 Okay, so I want to tell you about all the stuff and Here we go reality check the fork happened. It was painful. It was painful for all of us We lived through the confused narratives the changing the moving goalposts. They moved more than the ball did the ball was kind of stationary and The constant fighting the bickering the trolls and everything. It's over. Okay, so it's now time to look forward and And and so we have two different visions for how this currency is or how the original currency could play out One of them is a store of value Proposition and they've been pushing that that that that vision pretty hard and they've been celebrating with champagne or whatever that thing is the the idea of high fees and let them do that let them let them drink their champagne and Let's focus on the the world ahead and that world starts with increased adoption and on this front You all have been doing a fantastic job. Okay, so the number of good news in terms of adoption that's coming in is is Incredibly good to hear and with adoption comes greater scale the need for greater and greater scale So how are we going to get that scale? Can we possibly tweak the block size forever? that's a good question and Indeed, so let's actually sort of make sure that we've established the baseline Increasing the block size was exactly the right first move. Okay, so we have extensive measurement studies on this my group was the first study of actual resources that Bitcoin knows actually possess and we've done longitudinal studies Spending multiple years on how much these these resources the network bandwidth and so forth have been increasing So I can tell you from a position of scientific measurement that it was indeed possible to Increase the block size without any threat to centralization as as the Bitcoin cash community did and in fact Continuing to tweak the the block size is going to be important because we don't live in a static world We have more and more bandwidth available and as we do we can increase the block size But will is that going to be enough and here? I'm here to tell you maybe something that you might not be so prepared to hear but it is true regardless Greg Maxwell and friends are not entirely wrong. Okay, you can't tweak the block size completely to You can't continually tweak the block size over and over again and and scale by just simply upping that one single parameter Demanding use cases like microtransactions internet of things and so forth Require a level of scale that you just can't get with bigger and bigger and bigger blocks Not only that but now that we have a sort of vibrant community We have to lead intellectually it is not sufficient to sort of take the code that's developed elsewhere and Incorporate here and it's very dangerous to be in a position where you are on an island some place And there's another island somewhere else and these people are living under a perhaps kind of insane perhaps self-imposed Resource restriction and developing a whole bunch of resourceful protocols as a result They can just come over to your island and suddenly bring you you know all sorts of things That that that you you might not be prepared for you need to be on the lead here So it's essential that we understand how to scale up further from here And what I'm going to do in this talk is talk a little bit about options for scale both on chain and both also off chain Because the main my main point there is going to be you've been indoctrinated with the lightning network But there are other options available and there was a coherent vision for off-chain scaling that fits into the Bitcoin cash world And I'll talk if there's time available about scaling in the network the often underlooked Sort of orphaned child of this whole stack is the network stack So let me tell you a little bit about Bitcoin and G and G is short for next generation This is work from a long time ago now maybe four years ago that we did when this whole thing was just starting We decided me and and it I a all and Robert friend in this and and another student if a ginger to to see how one might Scale up Bitcoin without changing any of the the core vision parameters if you will that is retain the miners Retain almost everything there is to retain about the system when it comes to security when it comes to assumptions of trust and Yet allow a couple of orders of magnitude by a couple. I just mean one or two Unfortunately, this is not going to get us all the way to a million This is just one piece of this like three pronged puzzle So I will tell you a little bit about Bitcoin and G But I would really very much encourage you to read the paper to take a look at it because this is sort of a road map for How I believe a proper scientific inquiry about scaling ought to happen The paper does three things it proposes a protocol and you've all seen protocols proposed Yeah, everyone seen those terrible white papers, you know a sends a message to be decent You know those sort of descriptive things about how you could do things But in addition to doing that protocol description, it also introduces metrics Quantitative measures for how we can achieve how we can measure whether or not we're achieving our goals and then it actually develops a full simulation framework for Running the entire Bitcoin network in a data center and simulating it all and then it actually looks at the result I want to give you a sort of a sense for what we did So just so that we sort of understand how not to fall into that pitfall where you've got sort of Developers making up stuff and doing design by gut. Okay, so they're good. No, we all know how that actually played out So let me tell you a little bit about Bitcoin and G the core core thing here The core insight is to notice that in Bitcoin blocks serve two purposes Okay, so here's a block and there's one component of this block That is the crypto puzzle and what it really does if you if you sort of put on the right glasses Is that the crypto puzzle gives the right to a miner to dictate what happened in the last 10 minutes? It's sort of like a leader