Peer-to-Peer Computing with Fluence Labs’ Co-Founder

A transcript of key takeaways from our fireside chat with Tom Trowbridge

Introduction (0:00)

Chris Tse: Hello Cardstack community. This is Chris Tse, the founding director of the Cardstack Project, coming to you from the land of the Web3. Today, we’re joined by Tom Trowbridge, the co-founder of Fluence Labs to talk about not just blockchain applications but peer-to-peer applications he’s been working on. He’s a veteran in the blockchain space, and we’re so glad to have him join us for this fireside chat. We can take a look at Web3 in this inflection point and see where we can go from here and what the new applications and use cases are that we can capture with research and development. Welcome Tom.

Tom Trowbridge: Thank you. It’s great to be here. I appreciate you having me.

Chris Tse: Tom and I actually met about a year ago in the middle of COVID in Soho to chat about what we’re working on at Cardstack. Obviously, he was at the beginning of this new project, Fluence Labs, which is now making a lot more progress, and I want to give him an opportunity to talk a little bit about his journey into crypto and Web3, what he’s working on at Fluence, and what makes him excited about our near future.

Tom Trowbridge: Thank you. First a little bit on my background: my introduction into blockchain DLT was joining Hedera Hashgraph back in 2016 or 2017. I joined Mance and Leemon, who had a whitepaper at the time for a company called Swirlds and were looking to build a public ledger. So I joined them to help found the public expression of the hashgraph consensus algorithm. From a standing start, we raised a bunch of money, hired a lot of people, and brought in a blue chip governing council of companies that would actually own and govern the network. There were a lot of things to do very, very quickly. I kind of joked to people: When you’re building especially on L1 with a lot of other protocols, you are effectively building a global organization from day one. You don’t get the luxury of just having to focus on one little market and spanning it.

It was definitely baptism by fire, and Hedera is doing very well — it’s a top 30. There have been some great use cases announced on it just over the past week. From an institutional side, I think it’s kind of unparalleled. But Hedera still doesn’t have Web3 traction and really some of the credibility in the crypto community. But that will come at some point. It was a terrific introduction to the technology and to the choices and tradeoffs that people make with regard to different types of architecture. One of the key tenets to understanding crypto is first understanding blockchain. Then you have to understand Ethereum, and then other L1s, and then from there, you kind of branch out and start to grasp the whole ecosystem. Those are the basic fundamental elements that one needs to explore and have command of before actually expanding much further beyond it.

How do the various crypto communities resonate with P2P application stacks? (5:26)

Chris Tse: There are a lot of different communities in crypto. Many DAOs are cultural communities, like private clubs. Then there are development communities, with people who are interested in blockchain. And then there are people interested in programming languages and distributed applications. Which community do you think resonates most with this idea of a peer-to-peer application stack compared to a community interested in digital assets, enterprise or otherwise? How do you define the community where you believe you have found that mutual appreciation?

Tom Trowbridge: That’s hard because it crosses a lot of them. There are DAOs — and we are going to launch a DAO later this year, which will govern our protocol. That element is certainly a direct link to those levels of community. Allowing that type of decentralized organization to govern the protocol is super important and something that we certainly believe in and are moving forward towards. With regard to DeFi, that is a community which is not an obvious one for us because it doesn’t enable financial transactions out of the gate. But what it does do is: Composability is a huge core of what we’re doing and what we’re going to be offering; it’s the value that Fluence will ultimately bring to the table.

Fluence sort of rhymes with the DeFi community, and while it may not be a next stablecoin lending protocol if you will, the lessons of composability are things that we’re taking on. People who are invested in and understand what drove innovation in DeFi will appreciate the composability of what we’re doing. We’re trying to take composability and make it more applicable to the whole software world. That is pretty powerful when you start to understand and see how well that’s been exhibited and deployed in DeFi.

Computing in Web3 (13:01)

Chris Tse: Compute in the previous era was like, “let’s do compute” and then there were projects like Golem or Livepeer. It was mostly on commodity rendering and tasks. When we say “compute,” we really mean microservices, the pricing service that allows you to do surge pricing for Uber. That is the “compute” we want: real business logic, real value, not just commodity “I’m going to crunch some numbers and compress some video.” Not that that’s not valuable. But what’s cool is that those projects pioneered this type of payment system. Some of them do probabilistic micropayment. The attribution — whether you are storing files and getting some Filecoin or you’re running a program — is really where we can borrow from these more commodity-based versions of crypto infrastructure and use them for what is the most interesting thing: what if the backend service of the world is composable and is built as one single build but not controlled by one single company?

