Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Stefan Thomas: Understanding Ripple

Episode Date: August 17, 2015

In a long-overdue episode, we finally had the chance to dive into one of the most known but poorly understood cryptocurrency/blockchain project: Ripple. Reviled by many in the Bitcoin space, we put aside any prejudices and sat down for a fascinating conversation with Ripple Labs CTO Stefan Thomas. We talked about his early days in the Bitcoin space, how Ripple came about, what the Ripple network looks like today and how its consensus protocol works. Topics covered in this episode: ‘ early experiences in the Bitcoin space How Ripple was created The problem Ripple is trying to solve How the Ripple consensus algorithm works Whether Ripple is decentralized or not What the correspondent banking system is and how Ripple wants to disrupt it The recent FinCen fine and what it means for Ripple Episode links: Ripple Ripple Labs Ripple Labs Fined $700,000 by FinCen Richard Gendal Brown: The Deep Insight at Ripple's Cores Peter Todd Ripple Protocol Consensus Algorithm Review Ripple Consensus Protocol Whitepaper This episode is hosted by Brian Fabian Crain and Sébastien Couture. Show notes and listening options: epicenter.tv/092

This episode of Epicenter Bitcoin is brought you by Shapeshift.I.O. With no account or signup required, it's the easiest way to buy and sell gems, counterparty, dogecoin, dash, and other leading cryptocurrencies. Go to shapeshift.io to instantly convert your altcoins and to discover the future of cryptocurrency exchanges. Hi, welcome to Epicenter Bitcoin, the show it talks about the technologies, projects, and startups driving decentralization and the global. global cryptocurrency revolution. My name is Sebastian Kutjouye.

And my name is Brian Fabin Crane. We're heard today with Stefan Thomas. He's the CTO of Ripple Labs. And of course, pretty much anybody in the Bitcoin space has heard or if the cryptocurrency space has heard of Ripple and so have we many times. But we've never actually had that time to sort of really dive into that. So our knowledge, my knowledge has been very superficial. So really excited that we'll have a chance to change that today. Stefan also before he got involved in Ripple, he had sort of a long history in the Bitcoin space. I'm sure some of you will know the website he created. We use coins that was, I think for a long time, the sort of initial launch path to learn about Bitcoin in two minutes.

And he's also the original creator of Bitcoinjs, which is used as a JavaScript library, which is used for a lot of web wallets like blockchain. At least that's what it said on the Bitcoin J.S website. So, Stefan Thayer, thanks for coming on. Thanks for having me. I'm really excited to talk to you guys. And I hadn't realized that you were behind. We used coins.

Of course, first one on that website, whenever I first got into Bitcoin, about a year and a half ago. So, I mean, almost two years ago now. Awesome. Yeah. Yeah, so I guess that brings us to sort of our first topic, right? So you started really early on in Bitcoin. Can you tell us a little bit about that? Yeah.

Yeah, so I first got interested in Bitcoin in mid 2010. And at the time, there was already an established community, but things like the Bitcoin price were just brand new things. And you couldn't really, like there was a little bit of debate of what the price was and if it had value at all. And there's very little trading going on. Moncox was just starting out. And so it was a very, I don't know, it was clear there was some potential there.

What really attracted me to it was that it seemed to be doing something that in computer science I learned was impossible. You couldn't do consensus like that. And so that attracted me to it and I decided to get more involved and see how can I. How can I help? How did you get the idea to do we use coins? So people had, there was this one sub forum on Bitcoin talk, which was basically, you know, bounties or like projects that people wanted to sponsor. And one of the largest bounties with over 9,000, yeah, I know, very meme compatible, over

9,000 Bitcoins was the animated video bounty. It was 9,052 Bitcoins were actually paid out at the end. And I grew up with a friend of mine, Fabian, who actually also now works here at Ripple Labs with me. And he was finishing up his bachelor's degree in motion graphics. And so I called him up, we'd run up together. And I was like, you know, do you want to do this video together? And he was like, yeah, I need a bachelor project anyway.

And so we started collaborating on it. He ended up deciding to get paid in euros, not Bitcoin, which he still regrets to this day, I'm sure. Oh, my God. Well, you know, I could enforce him, you know. So yeah, we're, that's kind of how I got started, being more active in the community. I started going to conferences. I was at what I think most would call the first real Bitcoin conference in New York, organized

by Bruce Wagner back then. And yeah, I spoke there together with Gavin and Jeff Garzig. And yeah, just got more involved. Yeah, that sounds like that was a very profitable project to do that website. I don't regret anything. Yeah, so what about Bitcoin JS? When did you work on that? Yeah, so one of the lessons from we use coins was, you know, it was the website.

it was like you say the most popular Bitcoin website at the time aside from Bitcoin org and so a lot of people were coming in because they had seen the video on TV it was shown on a lot of news stations and so on so we got a lot of inbound traffic but very few people converted to actual Bitcoin users and I started looking into like why is that and my background is I worked at various web agencies so I know things like funnels and you know marketing one of the things that I noticed was that the funnel for us was just really horrible because you had to download the client and then a lot of people would drop off of the day downloaded and install the client.

At the time it took like three or four days to download the blockchain. And so it was a very high barrier to entry for people to actually get into it, especially if you weren't super technical. And so one thing that I thought would help a lot would be to do web wallets. But one thing I was concerned about is there was this website called Mybitcoin.com. And it was basically a web wallet that would hold your keys for you. There's several out today now. And they ended up running away with like 50% of their users' coins.

And so clearly that was not the right approach. So I wanted to create a library which would allow you to use Bitcoin in the browser, but in such a way that you didn't, like the website didn't have the keys. The keys were somehow stored on your local browser. And that's kind of always been the dream with Bitcoin. And that's how I got into writing Bitcoin software. Cool. And where did Ripple come in? When did you start working on Ripple? So I would say like there's a couple of things that attracted me to Ripple.