election by solving the crypto puzzle You become the leader for the preceding 10 minutes and you get to then Serialize what happened in those 10 minutes you get to add to the end of this distributed ledger Once you see things this way once you realize that this is a Retrospective protocol where we elect a leader who then determines what happened in the past We can then make a very small change and a very small change takes us to an interesting Interesting location now. What would we like to do? We'd love to be able to expand the block size We'd love to have huge blocks, but if we do that then we start getting forks We start leaving the smaller miners behind or We might want to have huge frequency of blocks But again if we do that we leave small miners behind again That's going to cause Centralization what I would love to do is to be at the intersection of these things I would love to have low latency and high throughput but without a fear of forks and If I go back to this picture when this stuff is happening when regular Bitcoin transactions are happening What we really have is a network that's mostly quiescent There's almost nothing happening except you know Joe pays you know Jane and so forth Just transactions are being floated around then suddenly a block is found and everybody goes into a frenetic activity They try to disseminate this block as fast as possible and then you know it goes back to quiescent state again It would be wonderful if we did not have such frenetic periods because those periods are really the stressful times So how do we do this? Well? I'd like to have you know as I said high throughput low latency secure protocol So here is how Bitcoin ng proposes to do this. It says take your blocks Split them into two that serve two different purposes we're going to have something that we call a key block and That key block is going to serve the purpose of selecting a leader if you will and And the number of micro blocks and the micro blocks will serialize what's what's happening as it happens in real time and Instead of a retrospective protocol that serializes what's happened in the last ten minutes We're going to have a forward-looking protocol that selects someone and And that someone then says okay this happened and this happened and this happened and this happened So how does that look? Okay? It looks like this inside every key blocks key blocks are going to be found with proof of work Using the same old miners we always have Inside the coin base. There is going to be a key just like there is a key today There is going to be no other identification of the miner no IPs no ports Those are all immaterial and not at all on the blockchain on the ng blockchain But there is a key just like there is today and that coin base key is then used to mint a series of small Micro blocks that contain transactions as they happen So the key things to notice here are that key blocks are very small and rare They come every ten minutes just like before Micro blocks are very very small, but they're frequent so instead of having a giant block every ten minutes And you're going into that frenetic mode what we're going to do is we're going to smear the creation of that block Across those ten minutes and take away the centralization pressure from the system So pictorially how does this look so here? Let's let's imagine that we have a blockchain those those two green things show us that the tail of the blockchain And then a miner comes in and they find a block a key block that says hey I found a good crypto puzzle and here's my key and From that point there's no other content in that block It's just their key that accomplishes the leader election task and from that point on as Transactions come in that miner takes that transaction signs it with the key in his coin base So he then starts minting these micro blocks up there that I've shown two of that That continually extend the the blockchain those blocks are threaded together in the in the typical blockchain manner Okay, so you can't just take one out without breaking the entire chain So so that of course gives us this ever expanding chain And at some point some other miner in this case the yellow miner might come in and find a key block of his own And he will hook in you know to this this ever expanding chain at whatever point that that he hooks in that Okay, and that might actually cut off a couple of the micro blocks, and that's okay so so the The interval between these key blocks as I mentioned is still ten minutes so the centralization pressure is completely off and And the the frenetic activity is reduced immensely the block sizes here are now tiny And they're actually distributed across a bunch of time now those of you who are technical in the audience would say hey That sounds Nice, but there are all sorts of attacks you could imagine yes indeed There are all sorts of attacks they come down to two Two formats if you will one is the leader could hide micro blocks or he could ignore micro Microblocks that is if you are the red fellow here You could you could hook in not at the tail end But earlier and try to steal some micro blocks for yourself Or if you're the blue guy then you create a long chain, but you don't give it to someone you squat on it Hoping that you will be the red guy and and sort of extend your own chain longest yourself Those are both reasonable attacks and and in fact that's where the discussion gets a bit complicated I'm not going to go into this too much except to say the reward structure is a little different Okay, so to gain theoretically incentivize people the right way Here in this universe what we what we do is we set up a slightly different game Where the miners are compensated both for the work they did and also for the work that they extended Okay, so we're trying to make sure that everybody works most for the public good And so that red fellow is not only going to get some money for the red micro blocks He means but he also gets some reward for the blue micro blocks that he extends So that particular setup will require some kind of math and I'm