Tom Trowbridge: Correct. And any smart author can come up with new ways or useful services that can be uploaded and then discovered and incorporated into that — no matter where they are. They don’t have to be in Silicon Valley or New York City or Berlin to create code that is useful and used. They can be anywhere; it can be uploaded, made discoverable, and then they can earn from that. That’s also super interesting because it democratizes the accessibility of useful code globally as well.

Localities and global blockchain distribution (14:43)

Chris Tse: Is there a locality to where the code goes? Or does it replicate to all the peers? On Ethereum, for example, if I run a laptop with Ethereum or even a data center, it runs the same transaction for the same block all the time. What do you think about this global distribution and parallelization that is possible when your compute is not on that base blockchain layer?

Tom Trowbridge: It’s interesting in a couple of ways. First, as a host, you’re free to choose whatever you want to host. So you may want to host some things that may make sense for you and some may not. We ultimately think there’ll be such a number of services that not every host will be able to or want to host everything. First of all, as a host, you choose — point one. Point two, as you author something that’s pushed out to the network, you as the author no longer have any control over that whatsoever. That code is out there and is running. You couldn’t stop it if you wanted to.

It also means you can’t insert any malicious code into it anymore because it’s out there and is gone. Point three: you want to upgrade it? Fine. You can set a new upgrade cycle for it. But if I’ve built an application running on V1, I’m in no way forced to do V2 as long as I keep paying the host for V1. My dependencies within remain consistent and unchallenged and valid, given every host is going to continue hosting something that people are paying for. You can break this kind of constant upgrade cycle that requires you to retool because who knows what breaks in your application, once one of your core modules is updated. That to us is also a very interesting part of it all.

Chris Tse: One of the challenges with centralized applications is not only version upgrade, which they don’t announce and then there’s downtime. The way that venture capital funding for SaaS products works is that a lot of them have plans to get acquired. And once they get acquired, the acquirer may not want to keep that service around. How many products have been killed by Microsoft and Google alone? This idea of the survivability of software and the data around the software seems to be a distinguishing point between a decentralized version of the software economy versus the centralized one.

What are some use cases you’re seeing in decentralized computing? (22:12)

Chris Tse: The imagination of these new projects coming in that may be one step beyond the acronyms of NFT, DeFi, and DAO — I consider that one cluster of EVM-enabled projects, and we’ve seen implementations on other chains. What do you think is the next area for decentralized computing? You mentioned Snapshot, which is currently the leader in governance, a really important thing. We’re seeing people coming in doing more enterprise type of things, like tools to run your DAO. DAO tooling is different than just making a Discord server and gating it. What type of use cases are you seeing that people are seeking out for decentralized compute?

Tom Trowbridge: There’s a couple. One, decentralized identity and decentralized messaging, both of which are actually linked because you can’t have properly decentralized identity without having decentralized messaging.

Chris Tse: And vice versa. Who am I messaging?

Tom Trowbridge: Correct, exactly. But decentralized messaging or identities now are shared via encryption, which is great but still centralized. Decentralized messaging is a critical key thing and we actually did a demonstration back in December of 2020. We built a decentralized messaging application live at our launch, and we were able to do that in 10 to 20 minutes to show what’s possible. I think that is one key piece of it. Just stepping back for a second, I think what we’re seeing globally is increasing observations of censorship and increasing observations of network failures from scale becoming more and more complicated and leading to more and more issues. Those are two things that are going to drive interest in both messaging and applications, for which those are problems.

Chris Tse: It’s interesting to look at centralized messaging as what has been gluing together this decentralization movement with Telegram and Discord. We succeeded as an industry, despite not solving the decentralized messaging problem, not because we didn’t solve it. When it is actually solved and the rails of the messaging carries with it the message of business and finance in more substantial ways, it will connect us to users’ lives a lot more. It seems like we’re chatting about something and then going to very convoluted means to settle using MetaMask and stuff like that. You’ve got to give the team at Status a lot of credit for first imagining this combination of the Ethereum wallet with a messaging solution. If you look at it zoomed out, there are a lot of people working on this.

