Epicenter - Learn about Crypto, Blockchain, Ethereum, Bitcoin and Distributed Technologies - Jonathan Levin & Peter Todd: CoinSummit London – Interviews

Episode Date: August 1, 2014

CoinSummit London Conference Series – July 10 and 11, 2014 Our second CoinSummit Episode is out! In it we have two fantastic interviews. In the first one we discussed transaction fees, the economics of Bitcoin and the coming relaunch of Coinometrics with Jonathan Levin. Jonathan is a co-founder of Coinometrics and just finished his Master’s thesis on Bitcoin at Oxford University. In the second interview we talked with Peter Todd, one of the thought leaders in the Bitcoin space. We covered the economics of transactions, mining centralization and the state of Bitcoin development. Peter is a core developer and also an advisor to projects such as Mastercoin and Dark Wallet. He’s also come up with a number of important innovations such as treechains and stealth addresses. Episode links: CoinSummit Coinometrics This episode is hosted by Brian Fabian Crain and Sébastien Couture. Show notes and listening options: epicenter.tv/coinsummit-london-02

Welcome to Epicenter Bitcoin, the show which talks about the technologies, projects, and startups driving decentralization and the global cryptocurrency revolution. My name is Sebastian Couture. And I'm Brian Fabian Crane. On July 10th and 11th, we were in London for the Coin Summit conference. This two-day event gathered approximately 250 investors, entrepreneurs, and developers to discuss some of the most important issues facing the Bitcoin and cryptocurrency ecosystem. This episode features two interviews. The first is with Jonathan. Levin, co-founder of Coinimatrix, and he discusses transaction fees, Bitcoin economics, and the upcoming relaunch of Cornometrics. In the second interview, we talked with Peter Todd, who's a Bitcoin co-developer and a consultant to a number of different projects such as MasterCoin and Dark Wallet. He discussed with him mining centralization, the state of Bitcoin development, and the economics

of transaction fees. Yeah, so we're here with Jonathan Levin. until recently he was at Oxford during a dissertation master's thesis on Bitcoin which I'm extremely eager to read and he's also a co-fan of coinometrics he was just presenting at the startup showcase perhaps to start off

do you want to introduce yourself and tell us a bit about perhaps let's start with your dissertation sure so yeah we know each other we've been on the podcast before but Jonathan Levin for the the people that haven't listened. Jonathan Levin, I was doing an economics masters at Oxford

and I wrote about the economic incentives that are introduced in a system that spends time away from consensus. So we know that Bitcoin has been designed as this really elegant computer science solution to the Byzantine Generals problem.

But that does mean that although we come to consensus about what transactions have happened, we do spend time as a network away from that step. and what that means is that Which are you talking about the time after a block The time between a block's mind and the next block So the time between a block is found on the network

And the time at which everyone Now finds out about that new block So when a miner finds a block They individually find it And they tell say eight peers about the block And then those eight peers verify that the miner has done the It's got the right proof proof of work and all the transactions in it are valid,

and then they propagate it around the network. So slowly the network builds consensus in knowing what the latest block in the chain is. But you can see that initially, only one person knows the new state of the Bitcoin network and everyone else is left in the dark. So as the block spreads around, so consensus is built. And essentially that means some interesting things on the mining side.

So my thesis was about, well, what does that information delay mean for someone who is trying to mine a block? And if blocks that are larger propagates slower, is that costly for the miner? And what I found was that basically the marginal cost of including a Bitcoin transaction in a block was proportional to the information delay, the network system. What did you define was the marginal cost of including a transaction? Yeah, so you need to make a couple of assumptions to make some progress here. But essentially, I found that Bitcoin currently at the low fee is not incentive-compatible for a small miner.

So if you were solo mining on the network, you should not be including any transactions, which I thought was really problematic. And the fee would have to go up by about four times in order for you to break that incentive compatibility problem. four times not that much. I mean, it's then 20 cent on the transaction of... I was interesting because I was asking Gavin. I was talking with Gavin at the Amsterdam Conference. And I asked him something very similar,

and he said, according to his calculation, maybe he uses some other methods. He said he brings his 10 cents, the marginal cost of including one transaction. Yeah, so Gavin uses the average propagation delay figure from a Christian Decker paper back in 2013 as his way of calculating. So poor data.

