, I know you guys had a big announcement earlier this week with the I know Protocol Labs has also been working on some cool stuff. So I'd love to hear what everyone's sort of been working on for the last couple of months. Yeah, sure. I'll jump in first here. So yeah, I mean, as you mentioned, we announced NFL this week. We also announced La Liga this week. And I think I've been sort of been able to broadly hint to people that we've been talking to other leagues. But I think that's a you know, it's it makes a lot of sense that once people saw the success of NBA Top Shot that other leagues would want to get involved in this. And so, you know, so we're sort of going to the leagues that we feel are going to make the best partners for us and working with them on similar products. You know, the NBA took a real big risk on us signing an agreement with us back in 2018. I think before this idea of sports collectibles as NFTs was was sort of, you know, well established, and they've just been an excellent, excellent partner. And so we're really excited to bring more leagues onto that. And especially excited about NFL, because it's obviously, you know, really, really intense fans. And I think it's going to be a great way for them to express their fandom. On the technical side, you know, I think the flow blockchain, which underpins all this stuff is just, it's just been moving super fast. It's a, you know, it's, it's been live for about a year now, but it's still a relatively early stage project. And so, you know, we're, we're working really hard to help people understand how to build on top of it. We made a lot of decisions in flow that are different than the way other blockchains are built. We made all of those decisions from the standpoint of being dApp developers ourselves having built crypto kitties. And so we think that all of those changes make life easier for application developers. But, you know, if you know, some of those differences need a little bit of explaining. So we're working hard on the docs. But, you know, all that stuff's moving forward. And we're seeing lots of excitement in the community for building on top of it. Awesome. I know. I mean, personally, I'm a massive fan of flow myself. Another protocol I'm really turning out to be a big fan of too, built on top of IPFS is Filecoin. Dietrich, what's new in the Filecoin world as well as IPFS? Yes, the bringing those two ever closer, reducing friction, helping people onboard more storage, especially as NFTs blew up our world. I mean, they were using IPFS for storage. So many chains were NFTs. But then of course, you know, earlier this year, massive increase in interest. So many people, developers, especially coming into our channels. How does this stuff work? How do we get it to where, how do we put our stores into IPFS? How do we store it in the Filecoin? How do we make sure that this data stays accessible, resilient, make sure IPFS is there to avoid rug pulls. So we've launched a couple of different services over the last couple of months. NFT .storage, making it very easy for NFT platforms to be able to get their assets and metadata onto IPFS and Filecoin. Web3 .storage for storage of even larger files and working directly with individual developers to be able to build tooling and reduce friction to be able to use these services, but also with platforms to be able to onboard and mass and in bulk for these needs. And then finally, estuary .tech, which is a basically unified IPFS and Filecoin node that handles a lot of the kind of deal logic and abstracts it away. So it makes it easier for storage providers to be able to accept more deals faster and onboard storage, but also platforms and individual developers to be able to onboard into this world using more HTTP friendly APIs until we get things like IPFS supported more natively in browsers and places like that. That's awesome. And I know that's also something you spend a lot of time too, thinking about, you know, how do we do native IPFS support? One thing I heard from both those answers and both those announcements are, was the word developers. And so having an ecosystem of products and of services online, you need people to build them. So for, I guess, each of you, we'll start with Dieter and then go to Dietrich. Can you tell us a little bit about what the experience is of a distributed app developer? And maybe some of our audience here may not have experience developing things in a fully distributed way or using decentralized tools. How would you just describe what it takes to go from developing traditional apps to something that leverages this new landscape of technology? Yeah, I, it's a tough one because it's, I think, I think that there's a, there's a few aspects, right? One is that it, as a centralized developer coming into this space, unfortunately what people see usually for the first thing they see is what they're losing. And it's harder for them to see what they're gaining. But if you're willing to sort of let yourself, like, you know, just sort of like let yourself, you know, look for those, look for those upsides. Right. I think that's, that's the most important advice I can give to people. What, what people don't realize is it's entirely plausible to build a decentralized app today, such that you don't have to run a single server