So actually hard to think of where I could start. One of the things that was happening sort of in 2012 was that I had worked on Bitcoin Jazz for over a year. I was starting to get kind of tired. It was very hard for one person to keep up with, you know, all the changes and all the new things that people were discovering in Bitcoin, all the feature requests. and so on. And I had always like this little race going on with my current where we would try to get our respective implementations like him working on Bitcoin jazz, me working Bitcoin Jay, we were both in the Swiss community, trying to get them to feature parity with Bitcoin. And it just, it got to a level where it was just like not sleeping, not eating, not drinking, and just very, very tired all the time, trying to maintain Bitcoin jazz and keep it up to date.

And to be honest, like the browser library got very popular, but there's also the big, most of the work was in the server component, which implemented like the script interpreter, mining, all this kind of stuff. And that was a lot of work and not a lot of people used it. And so I started to kind of reevaluate like what I was doing and like what, what, you know, what would actually help Bitcoin the most and what would help this technology the most. And one thing that I noticed was that I always had to live off of my Bitcoin investment. And I knew that a lot of other Bitcoin developers were in a similar situation where they actually didn't have a ton of Bitcoins.

You'd constantly hear stories of people who had like a huge stack of points and, you know, totally forgot about them. And then they come back and suddenly they're like millionaires. And I always thought that it was kind of weird of how the Bitcoin economy worked. Like all of the Bitcoins that get produced get largely put into electricity. So they go out to electricity companies ultimately and don't really benefit the community. And right at that time I got approached by Jed McCaleb, who was starting this company called OpenCoin. And I was looking at my dwindling investment in Bitcoin because I had to even live off of it. I was looking at, you know, I want to get a job, but I don't necessarily want to get a job and something that isn't cryptocurrency related.

And so first I told him no because it was a pre-mined coin. I didn't, wasn't interested in that. But all the things sort of came together in the end. came together in the end, Jesse Powell, who's a CEO of Cracken, now I think. He was a good friend of mine. He had worked with him before I trusted him. He invited me to come over to California

and meet the team of OpenCoyne and kind of hang out with them for a little bit. And the other person who was an investor at the time was Roger Bear. And obviously, I respect him a lot. He was a good friend. So I ended up coming out and I ended up spending some time with a team.

And that's when I noticed, oh, holy private, these guys are really smart. And if I've learned anything, it's a good idea to hang out with smart people because, you know, it rubs off. So I ended up staying and I ended up moving here. So that was really early on with Ripple, right? So OpenCoin, is that, did it just change the name to Ripple or was that a different project? So, yeah, it just changed the name. Originally, the company was called OpenCoyne.

The project was already called Ripple at the time. And it was actually, you know, we intentionally tried to keep it separate, but it just caused a lot of confusion. So eventually just changed the name of the company to Ripple Labs, which is what it's called today. And so how did this idea come to be, like the idea of Ripple with these people that were obviously working in Bitcoin at the time? So I would say that like a lot of the issues that we saw with mainstream adoption were things like people wanted to use their own currency. There were little companies that I talked to where I tried to get them. to accept Bitcoin or try to get them to deal in Bitcoin, but they were like, well, look, we're paying our suppliers in dollars.

We're paying our suppliers in Swiss francs or, you know, whatever the case may be. And they said, you know, we have to accept Swiss francs. And so the only option that you really have at that point was, you know, there were companies like BitPay out there who would convert it for people, right? But then you have like a single point, like a single company that's giving you the rate and that's converting your points for you, whether it be, you know, I don't want to single out bid pay, but great, I think, but like, you know, do you have the great job, I think, but like, You know, do you have the coin basis, the bit pays.

And we didn't like that idea that, you know, the entire currency exchange would be controlled by that one company. And so one of the ideas that we were pursuing early on was kind of this idea of a distributed exchange. So you would have like, you would just have an exchange network. And if you wanted to get from Bitcoins to Mexican pesos or whatever, you wouldn't have to go through, you know, a central exchange. It gives you one rate and you're stuck with that rate. but rather you could go through a competitive sort of exchange market that was on a distributed ledger. And it was kind of like the original idea, I think. And yeah, that was one of the main things.

Another thing was kind of mining. So Jet had a big thing about mining being wasteful. I recently looked at some statistics that Bitcoin, when it was at $1,000 was roughly putting out something like 8 million tons of CO2. that's more than the country of Cyprus. So there's definitely something to be said about mining as being a wasteful process. And I think we owe it to the world, I guess, to try and find other options if they exist. So you said the original idea was to have a kind of decentralized exchange so you could trade like Paisers and different currencies. Of course, the question then is, well, how?

do you have a peso on a ledger or a fiat currency on a ledger? Is that sort of the next step that then came and it was like, oh, we need to do sort of a blockchain like assets for fiat currencies? Yeah, so you definitely still need a custodian. The goal there was to make the actual exchange rates competitive so that if you wanted to trade out of a currency, you could always do that. didn't depend on an exchange being online, exchanges being DOS was a big thing at the time.

So again, that was kind of like what got the ball rolling. The other idea that came into it was Ryan Fuger's original ripple concept, which he actually allowed us to use the name, which was basically this credit network. So you could have different people who could owe each other money and then move that money around. And so all of these things came together, And what Ripple then became was essentially a graph that describes everyone's financial

or credit relationships to each other. And from there, once you have this very general language to describe relationships, we could allow people to use it however you wanted. And the most popular model of usage that emerged was kind of what we now call the gateways, which is basically you have someone who's the custodian who's holding the funds. Then you have a bunch of people who are market makers who are trading. funds against other assets, could be XRP, it could be another gateway's assets. And then you have the users who are kind of using that network.

And that's kind of what it evolved into, circa 2013, early 2013. It's interesting because in the Bitcoin space, right, one of these sort of conceptions about Whipple is, oh, it's this horrible centralized thing. But actually the original idea was that, oh, we want to do something more centralized than what we are seeing happening in Bitcoin with these payment processors. Well, the way I look at it, it's like, you know, I don't mind people criticizing us as being centralized. Like I, you know, our official company position is we're not centralized because we can't actually change the rules and that's true.

Like if we tried to do anything that people really didn't like, they would just not longer trust our validators, we would start trusting other ones. But to me, the real answer is, you know, that criticism right now is totally valid. I don't mind that criticism. It's just one day we will decentralize. We will add more validators as soon as there are enough good ones out there that we can recommend and so on. And people will start trusting other validators and the network will become decentralized. Just like when we originally started, we were closed source and people said that it'll never open source.