not going to go into it Except sort of to flash it at you and say okay Well, so if you read the paper you will see that there are some constraints on On what that payoff ought to be and it turns out that if you actually compensate somebody 40% for their own blocks and 60% for the blocks they extend It leads leads to sort of a virtuous cycle where everybody has the right incentives to work together and extend the longest chain So so that's the core idea. I'm very happy to take more questions about this After the talk if you will but but then now what what next and the reason why I came here is Probably to try to sort of talk to you about how to really do this Okay So at this point one could start emailing and and start hiring troll armies or a small army of people who fixed the narrative or whatever The proper terminology is but the real right thing to do at this point is to actually come up with some metrics And I was shocked when after three years of this block size debate nobody actually did this So what we did was okay. Well, we're going to propose a new protocol. We should know how to measure this This is this is the only way to do science. So we came up with with about five metrics Mining power utilization fairness consensus delay time to win and time to prune. I'll talk about some of them later on But you need metrics if you're going to be doing this And the second thing is you need an apparatus for measuring and evaluating protocols We then went into the lab and we measured everything There's the measure for every single public Bitcoin node and we've been doing this for years now So our goal was to create a miniature world a replica of the Bitcoin network inside our own basement So for every node out there I want to have a node in the basement Configured with the exact latency and bandwidth characteristics between that node and every other node So that's that's the miniature world idea so Let me sort of show you some of these metrics mining power utilization is very simple It's the amount of of hash power that went into the chain And if you look at this for Bitcoin, it's in the 99% if you look at it for Ethereum. It's actually by design It's lower It's an interesting metric. It's a simple one the time to win and time to prune are the complicated ones Fairness is an interesting one You want to make sure that the people who experience orphans I experience them in regard to their hash power You don't want the small p small miners experiencing more orphans Because if that's what you have then the small miners will tend to coalesce and that will give you centralization pressure So and there are a bunch of other metrics as well So we built this miniature world thing It's actually very difficult to figure out the network conditions between nodes But we did our best to do that and that we created a one-to-six replica for the purposes of this paper back in 2014 one-to-six the replica of the Bitcoin Bitcoin network So let's see and then we took realistic data from the hash power distribution. So here are some Interesting graphs. I think I'll flash only one at you On purpose back then we did not want to take a side in the Bitcoin side of Bitcoin block size debate So we we changed the x-axis quite a bit So that it wouldn't be directly comparable to to any of the sizes that you know So here we essentially changed a couple of parameters so that we could see interesting effects But they're not going to correspond directly to sizes that you know because the frequency has been changed And we also don't have no we can't wait ten minutes with nothing happening in between key blocks We don't have that much life to live If we did it that way we had to speed everything up So just to give you a sense though the entire block size debate is being was was being waged on What would be considered the left-hand side of this this graph? So going from one one megabyte to eight megabytes is essentially always Summer over here. So the big fight was around from around here to around over here You can see the error bars so people are getting really worked up over over that over essentially noise Okay, so you could have gone to the right and there would have been no change in fairness At all and so that entire narrative just quantitatively did not make sense We tried to sort of bring this up with the the core devs and so forth But of course they have a bunch of mechanisms for for shutting down any any scientific discussion. So No, I didn't mean to I don't maybe that was too harsh. I mean, they're individually great people But but the forums did not lend themselves to a scientific discussion. So Let's see So you can see the the interesting thing here. They are not wrong So as you make the block size bigger and bigger Bitcoin does decay its performance drops In fact immensely and you can see of course that Bitcoin and G does not and it should not be surprising because it's by design still issuing one block every ten minutes and Is not going to suffer from from the same problems as as regular legacy systems So I have a bunch of other graphs. I'm going to skip this stuff so What's the sort of takeaway from this portion of the talk reparameterization can only take us so far and So maybe Bitcoin ng can give us in the medium term a way to scale up if if reparameterization Runs into trouble. Okay, so I'm not here to push Bitcoin ng You know, you guys can take it or not But at least you should know that it's there Okay, and it's good to know that there is actually a path out in the future so But how much can it actually help it gives us about a hundred transactions per second or thereabouts and that's good Okay, I mean it gave us a hundred transactions per second back back then right so or thereabouts It's it's good, but it's not quite Google scale. It's not going to be IOT scale So what should we do? Can we do better than this and I want to talk to you a little bit about going off chain Okay, so you've all been indoctrinated you've been sold the idea that the only layer two solution is the lightning network and I get I will grant to you. I will grant to