anywhere. That your logic can be mediated by smart contracts, that your, your front end can be still hosted on IPFS. That, you know, maybe you want to have some sort of gateway like static URL, like, you know, you know, mydap.com or something like that, that points at it, but you don't have to, you don't have to. And when I say you don't need to run a single server, I mean, you don't even need to pay Google or Amazon to run your servers, right? Like all of this stuff can be done in a completely decentralized way where you can disappear from the face of the earth and stop supporting this and your users still have access to this application. That is like, yes, I'm stealing Dietrich's talking point here, but it's, it's, it's, it's an amazing thing. Now you're giving up some level of control then, right? And, and, and, and we're still in a transition period. There are certain things that aren't yet well established how to do them. Right. So I'll give a really concrete example with CryptoKitties, right? So CryptoKitties does have an off-chain backend. We do have a Postgres database with all the kitty data in it, and that powers the user search, right? I want to go and I want to look at all the cats with blue eyes. It's not convenient to, to ask the blockchain. It's not efficient to ask the blockchain to give you that data. And you can't just store that data. So like sort of as a static index and IPFS. And so right now today, we don't yet have robust tools. Now, maybe this is, can lead off into what, what Dietrich would like to talk about. Cause I know that the community is working on this stuff. And, and I think that that we'll get there, but right now today, there is going to be a bit of a transition time where most projects will probably need a little bit of backend, but probably way less than they think. And I think, I think that's probably the most exciting part of it, but honestly, I think it's the mental shift more than anything else. So let's I want to hear what Dietrich has to say on this one. I was really disagreeing. Like it's, I think that longevity of availability of applications is one of the unsung wins of the space. Like that, what the, the small business turnover rate in the United States is something like 90% per year. Like we want applications that work for people that can stick around for as long as people want to use them. And that's part of the vision of like, for me, of why it's so important to be here. So I was just really agreeing with that part of it from a developer onboarding perspective. Yeah. There's some challenges for sure. And I really agree also that the kind of mental shift around different primitives and different network architectures, different forms and places where identity can come from and what that means, different trust models, that shift is harder, I think, than the technology. I've ran for, since February, a hackathon-a -thon program where we tried to basically be in a hackathon all the time. And now we're at about five hackathons a month that we're doing. And we are learning a ton about what new developers are, or what the challenges that they're having in the space. Because about, I don't know, it depends on the event, right? Like some, like a trailing hackathon, a lot of people are already involved in the space. But I mean, what we're seeing for ETH Global and some of these events, more than half the people that join are new. And we've also done some hackathons with like indie game developers who are all Web2, where it's their first experiences with Web3 at all. And I think there's a lot of people don't understand the superpowers that they get with things like data portability. So like, we've talked about data portability in abstract. Now we have tools that kind of make that possible. And seeing, especially like seeing game developers recently be like, oh, like an NFT of a boss battle that I can share, that my users can share socially outside of the game that lives forever. Like these are really, really powerful, basic primitives of game building that now can be interoperable and portable. The shift around that, it's going to take some time. The tooling to catch up, absolutely going to take some time. I think that indexing, like you're talking about data, is going to be one of those ones where like, how do we do, how do we make it so that world, you don't have to have that big Postgres instance behind it. I think those are going to be some of the key pieces that fill in as these technologies scale up. But we're just seeing a lot of people, one, new to the space and two, willing to kind of hammer it out in some of these events. I think that's so fascinating. I mean, you know, I talked about it in my blog today and I love that you sort of keyed into it, Dietrich, and this idea that we're finally getting people to start understanding some of the superpowers. And so much of that superpower is content creator content, or creator content that's made by creators, like indie game developers that can be moved as NFTs into new spaces. I mean, we were talking about in the context of the metaverse today in the blog, and it's something that Thibaut, myself, and Nick have been talking about for a long time that I think one of the superpowers of the space is data portability. And this idea that you can suddenly get into a world