It's always going to be closed source. And then we open source, right? It's just like we do the work and we don't mind being hated in the meantime. You know, like hate us all you want. At the end of the day, it's going to be decentralized. then the facts are going to speak for themselves. So that is fascinating, right? So the idea is, oh, you are centralized today

because that gives you, I guess, a control over the process of development, right? Because that is currently quite a nightmare in Bitcoin, right? Like how do we reach a decision on what features, what choices need to be made? So obviously you guys don't have that problem because, well, you guys just make the choices that you find best.

It's not quite that easy. We do have a community to worry about. So I think every choice that we make is kind of based on what would people want us to do. And this is kind of like if you're this early and at this early of a stage, you don't have a ton of freedom. That's I think what we mean when we say that we're already decentralized. It's like, you know, our hands are pretty much tight. Like between what the government was and what the community wants, we only have very small area to maneuver. but we use that to kind of make sure that the core tenets of ripple are still there.

Like, it's an open system, anyone can use it, that kind of thing. Today's show is brought you by your friends at ShapeShift. Of course, Shapeshift is the fast and easy way to trade altcoins, and they now support over 45 of the most popular altcoins, including Counterparty, Ripple, gems, and so many others that I've never even heard of. And they now support, you guess, that Ether, since the release of the Frontier launch, ShapeShift, that's added ether to the list of supported currencies.

And so now you can buy ether and start building some sort of contracts on Ethereum. Now, there is a small caveat, and this is true for every exchange at the moment, is that it can take up for six hours for you to get your transaction confirmed. But just stay patient and you'll get some ether and you'll start building some really cool smart contracts very soon. And just as if ether wasn't enough, they're also doing this cool thing with a gaming company called Spells of Genesis that's putting some of these in-game assets on the blockchain. And one of those, they're selling in a very limited amount, like 25 every week, only on ShapeShift. So that's called The Wander and it's called Spells of Genesis.

So if you're a gamer, if you're interested in that, check it out. Because, of course, it's a very interesting way to make this sort of fake scarce asset actually scarce by using the blockchain. So head over the Shapeshift. I'll give it a try. We'd like to thank Shapeshift for their support. of Epicenter Bitcoin. So let's get into some of the technical components of Ripple. So there are multiple terminologies here that some people might find confusing.

In fact, I do. So you have the Ripple protocol and consensus algorithm. And then you have the Ripple network. And then there's also the company Ripple Labs. So perhaps we can go through these in order. Let's start with the protocol. Can you describe the different components of the protocol? So usually we break down the protocol into a number of layers.

So there's a physical layer, which is essentially a gossip network, where different nodes exchange information about what the ledger is, where the network is at, what people's opinions are, etc. The next layer up is what we call the validation layer. In the validation layer, it's basically a touring machine that, it's like a database where people apply transactions in order to move forward in time and in order to mutate that state. And that, again, gives you a touring machine that you can pretty much implement whatever you want on.

And then we have another layer on top, which is essentially what we call the application layer, which is basically the actual application that's implemented on this touring machine today, which is kind of this financial network with things like trust lines and XRP and all that kind of stuff. There are some interdependencies between the layers. Like, for example, XRP is also used on the layer below as a way to avoid spam or, like, limit spam. But other than that, they are pretty independent. So, like, you could take the bottom two layers of the Rubell architecture and build, like, a completely different system on top. So I read the web paper, and I wasn't aware of this, but there are different consensus algorithms that solve the Byzantine general's problem.

Can you describe how the ripple consensus algorithm is different than, say, Bitcoin's algorithm? Yeah, so I would say that if you look at the traditional consensus algorithms that actually, like a lot of them, predate Bitcoin, the way that they generally work is that they will set, they will take a group of nodes and you know what that group is, and then they'll set a quorum, like a number of notes that will make a statement. And if you have enough statements from enough notes, you know that there will not be another opinion that has as much support as long as your assumption, which is that only a certain number of notes have failed, usually called the fault tolerance F. You know that then you're safe, right? And so people have proven algorithms such that they provide safety, which means that the network can't fork, and liveliness,

which means that the network will eventually make progress under these assumptions. Like you only have so many failures and you have a network where messages get delivered after some unknown amount of time. So these are some, I think we've talked about some of these on the show before, like, is like Paxos one of them or PBFT? Right, so Paxos would be a consensus algorithm that's not Byzantine fault tolerant. PbFT is probably the most well-known one that is. And yeah, one of the biggest differences is that if you want Byzantine fault tolerance, you just need to add more nodes. So the number of nodes you need for a non-B Byzantine fault tolerance system is 2F plus 1.

So you need if you want to tolerate one failure, you need 2 times 1 plus 1, which is 3. Whereas in a Byzantine fault tolerance system, if you want to tolerate one failure, you need 3F plus 1. So 3 times 1 plus 1, which is 4. So those are just like your constraints in terms of like how many nodes you need in order to gain a stress. type of tolerance. Now, Bitcoin is kind of interesting because it breaks that model a little bit, or it breaks that mold a little bit, because it doesn't provide safety theoretically, right, because the Bitcoin network can fork in different directions. It's just that we always pick what to pick a winner. We just pick the longest blockchain to, like, come back to. And so from a

consensus perspective, Bitcoin doesn't provide safety because you never know for sure that isn't another longer chain out there. But in order, but by, by my mind, making this trade off, it gets a very nice property, which is the fact that you only need a simple majority, depending on, you know, yeah, you only need a simple majority to decide what the official chain is. Okay, cool. So let's dive into, let's dive into some of these. A lot of questions. So first of all, blockchain. Is Ripple a blockchain or not? I'm not sure if this question makes a lot of sense. Yeah, yeah.

So I would say that, first of all, the term blockchain, like, you know, we're just in the process of defining it, really, right? I mean, blockchain didn't exist until recently, so I think it's up to anyone to define what they mean by it. We usually call Ripple a distributed ledger, but it's still a chain of cryptographic blocks that refers back to the previous one. So I wouldn't blame you for calling it a blockchain.