you that it's got the coolest name Okay, it's got a cooler name than what I'm about to show you, but it's a false narrative Okay, it wasn't the first layer to solution and it's not the best layer to solution There are other alternatives that every community should should explore and I'll talk about one of them So let me tell you a little bit about the lightning layer problems I'm not sure if I have enough time to do this, but I'm not gonna be able to fix all of them Some are inherent to layer two. Okay, so I have a slightly different vision for a layer two than what you hear from core But let me actually all the hands are going up. Yeah, okay So I've been hoping to have some time to write a blog post about this Well, the first and biggest problem is it has limited capacity the capacity of that network is unknown Despite so many years of noise around it. Nobody has actually gone and done a scientific study and That actual emergent capacity is going to depend on the credit relationships between people and we don't know what that's going to be like And it's not that that credit graph is going to look nothing like the Facebook social graph The fact that I know and love Tariq does not mean that I'm going to extend to him You know a hundred thousand dollars worth of credit. It's just sorry, but it's not gonna happen so that network may or may not actually have any carrying capacity and It's economically broken So I was a broken there is no benefit unless the endpoints have frequent interactions So Tariq and I and end up interacting once a year that is not enough of a reason to open up a channel between us even if he were credit worthy for a hundred thousand and Worse an exchange needs to have funds on hand tied into channels that are proportional to the float for per person times the number of people that they want to support so Suppose somebody could receive up to ten thousand dollars from Coinbase and Coinbase has I don't know ten million users So ten million times ten thousand Ten billion a hundred billion dollars gets tied up. Is that right? I don't know That was so you know in that that future world where they want to do that if LN really takes off a lot of coins Are going to be all tied into these payment channels The system is insecure requires new intermediaries, you know So you all know about the Mali ability fix the kitchen sink known as Segwit Segwit That if I were to propose it, I'm sure it would not have been accepted and yet somehow Somehow it got into Bitcoin core So so it requires keys to be kept online You need this new thing called watch guards to watch the chain 24-7 or else you lose funds you blink you lose But I close the channel with an old state and you lose your money and counterparty misbehavior can lock up funds Some bad thing happens and now we have to wait for time out. So Let's see it erodes privacy I Don't know. So yeah. Oh, yeah, this is important. I should mention this routing involves an inherent trade-off between finding out what paths are available in the network and That is an inherent trade-off between efficiency and privacy. So to find a good money route to you I need to know about everybody's credit relationships And I need to be able to do this in real time as it changes And if I were to do that then I'm actually monitoring you right? I find out that that gentleman over there has a credit relationship with the the the person behind him Well, I do it over and over again and suddenly if the capacity changes I know that somebody paid someone else over that link and that is is that gives me the ability to do what we call network Tomography and that is not at all a good outcome Routing is difficult except for hub and spoke models and you know, the user experiences overall a mess So will I be able to fix all of this? No, I sadly it requires much more research to fix all of them But I will tell you about how to fix some of them in particular The capacity of the network I can't fix But we can fix the economic model and we can fix the the insecurity the malleability fix the requirements for malleability fix So I will let me present to you one way of doing layer two that works on Bitcoin cash today Without segwit without any malleability fixes and the core sort of foundation for this is a new Capability that new chips have many new chips from Intel arm AMD and ledger and so forth have a secure element capable of doing three things this element can run some code with Confidentiality without revealing what it is do what that code is doing internally with integrity without allowing you to attach a Debugger and change what the code is doing or allowing the code to be manipulated in any other shape or form Even if you are attached to the bus even if your unepoxied USB port has some foreign device Even then you cannot molest what that code is doing and then the third thing that you get out of these These new chips is attestation and this is the critical part I can ask a question to Omri Let's say and say what hash are you running right now and he has to give me as his chip Will give me a fingerprint of the code He's running so I know exactly what behaviors he can engage in and I also know what behaviors He won't engage in and this is the critical insight so instead of a Byzantine world these chips carry us to a fail-stop world So the worst you can do is stop execution, but you can't willy-nilly misbehave so Yeah, I mentioned this you only need to trust the code that's running inside the chip and it has this fingerprint thing That's so Alice can ask Bob What's what's your hash and he says something like it's a dead beef and you know exactly what Bob is going to do? Bob himself cannot molest his chip to execute something other than the dead beef code after that attestation So let's see the key thing with t-chan is that both parties exchange their secret keys and become capable of Closing the channel on their own without cooperation from the other side and And so there is a typical very similar to lightning There's a setup event where one of the parties talks to the blockchain, but subsequent to the