where you have really easily moved authorship and ownership. I think that's what's so fascinating for me about what's probably coming next. And what, you know, I think the distributed web team at Cloudflare here is so interested in is like, how do we help build some of these next generation metaverse concepts? How do we help build some of these next generation networking stacks that, you know, all these developers are going to need? And how do we make it as easy as possible? Because I know for myself, when I was trying to get into the space, it can be super confusing. You know, so that's why we built Gateway in a lot of ways. You know, how do we make this as easy as possible for people to make that web two to web three transition? Yeah, it should be around a little while. I think it's one of those things that's hanging around. I don't think we have any RFCs getting rid of it currently. So I think it's hanging on for a bit. You know, Nick, you were here four years ago, and we started talking about even the possibility of doing the IPFS and Ethereum Gateway. You know, what's changed in the market for you over those four years? What makes this such an exciting space now? Well, I think when we first started exploring, there were a few applications out there. And it was a relatively niche market. There were sort of innovators in the space, like CryptoKitties, for example, who were exploring what are the possibilities of things we can build in this world. IPFS had been around already for a couple years, and people were building stuff on top of it. There's this sort of like early, early interest. And I think over the last, particularly over the last year, there's been a real acceleration of and interest from developers and interest from end users on technology that's built with some of these great properties that we talked about that come from decentralization. So I'm excited. There's always new ideas. There's smart contracts that have been tried and true and get reinvented in many different ways for different applications, whether it's for, as we mentioned, some of the examples here or in the art world. It's sort of really found a market there. So yeah, it's kind of come from a place where it was a lot more hobbyists to what's becoming, I guess, a relatively big business. Well, I think the other thing to interest, like, I'm going to ask a leading question, which is we've been working on this for four years. Why are we talking about it again now? I mean, obviously there's a market opportunity here, but what have been some of the technical challenges that Cloudflare have had to overcome to even put us in this position to talk about some of this stuff? Yeah, I think one of the hardest parts about decentralized technologies and one that most people who are casually acquainted with it don't really appreciate is just the difficulty of networking and doing network software. The Internet itself has been around for decades and decades, and we have TCPIP, and we have UDP, and we have HTTP, and a lot of these things have been tried and true and tested and have made up the Internet as we typically use it. But coming up with something that works in a peer-to-peer manner as an overlay on top of the existing Internet has been a challenge for a long time. Think back to BitTorrent or file sharing in the early parts of the century. That software wasn't always the most reliable, and it had some interesting, difficult challenges that had to be solved. So what we're seeing now is a lot of commitment and work and resources being put into making sure that we have software that can communicate in a peer-to -peer way that is reliable. It's something that you can actually run, that developers can use, and that it's not a huge burden to be able to interact with. We're really excited for our gateways because while we've learned how to run these nodes, IPFS nodes and Ethereum nodes, and keep them up, which is something that is sometimes challenging, but this idea of having the network layer itself kind of hidden from the developers is something that enables a lot of innovation. And so that's kind of why we're talking about our gateways again, is we have so many more developers interested in the space, and there's a lot of the tools that have been part of the Internet for a while, like HTTP, are things that we think is a good gateway. Pun intended, right? It's a good way to get involved with this technology and sort of take the next step. I want to jump in on that one, if I may, because I think Nick was talking about the supply side, but I think what we're seeing this year in particular, there's actually a marked change on the demand side as well, right? One of the challenges, one of the other challenges of decentralized technology is when we do it right, it doesn't look any different than the centralized technology, right? And now it has properties that centralized technology doesn't have, but if the users aren't aware of those properties, aren't asking for those properties, then all it does is it just, it's a bunch of complicated education for engineers, it's occasionally some delays for users. And if they don't know what benefits they're getting from that, then it's just making life harder for everyone. And I think what we're seeing this year is, and I will put a lot of this on the NFT