I think one reason to use a different term is we kind of talked about sort of the Paxus and consensus history in computer science. We talked about Bitcoin as a different approach. Ripper was probably closer to the traditional consensus algorithms. The main thing that it does differently is rather than having a known set of nodes up front that is fully connected. One way to model PBFT is basically think of a graph of nodes and every node listens to every other node, right? And so you basically have a fully connected directed graph, right? And so Ripple would

basically be a consensus algorithm that uses similar principles, except it allows the graph to not be fully connected. You have to start thinking about what if this note doesn't use the same set of nodes as everyone else, right? And so somewhere between the fully connected graph, which we know is proven safe, and the completely disconnected graph, which we know is going to fork, there's going to be a point where you have a certain amount of tolerance and that amount of fault tolerance is going to be decreasing as you remove connections. And so what we're kind of interested in is sort of this middle ground. And the reason that's interesting is because the assumption that everyone's going to set their

same, the same nodeless is just not a very strong assumption, right? Like if you have a validator, let's say, in the U.S. and you have a validator, let's say, in Iran, there's a good chance that the U.S. validator won't ever trust the Iranian validator, right? They might be forced not to trust them. And so we still want to be able to evaluate the topology of the network and say, and say, this network is still safe. This network is still live, even with the disconnections. Okay, so we'll get to this node list thing in just a second.

But can you first just walk us through a transaction? So someone creates a transaction, signs it submits it to the network, then what happens? So when you sign and submit a transaction to the network, it would first come in on the gossip network layer and be spread around the network. If it's valid and it's paying enough of a fee and the note that you originally sent it to is confident that it will be included in the ledger. After it's spread through the network,

it will be a part of the next proposed set from a lot of nodes. So a lot of notes that have seen this transaction, they think it's valid. They're going to say, we propose that this should be in the next ledger. Any transaction that's uncontroversial... So is the proposed set...

Is that kind of like the mempool in Bitcoin? I would say it's like it's one step further because like it first goes into something very similar to the MAMPool which is kind of like all the transactions that I know about

right and then the proposal is more about like what I think is going to go into the next ledger it's like I actually think this is going in so who sets these criteria

who sets the criteria for what is a valid transaction to be a next proposal so any transaction that pays enough of a fee is syntactically valid. There's no conflicting transactions that I already think are going in.

Those transactions would be considered likely to go into that next. So the rules are the same for everyone? Yeah. As long as you run the same software, obviously. I mean, you can't stop a single validator from running different software, but that would just be like a faulty validator, essentially. Okay. And so then it gets sent into, to be a proposed transaction in the next page of the ledger?

Yeah, so the next step would be every validator receives proposals from other validators. And so now you can see what actually people think. And so once you know what people think, then you can say, I actually think that this is the ledger that we should construct because these transactions have an actual majority based on the proposals that I've received. And so once that's agreed upon, that ledger will be published, the validation will be published. And then the final step is every validator will be listening from other validators, published ledgers, and if they see enough of those published ledgers, they will consider those ledgers validated, and they will say, okay, this is now the state of the network and we proceed from here.

So that's what the point where, like, for example, a gateway would process a deposit or something like that. Okay, so where does, for example, with Bitcoin, we have proof of work providing some security what protects the ripple network from someone, you know, creating all the validator and doing something funny? So if you think about proof of work, like what proof of work essentially says is, well, let's actually start a little bit earlier. So let's think about what is the actual problem we're trying to solve. The problem we're trying to solve is the double spend problem. We're trying to make it so that if someone wants to spend their coins, they can only do it once. They can't later change their minds and say, like, no, actually, I wanted to send it to and stuff like that.

So in order to solve that problem, you need to have some sort of authoritative decision. It doesn't necessarily matter what the decision is, only that someone is making it, and then after it's been made, it can't be changed. And so if you imagine a peer-to-peer network, if everyone was always connected and always online, everyone could just watch the transactions that happen and make their own decision, and you wouldn't need any validation or mining or anything. You would just see which transaction came in first, and then everyone would see that one first. I'm making some assumptions here. And you would just say, okay, this one came in first, so we'll process this one, and we won't process the other one. And you could make a rule, like if they come in very close to each other, then they'll both invalid or something like that.

And so people wouldn't double spend because they've just been destroying the coins. Now there's one big problem with this, and that problem is that people, you know, they join the network later, right? They go online, they go offline, and they need to somehow re-synchronize. How do you re-synchronize? Because when you connect to the network, all you see is a bunch of nodes out there and you don't know who to listen to it. They're going to say, well, this is the state of the network. Someone else is going to say this state of the network.

And you don't know if the majority is just a bunch of, it's just a botnet that's like trying to convince you of an invalid state. And so what mining allows you to do is it allows you to say, like, well, this state has this much mining power behind it and this state has this much mining power behind it. So essentially, you're not watching the network all the time, but you're trusting that there There are other people out there that are watching the network all the time and they're also mining. And so whatever they say, you can kind of rely on.

Now, mining or the proof of work portion essentially acts as a proxy for voting rights. So you're giving people voting rights based on how much they mine. So if you have 70% of the mining power, you have 70% of the voting rights. So what Ripple does by contrast is rather than deciding who gets to do that you get to the money, to vote based on mining power, we just tell each individual user essentially or each individual node that connects to the network, you decide who you want to have voting. So like if you want to listen to these notes, just put them in your list and then your software will adopt whatever those notes say.

If you want to listen to those notes, put them in your list, right? And so you're kind of like turning, you're making it the notes problem or the notes decision of who do you want to listen to. And it gives you some pros and cons. So on the pro side, it allows you to deal with malicious nodes. So for example, if there was a malicious Bitcoin miner that would like never include any transactions, for example, there's nothing you can do. It would just be like, okay, well, those blocks are like empty.

It wouldn't hurt the Bitcoin network a great deal, but you couldn't all, you also couldn't get rid of them. You couldn't shut them down or anything. Whereas like in Ripple, what happens is if someone doesn't include any transactions or they always vote, no, we would consider them a faulty validator. So people would take them off the list of validators that they listen to. And so their voting power in the network would go down. And so you have kind of like the ability.