Poor data. I'm in constant, I'm constantly trying to give Gavin updated data, but so far, we haven't managed to do that. I'm really looking forward to smart fees and actually providing some good data for them to make decisions about how that market looks like.

Because I think that's, I think fundamentally we need to do some modeling before that gets implemented. I think the miners also need to understand this. So one of the things is that it's an underappreciated issue, under-researched, and no one knows what the marginal cost of a transaction is, and miners should be aware of it, specifically small pools, actually. So what would the consequence of this be?

I mean, do you expect that maybe if you're going to post your dissertation or it's going to get some attention maybe now that people hear about it, do you expect that miners will stop, more miners will stop, including transactions in their blocks? No, so right now, I mean, when we say it's not incentive compatible, I think, I mean,

you're still talking about small amounts. You're still talking about $10 worth of returns on a block which is worth tons more. So you're talking about a pretty marginal decision for the miners, but I think generally you, this is an important issue to bear in mind for sort of long-term,

long-term development priorities. So when they're thinking about smart fees, that's a long-term development goal, and we need to actually model this first so that they have the best tools to really structure this market correctly. So what do you think fees should look like in the future? So I think fees eventually must actually provide the incentives for security. I think it's very interesting that right now the model of Bitcoin security is that the inflation rewards,

the block reward essentially provides the only security for the network. And initially that's like the best idea ever because you don't want the security of the network to be proportional to the amount of activity if there's no activity. But once there is activity, then you actually do want your security to be proportional to that. And so transaction fees are a really big part of that. I think you'll see Ethereum and some of the alt coins take that more seriously in the next 12 months and try and see how you can create a decentralized cryptocurrency where fees provide a large incentive for miners to contribute.

I mean, it might take a different proof of work. It might have to be more proof of stake driven. But, yeah, I think that's the crux of cryptocurrencies right now. Yeah, no, I think it's a super important issue. And so you think that some of these other currencies, like Ethereum, or does Bitcoin 2Baloney's stuff can solve some of these problems? Yeah, so, I mean, like, I've been reading a bunch of stuff that Vitalik has been up to recently.

I'm quite encouraged that he's thinking very hard about how transaction fees should incentivize security. but I think that Are you talking about his blog posts? His blog posts and also he posts in there's an Ethereum Economics Skype group that he often posts

and he's also got some stuff on his GitHub if you go and check out Vitalix but GitHub he's got some interesting Python code to model the economics of a decentralized crypto. So he's a mathematician, he's pushing the boundaries for sure

and whether they have a solution for this, I don't know. Well, it's great to, you know, it's great to see that you talk, I think about this and I think it's something we also talked about with Mike Kern recently. We had one on a podcast and he thinks it feels the same way. Anyways, he doesn't think that transaction fees, although he's perhaps even more skeptical. He said he doesn't think transaction fees will be able to provide network security in the future and he felt that there

should be more some sort of contractual something assurance he's talks about assurance contracts like maybe I mean I had any idea well you could integrate in the wallet software that there's some sort of

insurance contracts or basically a sort of a donation model right I mean which of course seems problematic too so yeah I think one of the other

interesting discussions is if there are, for example, services that depend on the security of the Bitcoin network to function, something like a coloured coin, master coin, etc. And there are services that have essentially high value propositions built on top of Bitcoin. Then actually you might be able to sustain a kind of fee model based on those applications rather than simply just the Bitcoin fee structure. and that might provide security, although I think there's a lot of thinking. That's a very early stage idea. I think Peter Todd has said that the security of Bitcoin in some ways increases if you have multiple applications built on top because 51% attack on all of those is what's needed to 51% attack on one of them.

Really? It seems like it should be the opposite there. Because you have, I mean, let's say the block reward now gives an incentive for security, but then if you have things that run on top of it, where you don't get a block reward, you can have maybe your incentives to attack those. Your incentive to attack those will be higher for sure and potentially to attack the Bitcoin network

to gain 51% of the power. But essentially you have multiple targets that you should be that you need to double spend on. in order to make the attack worth it. So each one of those is going to provide a level of security. And in order to attack one of them, you're going to have to have the capability to sort of overcome the security

that's provided by all the services. So let's just be simple. So you've got Bitcoin. Bitcoin provides security through this block header. Colored coins have a fee structure that feed into Bitcoin security because miners, only mine colored coin transactions if the fees high enough.