boom this year, is that people are realizing that this stuff matters, that this isn't just a niche thing, right? We have a million people signed up for NBA Top Shot, and we have thousands of developers building on Flow right now. This is something that people are really taking notice of in a wide area. And it's because they're starting to finally internalize that what we used to call the cloud is just somebody else's computer, and this decentralized web is a place where you can have things, you can put data there, and the data stays, you can have code there, and the code runs, and you're not dependent on somebody else's infrastructure to make sure that's all running. Now, somebody else's infrastructure is, it needs to be there, right? But the neat thing about all of these protocols is that you can swap in alternatives, and you can, like Cloudflare, right? They're always going to be one of the fastest IPFS gateways, just because they have all that edge computing and stuff. And so using Cloudflare for this is great, but you're also not locked into Cloudflare. Now, maybe that makes Cloudflare a little sad, but as users, that makes life much easier for us, it makes it much easier for us to adopt it. So I think the big part of this is that people are understanding the value of having systems that can remain running and robust, even in the absence of the teams that created them. Well, I just want to piggyback right off that, because I think you said something really important there, which is, and it's a little interesting, right? The idea of there's not vendor lock -in. And I think that's actually something that's really important for us to talk about. Cloudflare is a really big network. I mean, it's one of the things that, let's talk IPFS, right? We have the ability to cache IPFS content in more than 200 cities around the globe and serve it to users in milliseconds. That's a big, big, big benefit for the network. At the same time, if we go down, there are other IPFS nodes that can serve that content. I mean, I think what's so interesting from my perspective is that this isn't just building cool protocols. If you go back to Cloudflare's mission, our job is to help build a better Internet. Well, a better Internet is an Internet that your application stays up, your data stays persistent, systems keep working, even if you have a vendor failout. And I think building a world where we have those types of systems is really, really important. So that's why we're also really interested in these systems. It's because we can help build a better Internet through it. And I think that means, look at what we did with R2 storage this week. I think it's really important that we try to help build better systems beyond certain centralized vendor components. And I think we're doing that in a lot of different areas. So yeah, it's definitely an interesting idea from a business model perspective, but it's a mission driven perspective, certainly. And then Dieter, I want to get you in on this too. I mean, IPFS is certainly one of the most interesting protocols and certainly Filecoin as well. What have been some of your scaling challenges? What have been some of your technical constraints that you've been working through at your side? I think, like you said, demand side has presented some new challenges, great problems to have. It's an exciting group. And so we've been really scaling up our IPFS gateway, of course, and we will work closely with you all in figuring out how those gateways are going to work, how we add more features and capabilities to those gateways. I think especially as there are primitive use cases, even in an NF2 world, and those had primitive demands on a complex system. As the use cases have become more sophisticated, we need to do more things through the HTTP gateway, as well as parallelizing our presence natively in systems. So some of the scale is definitely being able to onboard storage onto the Filecoin network, be able to scale up deal size, deal speed, and be able to really smooth out the pathways between saving things to Filecoin in a way that keeps it available permanently on IPFS for those people that are archiving things that way. I think that there's some client side as well, and that we definitely hit that in the... And maybe scaling isn't the right word really for there, but let's say you have an NFT that goes viral. They're all hitting a HTTP URL for that IPFS CID, and we see our gateways spike, right? The system is designed such that you could take that CID and just plop it at the end of Cloudflare's gateway and distribute that load. But we don't really have the flexibility in the HTTP URL system yet, or client side logic is not really developed as well enough yet to say, randomize between different IPFS gateways, because that's a thing that you can do, and that's built into the addressing of CIDs, that it doesn't matter what gateway you get it from, you can trust that that is what you asked for. I mean, not necessarily with gateways all the time, but with the gateways that have verifiability and that you trust like Cloudflare or like an IPFS. And as soon as we can push that trust out, or trustlessness actually, out into the client, then we have less of these types of scaling issues. We can really start taking advantage of the native resilience, reliability, and