setup They no longer have to communicate with the blockchain Alice can pay Bob without monitoring the blockchain at all Hey, this is crucial. So it's kind of like an open-dime device So you and I can go off off off the network. We're in the middle of the Sahara and we exchange exchange cash securely without ever ever touching the blockchain itself at all and when we go back We then one of us will close the channel and we can't close the channel with a stale state Why not because this is a hardware solution if it were software. I wouldn't know what you're doing You would have access to all previous states. So if you read the lightning paper about it's what 60-something pages and 59 of those pages are all about what do you do about the stale state? What do I do when the counterparty uses an old state? That's advantageous to him to close the channel that cannot happen here I know that he discarded the old states because I know what Cody was running because it was running on the secure hardware And I knew the hash of it. So We implemented this and we're seeing enormous enormous throughput per channel that's on my dinky laptop I was getting 78,000 transactions per second when sending money from London to Ithaca and the delay the the latency of a transaction in this universe is 0.4 milliseconds plus Network transfer so in this case about 40 milliseconds to cross the Atlantic and then fraction of a millisecond to actually make the payment So if you have multiple channels the throughput scales linearly and as I said it doesn't require a malleability fix We just don't care about about misbehavours involving malleability So now you can say well, aren't you trusting Intel here or aren't you trusting somebody? Yes. Yes, we are absolutely that's the trade-off we're making so That that is true But I would like to point out that you also are trusting Intel last time I checked all of you just issued a Transaction without manually checking the computations, right? So nobody has a little abacus where they read it the the modular Exponentiation it's a little tough to do so so you are already trusting that hardware base anyway But it's also important to notice that the trust is channel bound So if Intel decides to misbehave they cannot affect the system they can perhaps manipulate and and allow my counterparty to steal the funds I have in the channel with my counterparty that is the max So if I have a thousand dollars open to you and you're in cahoots with Intel you could steal my my thousand dollars and Intel makes about fifty eight billion dollars a year whether or not they would do that with you seems seems kind of nonsensical to me and But much more importantly we're not beholden to just Intel. There are many other vendors and And one could could actually conceive of other ways of doing this so that you you don't trust any single vendor alone So overall if you were to compare t-chan to lightning network, it's more secure except it's it's more secure because the keys Are not online and software They are in hardware itself. So it's it's hardens the implementation Counter-party misbehavior does not lack of funds the t-chan approach requires no change to the underlying blockchain No segwit. It doesn't require a change to the ecosystem. We don't need watch guards. We don't need banks That's the whole reason why we are here, right to be our own banks So you don't you don't have to be beholden to someone who's connected to the blockchain 724 and monitors it for you Or you don't you are not beholden to a routing hub It's economically efficient. I didn't mention this perhaps but exchanges need to tie up only money that's proportional to their float and So that gives us a factor of n improvement. Okay, so instead of n times float. We just have the float Coinbase will spend at most ten million dollars per day paying out. That's the amount of funds they need to have in the channels and So what's the vision here? It is not my vision that we employ this to create connections between you and your aunt Bertha, okay? That is not a workable model. That is the model that leads us down that UX user experience That's going to be absolutely terrible, but these kinds of solutions are crucial for frequently interacting businesses This is the kind of thing you need when you have two businesses like exchanges that frequently communicate It can give us great improvements in arbitrage and prices between exchanges So this is that's I think my vision that that in fact, that's the vision we're working towards which is not This deployment for users. Although the users could actually just as equally participate in this model I think that's a that's a bit of a pipe dream The sort of the the more sensible approach is to actually connect this to exchanges So and of course we do multi hop transactions and so on and so forth But these are all kind of if you ask me as an academic I will say these multi hop transactions are an academic idea. They're very cute very very cool We worked really hard to make sure they're atomic So your money doesn't get paid doesn't get stuck between your aunt Bertha and and like the person She knows in Boston and so forth on when it's getting routed So either payments happen up fully or they don't happen at all. So that's a very nice thing to have but But I don't think that I just don't think that that's where this this vision is going to take us The this I don't think that layer 2 should be should be deployed to the masses It's just not ready the routing will not be ready for another five to ten years at at best the routing protocols So let me just take one more sort of segment and talk to you about another layer the long Forgotten the long overlooked network layer and tell you a little bit about the work that we've been doing to speed up Communication of financial data between people who have it and the people who want it So it's users exchanges have it miners demanded users have it miners demanded or vice versa Miners have it and exchanges and users want want access to blocks So how what are the sort of the options here? Well, let me tell you