Yeah. If the fee on colored coin transactions is significant, then Bitcoin hash rate will go up in proportion to the colored coin fees. And therefore Bitcoin becomes actually more secure based on the fees of the color coin. Oh, that's interesting. I mean, yeah, there's obviously attack vectors. So the incentive to attack colored coins, I think, is a very big. big issue. So double

spending on a colored coin transaction might be really easy if they're based on the Bitcoin blockchain, but double spending a Bitcoin transaction becomes less likely if a colored coin is implemented on top of the Bitcoin blockchain. That's one way that I think about it. So you founded

Quintometrics, and I think you just released, you guys just released a new site recently? We're releasing a new site in the next week or two, so maybe by the time the podcast has come out, you would of it will be up and running. And so can you just tell us about, well, the new site, what are the new things that you'll

be offering there and perhaps where that's going? Yeah, so we decided to launch a new site. We wanted to make the information even more accessible to someone who's not from a Bitcoin community. We launching a industry grade price index that has four quantitative rules attached to a it rather than qualitative rules that we see in all the other indexes today. Can you just explain that that means? Yeah.

So essentially, what you want from an index is you want to try and capture the fair market value of a Bitcoin today. And now that's a very subjective question. And people have implemented price indices during the Mount Gox kind of saga. And all of those price indices were designed in such a way so that you could basically create an index that excluded Mount Gox. They set some rules saying you've got to be able to process transactions like this or you've got to be able to do stuff like that. And essentially that doesn't give you great transparency moving forward

because you don't know how often that criteria is applied. You're not sure if that criteria is going to change next week based on some other exchange that comes along that behaves quite strangely. And the only way to really do this is actually to look at the numbers and say, does this exchange, does it behave in the same way that the model, is currently behaving. And that's a statistical question rather than a qualitative question. And so what we did was we sat down and we thought, well, okay, we need a certain

amount of volume on the index in order to have sort of price discovery. We also don't want its price to be consistently different from the median in the market. We also don't want its change in price to move very differently to the rest of the of the market in any given time period. So we apply that on a weekly basis at the moment. We update that every day as it currently stands. We take a volume weighted last price. So the last price traded on an exchange

is weighted by the 24-hour volume on that exchange in our current calculation. And we also have a final rule that is a black swan rule where we say that if anything goes absolutely mad on any of the exchanges they automatically get dropped off. So if you apply those criteria and go back, would Mount Cox have been excluded? And from what point on?

Yeah, so actually you'll see, according to our criteria, exchanges go in and out a little bit. There was like a period where Mount Gox was diverging in like sort of October time and they then get cut off our index. but it happens in and then in February I mean they get dropped off as well but if you were tracking Mount Gocks

throughout a lot of the indexes were tracking Mount Gox throughout the whole thing and I think that that's problematic because there were obviously signs that stuff was going wrong and it was somewhat of an isolated market so our price index looks quite smooth

throughout that time and any traders who were sort of using our index would be able to apply the same criteria that we apply. We apply it consistently so they would know when Mount Gox was going to be excluded. So I think the transparency, and as you say, we've back-tested the index. We've had a look at what would have happened had we applied the index going backwards. And we use that to calibrate the index. So, yeah, we've prepared some documentation and we're open for people to approach us to license it out.

So are you going to charge fees, I guess, for API calls above a certain rate? Yeah, that's the idea. And depending on the use case as well. So I've had some conversations at Coin Summit about implementing a price index for a derivatives product. They need to do it slightly differently. They might need to tweak it and we'll do some licensing agreement and charge for that service. Yeah. Going back to quantum metrics,

some of the other features that are coming out on the new website. So we had a volatility index on our previous site. That's going to be expanded to cover all the exchanges to cover the price index. We're going to have multiple models

if you want to compare them. We're going to have some products, probably not on the website, but if you are a derivatives platform and you need to do some volatility modeling to know whether you can offer 10x leverage, then we can help you out and figure out whether that's a good idea

or whether that's going to blow your bank. So those are a couple of the features. We're doing some more blockchain analysis at the moment. Lots of stuff about estimating volumes of casino activity, trying to get into understanding how much commercial activity is actually happening on the blockchain and really trying to provide some more high-level metrics about velocity of money, circulation, wealth concentration,

and all of that stuff will be coming out quite soon. Is anybody else doing any of this? Yeah, I don't think we're the only people doing this. I think that there's an index started by Pantera Capital, which was trying to track, Bitcoin adoption, which was quite good. But I think that some of that is more high-level stuff, sort of GitHub repository stuff.