So is there on a client level locally? Are there some rules that determine who is a faulty validator? There's no rules. It's completely manual, right? So like the note, each note has to decide who they want to listen to. Okay. So if I was running a ripple node and I noticed, so this note is always saying node and I would like manually

go in there and remove them. Yeah. It's actually kind of, I don't know how deep you want me to go into this topic, but it's actually kind of interesting how ultimately the consensus is based on the

code that you're running. For example, let's say the Bitcoin developers decided to change the mining algorithm to another mining algorithm. For whatever reason, let's say Shaw-256 is broken or something. Then the blockchain that previously

would have been considered formally valid, the one that has the most work under shot 56, would suddenly no longer be valid. It would be a different blockchain that would be valid. And so, ultimately, the trust always lies with the developers. The trust is always a manual process, right? Because the developers define the rules. And then all that mining is a shortcut, a decision-making framework that's faster than the manual changes. In order to get everyone to upgrade their Bitcoin software, take a long time.

So you need something that can react on the order of minutes. And so that's why you use mining. And so with Ripple, it's similar. It's like there's the developers to make the software, and they decide certain ground rules. And then you have the people run the validators, and their job is to make faster decisions. And so you're kind of like trying to get to a point

where you can make faster and faster decisions that are less and less secure, right? So Peter Tau wrote an interesting paper analyzing Ripple, And I think he brought up a point that I'd like to address here. And that's the question, you know, can this be really decentralized, right? Because if you, let's say I'm running a ripple node, then what I don't want to do is I don't want to listen to nodes that, you know, do the wrong thing, right? I don't want to listen to notes that maybe give me fake data or malicious because then I'm at risk of losing money of being defrauded or something, right? So doesn't that just mean that there is a really strong, the sort of Nash equilibrium of this is everyone listening to the same set of notes and then you have this sort of centralization where you don't want to switch away, right?

And I guess that gets even more, you know, that gets even stronger because you guys have a default, a list of a default node with the clients, right? So I presume most people just download it and stick with default nodes. So how is that ever going to be decentralized? So it depends on what people are actually trying to optimize for. Like if you're, if decentralization gives you some major benefit, then you are going to choose a list of notes that is that is decentralized, you know? And whether you call that ripple or you call that some fork of ripple, right? Like you would choose whatever network has the properties that you're looking for. So if you think that a decentralized network is better than a centralized one,

you would choose a list of validators that also buy into that. The other thing that will influence your decision is like the people that you're trading with, what have they decided? What network are they on? Because you want to be on the same network in order to be able to interact with them. The other point you mentioned was the fact that there's a default list. We don't think that the defaults will be static in, you know, one year or two years. We think that we'll build, or we are already building tools that allow you to build a list based on validator uptime, again, what your trading partners, whose validators they are listening to, and so on.

So very soon we'll have the kinds of tools that allow you to build your own list and put together your own list. Still though, it seems to me that for this to be truly achieve a high degree of decentralization, right, you'd need to have, first of all, you need to have a fair number of different nodes, right? Maybe it doesn't have to be huge, but probably a hundred or more. And then some... I don't know if you need 100 notes to be decentralized. You need enough notes so that collusion becomes impractical. And I think that that number is probably a lot lower than 100 nodes.

Like, would you consider Bitcoin decentralized? Bitcoin right now is controlled by roughly four nodes. Yeah, no. I mean, I think Bitcoin certainly has a problem with being too centralized. And I don't think that's a good thing for Bitcoin at all, right? I think Bitcoin should be more decentralized. I don't know. I think it's, it's, you have to think about like what are you trying to get out of the decentralization, right?

And something isn't like, decentralization is not a goal in of itself. The goal is security. The goal is access. The goal is, you know, the rules come first and no one can, no one's above the rules. Those kinds of ideals, I think, are what we're trying to achieve. And I think that people will be attracted to whatever networks provide them with those benefits. And I think that with this, like, approach of, like, manually choosing the validators, what we're creating is a system where it will just end up with whatever people prefer. because we will try to provide the kind of network that people want and then people will opt into the kind of network that they want. So even if we screw up and we don't provide the network that people want,

people will just go onto a different network that does have what they want. It's just important to understand that what people want isn't maximum decentralization. It's not that you want to be on a network that has like a billion nodes and the work is replicated a billion times. You want to be on a network that is secure, that gives you access, you have confidence about its continuity, those sorts of things. So you always have to come back to what are you actually trying to get out of the decentralization. And this is why people hate you on Reddit, my friend.

This type of talk, man. You got to. So is there an incentive to run a node then? Is it purely just a contributed network or do nodes get paid in some way? So you already have to run a node if you want to do anything interesting with Ripple. Like if you want to listen to transactions and so on, you have to connect to some node. And so you can either run one or you can connect to someone else's note. And of course, they might charge you for that or they might restrict your, how many things you can submit and so on, how many queries you can do.

And so there's already a strong incentive to run a note if you're a participant. And then whether or not you publish validations is literally just a flag. It's just a switch to you flip. And so we, what we observe in practice, so a lot of the businesses that are on Ripple, they also publish. validations and we're starting to track how good their validators are in order to be able to start to decentralize. And then also, or in order to be able to start to add them to the recommended node list. And so it's, I don't think there needs to be much of an incentive.

And one nice thing about not having an incentive is that you avoid a lot of the attacks that are based on trying to get the incentive. So like if you think of selfish mining, the things that are issues with Bitcoin, those all come back to the fact that you get incentivized for mining. And so if you don't get incentivized for validating, there's actually, there's some benefits to that. And so right now this is the approach that we're taking. But obviously, you know, this could change if our assumptions turn out to be wrong.

Today's magic word is Waves, W-A-V-E-S. Head over to let's talkbidcoin.com to sign in, enter the magic word, and claim you're part of a listener award. So we've talked about decentralization from sort of a node level, but another, I mean, of course, you can talk about decentralization from all sorts of perspectives. One is also the software you run, right, as you mentioned, which is, of course, super important. So with Ripple, is it correct, that there's only one implementation? And do you think this is a problem? Would you like to see other people develop other Ripple implementations?