trustlessness of IPFS addressing. So that's why HTTP gateways are great for now as a gateway, a bridge, on board more people into that space, but ultimately the scaling is designed by the characteristics of the protocol and the protocol design itself, and IPFS. That's why we need to push that native IPFS out of the edges as soon as we can. It makes total sense. Looking at, we have about five minutes left, I want to do one final round the horn here. We'll start with Nick. Look five years into the future. What does the future of Web3 look like? What are people building? What are people focusing on? Well, I certainly don't have an answer to that. I think that's something that I'm excited to learn about through interacting with the community, with looking at how customers use our gateways, what sort of features are requested. And I think five years is a very long time. Think back to how five years after the Bitcoin paper came out, what the world looked like in that amount of time. So I think there's room for new inventions. There's room for new business models. There's room for new technologies. But I think a lot of what we've seen over the last five years is going to continue and get stronger and gain use cases, both on, as Dieter said, on the supply side and on the demand side. Kind of a cop-out answer, I know, but I can't- It's not that much of a cop-out answer. It's fine. Let's go to Dieter. Which Dieter? You've used Dieter for both of us now. Oh, guy. Sorry, guys. We'll do Diet, and then we'll go Dietrick. I am sorry. Excellent. So I have a hope, right? I think that what we saw, I'd try and draw the third dot when I have two dots and assume it's all going to be a line, right? So that's flawed. And I'm not likely to know any better than Nick, but I'm at least willing to make a guess. So I think that what we saw when infrastructure as a service started becoming a way of building applications, and suddenly, as a software company, you didn't have to go and buy a fleet of servers and set up your own data center in order to get started. What we saw was software going from monolithic to more modular. We saw that applications were a little bit more focused, and that people would use more of them, right? So it used to be you'd buy Microsoft Office, and you were done, right? And then for a period of time, you'd use these suites. But now, we use so many different tools in our day from so many different companies. I think that the next step is to go even one step further, where today, we use different things for different data sets. But I think the true power of this decentralized stuff is that it's going to let us use different tools for the same data sets. So right now, if I want to talk to you on Slack, we both have to use the same user interface, because the data storage is in Slack's interface, or Slack's data centers. But if we can have a decentralized Slack, then I can have a different interface than you do. I can use a different front end than you do. And I think that that's maybe not a great example, but I do think that that idea that we'll have lots of little tools that are layered onto the same data is going to be the big difference there. And I'm here for it. I think that more people doing narrower things, but having those narrow things work together is going to be incredibly powerful. Awesome. Thanks, Dieter. Dieter? Yeah, that was fantastic. I feel like Dietrich, Dieter, we're on the same page. Before you joined, Wesley, we realized that we had exactly the same name. We just go by different names. But that's a great summary, I think, of the vision. And one of the reasons why this technology is so important to me personally is that I think, and what I want in five years, is for us to have ways to be able to have more choice in how we interact with other people online on a daily basis. And I see that happening in public networks, what other people might call blockchain. For me, that's a public network for compute and for storage, logic, public storage and distributed storage. I think that's going to be a regular and a default. We'll be in a better place Internet-wise, an Internet that works for us instead of extractive industries, if we can get to where distributed storage is the norm and open markets for storage. And that's really what we're trying to do with Filecoin. And then the long-term resilience of data online. You keep talking about this idea of applications that outlive companies. DNS certificates are rooted in companies, and those are transient and fragile. And I think that we'll be in a great place in five years if more and more applications and developers and end users are using things that are shipped on distributed front ends, the BYF pattern, using distributed storage, and then using public networks for distributed compute and application logic. Thanks, guys. We could not have timed this better. We have about 15 seconds left here. So I just want to say thank you to all of our guests, Dieter Shirley, Dietrich Aelia, Nick Sullivan, and myself, Wesley Evans. If you're interested in the distributed web, check out Protocol Labs, check out Depper Labs, read Cloudflare's Gateway blogs. We are very excited about the future of Web3 here at Cloudflare with all of our partners. And with that, wishing you a happy afternoon and a good day.