I think really there's actually there was a presentation here by a guy called Chen, who's at Blockview. They're starting to do some blockchain visualizations, which are quite nice, and trying to get inside the commercial activity angle. There's a few of us doing it. I like to think that we're being very rigorous and ahead of the competition, but maybe we're not. So I remember the last time, or one of the last times we talked,

you were still finishing your dissertation, your thesis, and you also were thinking about doing a PhD. Are you sort of committed to building a company working on quantum metrics, or is this still the idea of perhaps writing a PhD thesis on some Bitcoin topic? Yeah, I think, I mean, at the moment, I'm working absolutely full-time on coinometrics, trying to make it a company. If in October, in a few months' time, nothing has happened with coinometrics, which I would be very surprised about if that was the case. But if that were to be the case, yeah, I might consider going back into academia. I guess it's more, I enjoy far more working with our coinometrics team than academics.

Put it that way. I can understand that. Well, great. Thanks very much for taking time to talk about us. Thanks for having me on the show. And, yeah, so we'll try to have you on a more regular basis to give us the state of Bitcoin. The Economic Bulletin.

Yeah, we could call it the Economic Bulletin. Jonathan Levin. Perfect, guys. Great to have you here. So we're here. with Peter Todd, who's a Bitcoin core developer, and who just wrapped up a talk about Bitcoin mining and the risk of a 51% attack.

Can you just kind of give us a brief summary of what was said and your thoughts on the state of mining right now? Yeah, well, I think what's interesting about the g-hash.io situation is on one level, I think it really does show that the underlying incentive structure behind Bitcoin is flawed. you know, we should not have a system which has incentives for people to get to 51%. And indeed, the system should have disincentives for people to get to 51%. You know, even if GHS.io was completely honest and, you know, they were not going to attack Bitcoin.

The reality is they can get hacked, you know, they can get coerced, they can turn evil later. Not to mention, you know, it's just, it's not good PR to go and say, well, in reality, actually, yeah, one per person. controls most of the network. So it's certainly something we need to fix long term. In the short term, though, you know, I think there's rough consensus among people in the mining community, people in the pool community that we just have to really agree to limit how big pools are, which is on a voluntary basis, but in the short term, that's a kind of thing that has worked before and I think we'll work again. And in the long term, while we're researching technical solutions. And so what are some of those potential technical solutions? I mean, are there propositions out there to

limit mining so that the network will, like you say, disincentivize people from, or disincentivize pools from reaching this critical mass? So there's a couple different mechanisms here. Maybe the first one to go speak of is the actual process of solving the math problem that Bitcoin blocks need solved to exist, which is the hashing. And essentially, of course, hashing, you plug in some hashing power, you turn electricity into math problems and heat. And the nice thing about this is on a physical level, that's guaranteed to be decentralized to some level.

Because useful things you can do with the heat to run this hashing power at the lowest possible cost, that is not an opportunity to exist in one place. If I'm heating my greenhouse, well, there's a lot of greenhouses around the world. And there's going to be thousands of people making use of waste heat from mining. There's going to be thousands of people making use of pockets of cheap electricity, like wind power generation in the middle of nowhere. That's on affordable trend power line, back to civilization. What can you do?

You can go put some mining power at the base of your wind turbine, you know, with a sat link. And these are all good opportunities. The problem is currently all those people will still have an incentive to sell that hashing power to a big centralized mining pool, where you get your attacks or your incentives to control the network. And one of the things we can do is essentially make it impossible to run in a pool. Andrew Miller has a very interesting idea along these lines, which is some math techniques where if I solve a block for you and near the pool, I can then steal the reward later.

and you have no way of even proving who did it or that it was done. So I want to go back to these incentives in a second because I think I have some doubts about that. But when we talk about this idea, would there be any way of achieving a change like that because wouldn't the mining pools in the end have to support that and it's really against their own interests?