Okay, so the answer is going to get a little technical, but one thing that was one of the innovations in Ripple originally is that validators don't just validate the history of transactions like they do in Bitcoin, but they also validate the current state of the network. And what that allows you to do is it allows you to build lightweight clients that don't trust the notes that they're talking through. They can literally see what their balances. They can see what their state of their trust lines is, et cetera. and they can cryptographically verify that against the top hash of the current ledger. What that allows you to do is you can be a lightweight node, and as long as you have your own trusted node list,

you know exactly what those trusted nodes are saying, and so you don't have to trust anyone else along the chain. And the reason that that's important is that there are several implementations of those client libraries. So like if you have RippleLLIP, there's a Ripple of Version in JavaScript, there's a Ripple of Version in Java, there's obviously the Ripple D implementation, C++. So there are a couple of different implementations of that logic. Now, there's still some things missing with the validation logic there,

but in principle you could add that to those implementations if you care enough about that. On the validation layer, it's true. There's only one ripple devalitation implementation that validates actually run. But that's the same true as with Bitcoin. So I think something like 99% of the mining power is actually the Satoshi Bitcoin original implementation. And that's a problem. Like that's a problem for both networks, I think. And one of the things that we think is going to have to have

to have long term is that there have to be multiple implementations. But it's very hard to do that. And like one of the reasons that people don't run alternative implementations in Bitcoin is because you risk that if there's any difference in that implementation, suddenly you're on a side chain, right? So let's move on to, I guess a higher level above the consensus protocol and the algorithm and some of the technical components that are at the core of Ripple, which are gateways.

So we've mentioned them early on in the show. Can you explain exactly what a gateway is? Because it seems to me from my understanding of it, that it's very similar to a Bitcoin exchange, but I understand there are some differences there that are kind of important. So I kind of mentioned gateways as like we were talking about the early days of Ripple, kind of where we started out. One thing that that has evolved into is like, well, who are the? best gateways. They are banks and even better the central bank, right? So the bank,

central bank is like the ideal gateway, right? Because they can actually issue real dollars, they can actually issue real euros and so on. And so as we've evolved as a company, our focus has shifted away from people who are non-banks or non-banking financial institutions over to banks, credit unions, corporates and so on, who have the kind of scope in order to be able to issue in order to issue real dollars and FDIC and shoot dollars and those kinds of things onto the network. And that's right now our current focus. And one thing we found is that for them, surprisingly, I think, you know, we get ripped off

by the banks, but obviously the banks get ripped off by other banks. So we actually found that they have a very, very compelling value proposition adopting this technology, especially smaller banks do, in order to be able to do a real-time internet. transactions. And so our focus has pretty much shifted to financial institutions now. And we just found that it's a slam-dunk story. They really like this. You know, they want to, they like the certainty that they get from being able to see where the transactions are at. They like competitive effects. They like the speed of the network. And so that's been our focus more recently. The issue with the with the community gateways was always the licensing, right?

It was very hard. It's very hard to operate a community gateway. And it's very hard to do so. in a profitable way. Okay, so that was interesting. I don't know if you want to dive into the application a little bit more right now, but perhaps this would be a good idea. So you mentioned, so banks use this for international transfer. So can you talk a little bit more about how exactly that works? Sure, sure.

So any bank can just issue, for example, ripple euros, And then why would I accept that as another bank? And why is that better? So the way that international transfers between banks work today is through the system called the correspondent banking system. And it's kind of this disjointed hodgepodge network of different banks having correspondence relationships with each other.

So if you go to a small, mid-sized bank, and you tell them, do you want to make an international wire transfer, what will ultimately happen is that they will send the money to a larger bank in their country. There's only really a few banks internationally that have those international correspondence relationships. That correspondent will talk to its correspondent in the other country and then that correspondent will forward it to the final receiving bank. So that's kind of the usual flow. You'll go through at least two intermediary banks before you get to the destination.

There's very little transparency with that. So if you make a swift transfer, we've had, like, I interviewed one guy who, who was an entrepreneur who would start a company in Singapore, it sent the founding capital over to Singapore, and it just never arrived there. And then you went to his bank and it's like, what's up, where do my money go? They're like, well, we send it up to our correspondent.

If you want to know what happened, we can investigate for you, but that'll be $40. And that's just, you know, adding insult to injury. Banks realize this, and especially smaller banks, would love to provide a better service to their customers. It's just that, you know, they get the product that they get from their correspondent, and no one's really in a position

to fundamentally improve the system. And so when we were looking around the market in 2013, we kind of noticed this huge opportunity, a huge problem with with correspondent banking, and noticed that Ripple was like the perfect solution, right? You could just, as a bank, you become an issue on the network. You issue dollars if you're a US dollar bank, if you're a bank in Europe, you might issue euros.

And then people can come in, market makers can come in, which could be either individual traders or it could be hedge funds, it could be investment banks and so on, who have very competitive of X-rates and they can exchange between the dollar assets from one bank and the euro assets from another bank. And so that way you create this sort of global network where banks can send money to each other directly. So on the one hand, you have this disjointed network, which is the international banking sector where it's very complex to send money from one country into another.

And on the other hand, so what Ripple is proposing is a system where essentially Ripple acts as the channel where banks can send money between each other. But who, so for example, if I want to buy some Ripple Euros, who created that asset? Can anybody just create an asset called Ripple Euros? And is there only one? Are there multiple ones? Can you explain that? So when we talk about assets on the Ripple network, usually the way we express them is not Ripple euros, but, you know, know, for example, five-dollar euros, right? So if you have a bank that is issuing an asset

onto the network, it is that asset issued by that entity, and that forms an asset on the chain. And so people create markets between different assets. Now, if you have a bunch of people issuing euros, since there is the SEPA network that allows you to settle between these different assets, there's arbitrators who just, they almost trade these things one-to-one, because they know that they, as soon as they've traded them one way, they can just send it back through the SEPA network and then rebalance their holding. So what ends up happening is that all of the different euros

end up being worth the same. And so on the network, you can send them around at basically zero cost. But at the same time, the risk is still localized to that particular bank or that particular issue. So if I'm, how may I say to reface that? So let's say I wanted to send some money to you in the US, right from Germany. I want to send some euro to you. Then, you know, before it was like I tell my bank to send it, they send it to another bank,

that sends it to your correspondent bank and they charge us a lot of money. It takes a lot of time. Whereas with Ripple, my bank creates those euros, the ripple euros. And then the idea is that somebody, has is arbitraging that and then buy that euros and maybe