Data science in DAO organization (34:41)

Chris Tse: When we are dealing with smart contracts, we’re dealing with sixth-grade math. It’s mathematics that you can teach a sixth grader, even the most complex DeFi project with rebasing and stuff like that. It’s not that hard to teach. It’s addition and subtraction and multiplication most of the time. With Greek symbols in the white paper it can be sort of difficult, but if they’re in Excel it’s pretty obvious. What I think we are missing is using data science to look broadly at the whole. It’s very hard to fake when you have a lot of signals and patterns and what they call features. Part of the value that we’re seeing now in the proprietary analytics space like Chainalysis and stuff like that is that they can figure out intent and identify people to subpoena. But that type of analytics on top of blockchain or public data is not available to a DAO.

Part of it is bringing more data science into the picture so we can have better signals to say: “Yeah, you can have an NFT. You can participate and vote.” But it’s also like, “could I use more analysis to look at that same set of facts and give better gradation signals that then get contributed back to the on-chain decision?” This idea of adding data science seems like an opportunity that’s missed because I think analytics has been relegated to a report dashboard for Nansen or something like that on the side. It hasn’t been made core to our value judgment about what’s going on. We can bring some of that in. We, as an organization, have a data science team, and not many organizations building these kinds of things do. But we understand what’s going on and if we can productize that, then maybe we can bring more of that pattern seeking into the question of who is contributing and who can be fairly compensated and empowered.

Tom Trowbridge: You need the data to do that, and so problem number one is that you need to get people to participate. And besides voting participation, you must have signals beyond voting and that gets even harder. What you seem to be working on is a very difficult problem, and I am anxious to see some of the output of what you do because that’s hard.

Chris Tse: The hardest thing about data science is: how do you get the data? But the good news is: while we don’t have the data now, the nature of this data sharing and storage and compute and these kinds of artifacts being captured with base and storage layers is that this will be the most open bottom-layer data lake that anybody has ever seen. And state level players already appreciate and get the benefits here. But I think the tools to analyze that base-level layer are not there yet. There’s very few people in this land and very little data, but there’s a lot of land and you can see them from the top. There’s no blockage on it. This is a longer-term thing, but in the short term, even taking into consideration Discord or GitHub activity in governance, which currently isn’t done — some people do it through source credit and stuff like that — that would be a direction. At least this is what we think of as we think about fair attribution of power and fair attribution of rewards. I think those are quite related.

Tom Trowbridge: Yeah, I agree fully.

What is the most likely unlikely event to happen in crypto during the next year? (38:06)

Chris Tse: We’re in a very interesting juncture in crypto. I saw the Coinbase ad “Crypto is dead” where there’s another tweet from two years ago, “Crypto is dead.” And then from 2013, “Crypto is dead.” I thought that was brave and poignant, especially only four months after the Super Bowl with the QR code bouncing around like a DVD menu. That’s crypto for you. Look at the juxtaposition of mainstream broadcast advertising in the sporting event. What do you think is the most likely unlikely event to happen during the next year?

Tom Trowbridge: The most likely unlikely event.

Chris Tse: Yeah, like your prediction for crypto in the next year.

Tom Trowbridge: One thing I’ll say — and maybe this is likely or not: I think the main reason El Salvador accepted crypto was to engage in state sponsored monetization of their geothermal resources. I think other countries are taking note, and so we could see another country effectively start state-sponsored Bitcoin mining, which will probably not be a first world country. It’ll be a country with excess natural resources, whether they’re hydro, geothermal, whatever it is. That’s sort of one thing.

Chris Tse: Essentially, free energy to the state actor.

Tom Trowbridge: Correct. And why not monetize that? You’ve got unused turbines in a variety of dams in Africa. Why are those not being monetized? A Bitcoin-based bond is another thing. I also think we’ll see a huge — this is a little bit of bias — first transaction-based use case in crypto. Hedera has a couple coming up that they’ve announced which have potentially billions of transactions. That will be interesting and another big thing I expect to see happen.

Chris Tse: What is it? Commercial or financial?

Tom Trowbridge: Commercial, not financial. This hasn’t really happened yet.

Chris Tse: You need commercial. Many interesting financial vehicles like invoice discounting really depend on commercial to be the underpinning. It’s just a much more solid foundation to build.

Tom Trowbridge: Like putting all the coupons in the US on DLT. That’s in progress. But we haven’t seen that actually deployed at any scale yet. I expect something like that to happen next year. 300 billion coupons enables entirely new use cases. Those are a couple of things I can think of.

To get all our latest updates, sign up for our newsletter on cardstack.com, star Cardstack on GitHub, and join our Discord channel or our Telegram group and announcement channel.