Well, after all, the mining pools and the network can change. You know, one example that you could do to implement this is you would create the mining pool, lend all mining pools. And the rules are if you want to mine at this, you've got to go prove that you support this new protocol. In the moment, the majority of miners agree that this is a good idea. Okay. That mining pool ceases to exist. And I guess you could do some kind of crowdfunding that maybe you would even pay people join that mining pool more.

Absolutely, you can. Yeah. You can essentially subsidize. this to end mining pools. That's a great idea, actually. Maybe that is feasible. Yeah. You know, and what's interesting is, you know, this is really a political thing.

You know, this is implementing a piece of technology. We're not still sure yet what exact piece of technology is the right way to do this. And when time comes to implement this, it'll really be a matter of doing the politics to get consensus that this is the right solution forward for Bitcoin. I'm sure there will be stakeholders. You don't want this to happen. But, you know, it's just a matter of convincing. people, this is the right way forward for Bitcoin.

So I wanted to come back to the thing you said about, you know, like having heat and distributed places with kind of, and I agree that factor is there, right? But if you have economies of scale on the production of the hardware itself, right?

And that works against that again. And I've always, and I guess especially when you have the hardware depreciating value so quickly, it seems that those, at least at this stage, are much stronger. Maybe that will change in the future, but... Well, you know, in the future, we can

fully expect the depreciation of hardware to slow because we went through this period in the ASIC mining business where we kept creating ASIC miners that had smaller and smaller geometries, more and more advanced types of ASICs. You know, we went from, say, 250, was it, nanometers to 160, DE now 22, which is top of the line. And you're not going to, now we're at the point where the A6s are manufactured with the same technology as the best chips in the world.

And equally more as law, this whole computer's become twice as fast every 18 months. That's actually ended. You know, that's technological progression has ended, and we do not know what will be next. It may never advance at the same rate it does. At the same time, you know, it is trying to. true that ASIC production is extremely centralized. There's probably a half dozen companies in the world that can make a competitive ASIC right now. What kind of risk does that pose, do you think? I think it's very simple. Governments can go say, you know what, we're going to start regulating

mining equipment. And even though your ASIC, who we think is mining equipment manufacturers, they're really designers who design equipment and then subcontract the actual creation of it to other companies. Yeah. And that subcontracting, so long as it's possible to go pay someone to go make an ASIC for you, we're fine. And if that gets regulated, we're not fine. I think I know some people are in Berlin that's very involved,

that has been like researching that a lot. And at least what they said is that the kind of companies that can actually do some of the subcontracting work, that is extremely limited as well. So, you know, that's just something that, you know, you're not going to be able to change, right? Because that's how chip manufacturing works today. Yeah. We know of no other way to do it. On the other hand, I mean, keep in mind, this is a problem we face with all computing.

You know, Richard Stallman has said for many, many years that DRM technologies, digital restrictions management, they are very dangerous because we can end up in a world where you cannot buy a general-purpose computer. and the reason why you still can, it really comes down to politics. You know, we've pushed back so hard on people trying to introduce mandatory DRM that we still have a free computing environment. I think ASICs for Bitcoin mining is a bit more precarious situation because it's a special purpose hardware. But, you know, ultimately control of chip manufacturing is a very politically touchy subject.

You know, so we do have that angle. and it always comes down to the more people are involved with Bitcoin, the more beneficial uses it has for society. The harder it is for anyone to go say, you know what, we should go stop this. So what do you think is going to happen to the area? Because, I mean, I see the solution with the mining pool, and that's actually really interesting.

I think that's something that, you know, I think that's something that can work. It makes sense to me. But what about the mining, manufacturing? Do you think we will find some solution there? or do you think that's just too late because of the proof of work we have here? I honestly have no idea.

I mean, I can kind of think in terms of what are possibilities, but if you wanted me to put a bed on it, I wouldn't really know. And I think it's too hard to know. Equally, you know, maybe we will have some technology advancements. Hypothetically imagine if genetic engineering got way better and we could grow ASICs in a bat. You know, that'd be great.