sells some ripple USD that then sends it to your bank and that is a lot cheaper than going through these correspondent banks and faster is that? So I mean a part of that is set up right like the bank issuing assets that

happens ahead of time people proposing orders to trade these different assets for one another that's something that happens ahead of time. When someone actually makes a transaction, it goes essentially straight through from one bank to the other. And within Ripple, there is a path-binding algorithm, which basically finds like the cheapest set of conversions that you could go through. So if it's direct, it'll just go direct. If there's an orderbook between those two exact assets. If there are two less liquid assets, it might go through some more liquid asset, like, you know, dollars

or Bitcoin or whatever, or XP, and then get to the... to the receiving asset. So what role does XRP play in all this? So XRP has two main roles. One is it acts as a fee so that we can, we have some neutral way of charging people for making transactions. So one of the issues, if we used any other asset, you know, all other assets have some issues.

So that issuer can create as much of that asset as they want. So we can't use it as a fee. And so one reason that there's a built-in crypto. currency is so that that's a neutral fee. That's a fee that we know is worth something. No one can create any. And so no one, including ourselves, can spam or do the S network just by creating more and more XRP to spam. The other main reason or the other main objective for XRP is to make it a liquidity asset. So it's basically, I kind of mentioned these different paths. So we specifically encourage market makers to make paths through XRP in order to create a

a liquid market through XRP. And one of the reasons that that's attractive, one of the reasons that we want that to happen, of course, aside from it benefits us, it's how we fund the company. It's also something that, like, if there was a neutral cryptocurrency out there that a lot of liquidity went through,

we would live in a world with just much lower FX rates because anyone can just connect to that cryptocurrency as opposed to having to go through dollars and it being at the whim of whoever issued those dollars. I mean, that's kind of what happened with Bitcoin and the alt-coin market, right? I mean, most of these all coins don't have any liquidity between each other, but only versus Bitcoin. Yep, exactly. Okay, no, so that makes sense.

So I think that brings us precisely to Ripple Labs as a company. So is a business model that you guys own a lot of that XRP? Because, you know, it was like pre-mined, so you had that in the beginning. So now you sort of try to make the network valuable and then hope that the XRP will appreciate and value. And that's the kind of the path to obtain wealth. So the thing with XRP is that we're probably not going to get rich with it. The issue is that just like me, you know, working on Bitcoin Jass back in the day, I had a huge stash of Bitcoins. But if you live off of it, that Stash tends to get, you know, smaller over.

time and you tend to have to sell a lot before it appreciates a lot. And so the same same basic thing is happening with with Ripple. I mean we're the number two cryptocurrency now behind Bitcoin. But even so, like we are already we already spent a good chunk of our XRP and gave a good chunk away. So it's it's foreseeable that after I don't know another 10, 15 years or so, we may run out or we may run very low on XRP. And so we are already looking for other ways to raise revenue.

And one of the things that's very attractive right now is just that, you know, banks want to pay us to help them integrate because they want to have a service level agreement. They want to have guarantees. They want to have someone they can call if anything goes wrong. And so it's actually we found that banks are more willing to integrate with Ripple if you charge them for it. Right. So those are new revenue streams that are opening up for us and that allows us to to kind of transition off XRP and make that more of an asset that's independent of us and where we are just one of many, many holders as opposed to like an important holder.

No, that makes sense. And I mean, I think that having that several subtle agreement for banks is something that will probably be desirable for for corporations now. But with regards to use cases, are the use cases, are the use? Use cases strictly limited to banks and large institutions and large corporations using this? Or can anybody use, like, are there use cases for consumers, basically, or small and medium-sized businesses? So I'd say the enterprise use cases are the most interesting for us right now. As you know, there are many cryptocurrency companies that are going after consumer use cases. And, you know, I wish them the best of luck.

If we are very successful and we grow and we have the bandwidth, we're definitely going to go after other things. So we're going to, you know, whatever we think is the most attractive thing to build next. Right now, the most attractive thing for us to build based on how we're positioned, what our strengths are, the way that Google works, it's just banks are our most attractive customer right now. But I'm not saying that that will always be the case. So one potential competitor one could imagine to what you guys are doing. So Ripple makes sense to me, right? If I want to have USD on a blockchain, right, then what you guys are doing is sort of

the one obvious way to do it. But I know a lot of people are also trying to find a way to create this sort of synthetic assets that, you know, like track the value of a dollar without being backed by someone. Is that something you worry about and do you think if that does actually work out, that would be a strong alternative to the Ripple approach? So I would say that probably the, obviously the best kind of dollar is a central bank issued dollar. So if the Federal Reserve adopts one protocol or another, that would be the non-past ultra. It would be the best for the banking customers. It would be the best for end users.

And so that's, we feel in a position to pursue that. Now, obviously, you know, that would be a long-term project, but it's one of the things that I think is going to have to happen for cryptocurrency to really go mainstream. As for these other approaches, I think fundamentally, if you want to hear my personal opinion, this is not necessarily the company's opinion, but my personal opinion,

I think that any asset that is not fully backed by what it says on the can, is going to have trouble because people can't call that bluff. It's kind of like a bank doing fractional reserve. As soon as people lose confidence in that asset, there could be a run on the bank, it could be a run on that asset. And then, yeah, the fact that there isn't enough of that asset backing it will be exposed. So I personally think that the best way of doing it is having,

if you want to have dollars, you can have dollars that are backed by one particular institution, then the risk is localized, you know exactly what that is. that risk is. And if you want to make a synthetic asset on top of those, you can do that. And you can do that on Ripple today if you want. You can create an account that has trust lines with different gateways and issues its own dollars that are backed by those different gateways. So that's something that's that's super cool. So you can you can sort of diversify the risk by by issuing, you know, Ripple dollars that are backed by a whole variety of of different issuers.