I don't have any clue how that. could happen. I don't think any biologists have any clue how that could happen. But I don't know, maybe we'll see something that crazy. You know, the sky's the limit. You mentioned politics quite a few times. There's been some recent discussion about internal politics within the Bitcoin core development community. We interviewed Mike Kern a few weeks ago, and he seems to think his impression is that there is a fundamental problem right now within Bitcoin Core development where developers are not getting along

and development is not advancing and essentially only one person is actually doing anything to make the Bitcoin protocol advanced. So what are your thoughts on this? Do you share this view or do you think that this is an exaggeration?

I think that view is a very misguide. view of how Bitcoin progresses and how Bitcoin really works. If anything, I would say that the Bitcoin ecosystem is actually advancing and progressing and developing at a faster rate than it ever has before. And, you know, I say this on the basis of all the things people are doing with the Bitcoin protocol. You're seeing things like Multisig finally get implemented through multiple different companies for different purposes.

You're seeing things like escrow get implemented. You're seeing things like entire embedded consensus systems getting implemented, that leverage Bitcoin to do all sorts of stuff that were never envisioned before. You know, there's just so many ways that people are taking the Bitcoin protocol and advancing the development of it. Now, that's not to say that the Bitcoin core protocol self is changing very rapidly, but that's okay. You know, the analogy I would make is a web.

TCPIP doesn't change very much. It's a very slow process to implement upgrades, and that's fine. It's a core base layer that everything else is built on. But by the time you get up to web browsers, JavaScript, and so on, innovation happens very rapidly. And you should expect that kind of stack where there's base layers that are stable, less stable, less stable, till finally get really experimental stuff.

And that's what should happen. I think where Mike comes from is that, you know, he's proposed a lot of ideas that really change Bitcoin core protocol, and there's been a lot of disagreements on whether the ideas are any good, and a lot of them have been shot down. And, you know, what do you expect if you propose major changes? It's going to be very hard to get consensus on it, and that's fine.

You shouldn't expect Bitcoin Core protocol to change much. Equally development of refining the implementation, I would say that's, again, It's happening at a relatively rapid pace. It has to be very careful because we can make very expensive mistakes with this. It's an $8 billion industry. And predictably, people have been careful, and they've been pushing unit tests along. They've been pushing refinements to the protocol implementation along.

Peter Wool, I think I got that pronunciation right. He's done a tremendous amount of very interesting work. We had TX mutability. that was you know there's many aspects of it and a lot of them have been solved and we're working towards solving that

we have things like fee estimation which you know we're kind of working out bugs of that you know to go say like there's just one person working on things I think is very misguided so when he says that Big Concord development has pretty much

grounded stopped you think that this is actually the norm, that things should move along slowly and that... And after all, I mean, when he says that it stopped, it's running at the same speed that always has been, which is slow and careful

and deliberate. You know, it hasn't changed, actually. If anything, we're seeing it maybe increase a little. And some of that increase of speed has led to some bugs, which predictably leads to a pushback and, all right, let's be a little more conservative next time.

You know, the hard fork that we had last year, that was a mistake in part caused by the increased speed of development. You know, equally, I personally caught a bug where the next implementation of, you know, the next release of Bitcoin Core could have been spending $500 or more per transaction in excess fees. Unfortunately, that bug was caught and it's never made into release, but, you know, that's an example, the sort of thing you run into if you're not careful and you're not cautious.

Yeah, so one topic I've been thinking about quite a lot in the last, you know, a few months is a transaction fees. And I'm like a lot of questions, doubts about how these work. So I'm curious, can you briefly talk about, you know, because now they're working on floating fees or, you know, Gavin's working on floating fees. Can you talk about that and like maybe how those work and you've used on that? Well, first of all, it's important to remember that we've always had floating fees. And I think the viewpoint that says that floating fees are something new is one that takes a view that Bitcoin is a network with unlimited capacity. You know, there's some people who take the view that transactions should always be nearly free. There should obviously be tons of computing power to go around to process transactions, and block sizes should be unlimited, and we'll just manage it somehow.

Whereas I would go argue, and I think many people are in the same position here, Bitcoin as a system fundamentally at the lowest level does not scale well, because everyone who needs to validate the blockchain has to have a copy of it on an ongoing basis. And, you know, in computer science terms, we'd call that system with n squared scaleability, which is usually thought of as very, very poor. Fortunately, you know, we don't have to use Bitcoin directly. We can go build systems on top of it. we can go build trust-free systems, like micropayment channels where you know you and I can send money to each other rapidly changing how much we send and then a whole collapsing that to one transaction.

And, you know, what comes down to is fees are just a way of bidding for space in the blockchain. And you've always been able to bid what you want. you know, with this fee estimation stuff, we're just seeing a mechanism to go estimate that. And, yeah. So, I mean, right now, right, the standard way of, if you have a double spending attempt, right, where you have two transactions spending the same inputs. Right now, the miners take the first one in general.

Yeah, that's a separate issue, yeah. And I know you actually wrote a patch for Replace by Fee, which in my view seems to be extremely problematic because it would destroy zero-confirmed transactions, like the ability to accept unconfirmed transactions. Yeah, she argued the opposite. What it does is it, it aligns the incentives correctly. And what's interesting about this, so I'll explain. So replaced by Fee says, rather than you accepting the first transaction you see, you just accept what. which everyone pays the highest fee. So basically if I send money to you, I can then modify the transaction later, which may include not sending the money to you anymore. But what's interesting is that if then I modify that transaction, broadcast a new one that sends the money somewhere else, you can go respond to me and say, you know what, screw you,

I'm going to go take the money you rightfully sent me and sent it all the fees. And now, chances are that transaction will be the one that gets mine. And I'm still going to end up paying someone. But the chances are the second one gets mine. The third one. So I create transaction number one, paying some merchant. I create transaction number two.

Paying myself. Paying myself. With more fee. And then the merchant creates transaction number three. That spends the output of transaction number one, all the mining fees. So the person trying to go rip off the merchant still ends up paying. and the merchant may not get the money, but businesses do not operate on

the need for every single transaction happen without fraud. They operate on the need to keep fraud low. And by turning something where I can go get away with fraud into simple vandalism where I don't earn anything from it, really removes incentives. But that also presumes that somehow merchants are prepared for this. Well, they're already prepared. Very few merchants except.

Ziracons because the current system has been... Well, BitPay does, right? No, they don't. I guess they leave it up to the merchant, yeah. They don't actually. BitPay... The way they handle

XeroConf is a merchant who's in a position where it doesn't actually matter if a transaction gets double-specked provided they find out before they say ship the goods. They'll, of course, tell you your orders accepted instantly.

But it doesn't mean that they've sent you a package in the mail instantly. Yeah. And as an example, ATM machines were you deposit some bitcoins and you can get cash out. I personally been in discussion with some vendors of this stuff as well, some operators. And nearly everyone's turned off Syracomph transactions in a way that exposes them to fraud because the current system is insecure and it's very easy to double spend.

You just have to use some technical tricks. So do you think, unconfirmed transactions are just like, you know, it's just like unsecure. A point, there's no way to make them secure. Well, I think with replaced by fee, this so-called scoreth short strategy, it goes a long way to making them secure. With micropayment channels, you can make them secure. The point is that the existing system has been proven not to work for a way.

But then with replaced by fee, it would end up increasing transaction fees, no? No. No. Now, replaced by fee is independent transaction fees. Right, so I want to get back to that because I didn't quite understand it. So I send a transaction with a merchant, and then I send the transaction myself paying a higher fee, right? This is far I understand it. But then what does the merchant do? You have just tried to rip me off.

I tried to rip me off. Now, the merchant can respond by taking your first transaction, sending it all the fees. Yeah. Which means that the merchant still doesn't receive the money. So the person trying to commit fraud doesn't gain any benefit. So what does the merchant do with the first transaction? They spend it all the fees. they spend all the fees

they spend it all to fees because they have the money now they spend it old fees a miner will include both transactions they earn the benefit of all those fees and the person trying to commit fraud

has no gain but this if you have you know I recently there was this a pool which Jafel was very interesting with spit on new thing where they would have a mining pool just of miners supporting

this kind of basically replaced by fee type in. Bid undoes had nearly no traction. I know they have no traction. And what's interesting about Bid Undo is if you're a mining pool operating the service, you have an incentive to tell the merchants but the double spends so you can go get more fees. True. But it would be possible there to do privately, so when the merchant realizes it's too late. Sure, but after all, if you make the assumption that mining power is well distributed, there's never going to be one pool that has significant health.

hashing power to go pull this kind of attack off often. Other than the replaced-by-fee method, where you have the scorchurch defense. And, you know, that's not even getting into all these other mechanisms like green addresses and micropayment channels and, you know, off-chain transactions which protect this. Never mind the fact that few merchants are actually vulnerable anyway. I mean, what's important with replaced by fees is it ensures that incentives are aligned, so we're not depending on honesty or depending on economic self-interest.