Yeah, if you want to do that, you can do that. I mean, right now, if I'm totally honest, like, we don't have the diversity of live gateways that you could, like, seriously do that. Also, I think that, honestly, for most people, the risk that their bank will collapse, and if it's FDIC insured, you know, FDIC will not pay out is pretty negligible. So I think, I'm not sure how much of a benefit there actually is in doing that. But yeah, you totally could. And I think with Ethereum coming online, like other people,

our contract solutions being out there. Yeah, you can do a lot of cool stuff with these sort of cryptographic assets. So there's another story we wanted to talk about recently. There were some news that Rippel got fined by FinCEN. I think that the fine itself was not actually that interesting. I mean, if anything, it was sort of worrying in terms of how aggressively they seem to go after what seemed to me a fairly minor thing. which was, I guess you guys didn't do like proper KYC on when you sold some XRP.

But what was more interesting was that it was also mentioned that you guys would have to start doing some transaction monitoring across the Ripple network. What's going on with that? So there were several things that they asked for, obviously. And so, you know, as part of the negotiation, one of the things that was a big goal for us was to make sure that, We weren't forced to change the protocol in any way or triple the protocol. So we achieved that goal. Most of the things that we are now required to do are related to our hosted wallet, which is purple trade. So for example, we have to K-YCR users, we have to file suspicious activity reports and so on.

And other than that, we're only monitoring some of the public data. So the purple is a public ledger just like Bitcoin. So there's a lot of data that's available just to the public. And so we are building monitoring tools and tools that can analyze that data for on behalf of regulators. Cool. So how is the traction of Ripple? Are you guys making a lot of headway with setting up banks like that and setting up that kind of network? So obviously, I can't talk about any partnerships that aren't announced yet.

In terms of what we've announced, we've announced several banking partnerships. with FIDOR, CBW, Cross River. There are other networks that are exploring Ripple. That would be Western Union, Northport. And so we are working with a lot of different companies behind the scenes. One thing that's always an issue is if you want the stigma of cryptocurrency, kind of people still associated with risk and it's very new. And so they only dip in a toe, one toe at a time.

So a lot of these partnerships right now are in a proof of concept status. And as long as those proof of concepts go well, we have a very robust pipeline. But is the kind of outcome that you guys need to really sign on a huge number of banks so that it really sort of becomes this universally used system to move value around? Or is this also something that can work on a smaller scale and in some niche markets? It's interesting. So there's actually a couple things I would say that. One is there are a lot of countries with efficient local systems.

I think the EU area is a great example with SEPA. So within those systems, there's already a lot of progress being made towards real-time systems and very efficient low-cost systems. And so if you have one institution in that area, you can pretty much debit and credit any account in that area. And so our initial goal for the next few years is only to get very robust, solid banks in all of the key payments areas, right? So all of the key countries and payments areas around the world.

If we achieve that, then we can pretty much already serve most payments and move money internationally, especially if we also build all liquidity between those corridors. The other thing I would say to that is that Ripple is just one network. There are many others out there, and I think also in the Bitcoin community, just a lot of interest in building networks that were just kind of like FIRAP-style Fiat issuance and that kind of thing. So I think that one thing that we'll try to work towards is to tie those together with our network as well so that it becomes more of this like, you know, global crypto currency marketplace as opposed to,

as opposed to like different networks competing with each other. And in order to achieve that, we're participating in the W3C, web payments interest group, which is kind of right now it's looking more at retail payments. And later this year, we're going to try to start a community group around web settlement, which is essentially that idea that I just mentioned. Can you talk about that, the W3C, W3C payment group? Yeah, so the web payments group, it started as a community group several years ago. It recently, I think it's about a year now, turned into an interest group.

I won't go into the difference. Basically, an interest group is a group of industry participants. Right now, it's a few banks. It's Walmart, Apple, Google, a number of tech companies who are looking into better ways of doing payments on the web. And within that, since, again, there are some banks in there and just ourselves in there, there are a couple of people who are interested in web settlement as well,

which is basically, like, after you've made the payment, does it go through the credit card networks or does it go through another network? And so right now the idea is to add Bitcoin as a scheme to this, but we're also interested in adding a new scheme that can go through different networks. And that's kind of what we're working on right now. And hopefully we'll have more to show by T-PAC, which is the next major W3C meeting in October. Excellent.

Well, Stefan, thanks so much for coming on. It was really interesting, really enjoyed this, and enjoyed learning more about Ripple. Awesome, yeah, I love talking to you guys. And, yeah, if you ever have me on again, I'd be happy to. Yeah, no, no, I think that would be cool. There's some stuff we didn't get to talk to, talk about, especially code years and some things in this direction. But, yeah, maybe we'll do that another time. I definitely feel now I'm more confident now when people ask me what I think about Ripple to at least be able to explain it a little bit.

Yeah, I also feel my level of knowledge is clearly increased in the last hour, so I'm very grateful for that. There you go. So actually, let me ask you one last question just before we wrap up. So, you know, you had such a long history in Bitcoin and now you're in Ripple. And it seems like in the Bitcoin community, there's a lot of negative views about all kinds of projects that aren't Bitcoin. whether I'm here in the office with the Ethereum guys, you know, they definitely feel that, you know, and I just joined Ares and I'm sure there's some of things there. And so it's and of course, Ripple, you guys are perhaps the favorite target of, so how do you feel about that? Do you feel people are just misunderstanding it or?

When I think about it, I'm not in this to be popular, so I don't really care and or register it. I mean, haters going to hate what you're going to say. Yeah, okay. No, I think that's probably the best way to think about this. Okay, well, Stefan, thanks so much for coming on. And, yeah, it was really pleasure. Cool. So, also, one last thing.

So we've started this new bribing contest in which, like, once a week, if you leave an iTunes review, we will give a teacher to one of the people. Now, you can leave a review saying that we are clearly banking-operated robots controlled by Goldman Sachs for having. having Stefan on and talking about an evil project like Ripple, or you can say this is the best hour you've ever had in your life and you've never learned so much. Either way, it gives you equal chance to win this T-shirts.

So, yeah, check that out. And actually, if you do leave one, then send us an email at show at epicenterbidcom. So we know who you are because we can't figure it out otherwise from iTunes. So, yeah, thanks so much for joining us. So we put out new episodes every Monday. You can of course subscribe to it on iTunes, your favorite podcast app, SoundCloud, or get the video on YouTube. That's YouTube.com slash episode and BTC.

And yeah, that's it. Then we look forward to being back next week.