And that's a much better position for Bitcoin to be in than trust. Not to mention, I mean, replace Pepi is useful, too, for many protocols. You know, regardless of who's trying to rip who off, in many cases, it's just useful to be able to change a transaction after the fact. And if I pay three people in a row, it's more efficient and I'll spend less money. If I modify the first transaction to then pay the second person and the third person, and that makes Bitcoin as a whole more efficient. Okay.

No, that's interesting. It's interesting here your perspective on that. I guess another thing that may be very refilled, though I think it's maybe not as important of an issue. One thing I've also been thinking about that seems like it doesn't make any sense to me. Is right now, right, if I send a transaction, that doesn't be a fee. Like, eventually it will get included in the block because it, like, becomes, rises in priority over time. Oh, that's just a simple thing of miners giving way transaction space for free. Exactly, but it makes no sense, really, from a minus perspective. Yeah, it's one of those things where it has a low cost to them.

It's a very small amount of space per block that's reserved for this. So for them to give away maybe a dollar worth of transaction fees every block, it's sort of one of those decisions where it's not even worth their time to turn that functionality off. Some do, but would I bother spending five minutes hunting down the configuration option for that? Probably not. it's kind of an inertia thing I wouldn't be surprised if that goes away in the future but for now it seems to kind of just

happen by inertia okay well maybe one last thing on the transaction fees so do you think the way in the future we will have you know maybe the block limit will become a real factor

so that we'll have blocks that are full and you know people will have people who pay less you know they will have to wait for a future block with less transactions in it? It's a misconception to think that blocks are not already full. In that there is more demand for blockchain space than there is capacity.

However, that demand tends to be for extremely low value applications. So those people get priced at the market by transaction fees. Therefore, they use other ways to transact with Bitcoin. So you mean that? Is that correct? What you're saying is that if it was free, then it would be full, but so now people use off-chain transactions? Well, I'm saying that the transaction fees, as they are,

are already expensive enough that they price people at the market and they use other systems. Just Dice was an example, where they gained a lot of business from Satoshi Dice in part because per gamble, costless money. Yeah. and equally you see Coinbase to Coinbase transactions. That's actually a very common case.

Someone is a Coinbase account, they pay a merchant. It's entirely off-chain. No transaction fees. You know, and also you go see people who want to spam the blockchain with junk for advertising messages and whatnot. And what do you know? They don't get in the blockchain because they can't afford the fees. Equally, I'll also point out that there's this misconception

that miners don't fill up the blocks or creating small blocks. You know, you go see blocks that are nearly made by it all the time. there's a subset of miners who mine at theoretical capacity, so to speak, and they create nearly full blocks all the time. And to the extent that you don't see every single person doing this, well, there's not that much different between half a megabyte and a megabyte.

You know, when transaction fees are worth so little, again, it's an inertia thing. Why would you even turn that little knot and change things? Well, I was asking Gavin about that at the Amsterdam. conference and he told me that the cost of including a transaction was a 10 cents in that it makes your block bigger and increases your risk of orphan block. That's a very complex topic. He is right in the broad brush trucks. The problem is that there's no good way to get the cost down without creating incentives where, for instance, a larger minor has a lower cost than a smaller

minor. Because the larger minor will never orphan themselves. They'll always build on their own blocks. So if you try to optimize that number a lot, you further make it give an incentive for pools like ghash.io to exist. And they'll earn more money

than smaller pools, which is already the case, but the amount of extra money they earn is, say, 0.1%. And you could easily get that to say 1%, 5%, 10%. That's a big edge. Thank you very much for taking a time talk to us today. Hope you enjoy the rest of the conference. Yeah, you're welcome.

Thanks so much. Thanks so much for listening to our coverage of Coin Summit. If you enjoy this episode, please support us with your donation. It really helps us traveling to conferences and produce high-quality content for you. You could donate at eBitcom.com slash tips, where we have our tipping addresses, and also an option for donation subscriptions. Your support is much appreciated, and special thanks to those generous souls we have already donated to the podcast.

You know,