a little bit about what happens today in a peer-to-peer system like Bitcoin We have a source a relay and a peer And this is time going down in this case and the source might have a block that they have discovered And they want to transmit that block. That's the start time That's the end time and let's say that takes time beta relay gets it relay checks the block and You know it takes time alpha to check the block then relay transmits it to the peer and that takes another beta And this poor guy receives the block at alpha plus 2 beta. Okay, so if there are n hops is alpha plus n beta so Yeah, okay, so and sorry. No, no, I apologize. It's not alpha plus and it's n times alpha plus beta is the is the full-on transmission latency so can we do this can we do this entire process better and the answer is of course and This is something that we already did for the Falcon system Falcon relay network that actually powers both Bitcoin cash and BTC so let me tell you about what Falcon does. It does the following thing a source finds a block The relay receives it. He checks only the header to make sure that the source is not misbehaving and then sends it on to recipients so so indeed it takes the the the Transmission of the block still takes time beta and but the delays the alphas have been compressed and If you have n hops then we have a big savings of n beta n times beta Remember that beta is the the significant time here So we actually took out a linear factor We took out the whole and to speed up dissemination of financial data So where does that leave us? It leaves us with new infrastructure that can be used to serve as the foundation of Currencies and Bitcoin cash could use it but many other currencies could use it as well In fact, we're talking to various different projects that want to use it as sort of an Akamai for financial data Whereas Akamai is sort of destined or designed for sort of throughput of web information This is for low latency. It's an overlay network for low latency with a bunch of nodes scattered around the globe Whose purpose is to send data from people who have it to people who demand it? So that brings me to the end of my talk Oh, I'm gonna conclude by pointing out that it's a super exciting time to be working on cryptocurrencies I showed you three three projects one is NG. It's available as a medium term solution for For on-chain scaling I talked to you about teach and and I hope I sort of tried to impart upon you The the possibility that there are solutions other than the lightning network out there and with the right sort of Culture we can pursue them and we can do better than what LN can do. It is not the end all of layer 2 third is I think I tried to sort of Discuss with you the idea that the network layer is has been often overlooked and improvements there can can actually greatly improve the scale of cryptocurrencies and If I leave you with one message if I could leave you with one message it would be this Whatever we all do we have to use scientific methods in discussing protocol changes. Thank you all Yeah All right Can we please welcome that presentation with great Q&A questions We only have about five minutes. Do we have any questions from the audience that reflects the quality of the presentation? I'm Justin Bonds from a cyber capital first of all I want to say I have a lot of respect for for the work you've been doing and we need more real scientists in the space So I really love that My question was regarding mining Centralization in the way that I see it miners generally don't run nodes pools run nodes And because of the nature of variance and the way that the Bitcoin network works will only really be 10 to 30 pools And I would argue these 10 to 30 pools can afford to run, you know, very large for nodes. So I'll just I was like your thoughts on that. That's sure. I think you're right So how can I what you so what got said was the minor community? will tend to center will tend to sort of coalesce into Dozens of of players and these players have the resources to have actually strong machinery I agree with this and how could I not agree with this because we have eight years of history to look back upon and that's what we see so indeed that's that's entirely true and And yeah, I don't think there was anything that I said in my presentation That actually conflicts with what you said as you're absolutely right So but can those pools afford to increase? Let's say block size to infinity. No No, I would just if I could make one more quick question Well, I would say I think you have a point in that We can't scale to cover all use cases and I agree we need second layers for micro transactions and and so forth but but I would say that The cash use case and also considering that that pools run for nodes, you know I see that is being quite viable For being on chain. Oh, you agree. I agree. I agree. So look, I'm not saying something like Bitcoin it's a Bitcoin and you should be deployed tomorrow That's how I started the very first thing was to improve was to increase the block size the next step I think I was I have a bullet point that said next step should probably be to tweak it up as Technology permits us and we should keep on doing that But also I don't think we could tweak it up to infinity So there will be a time hopefully in our lifetime that we will see when these kinds of things start to bite us And at least we should be able to sleep at night. I personally can't sleep well at night unless I know That I'm covered, you know down the line So now I think I hope many of you can go well, okay If we actually hit that bottleneck, we know exactly what to do My old question one more Make it good. All right, make it good Great presentation So I was wondering is there anything that can be learned from Ethereum and whether you've looked at any of the stuff that they're doing around sharding for things for example Yes, there's a lot to learn from Ethereum So we did we just published a paper last February at the financial cryptography conference that looked at centralization quantified centralization for Bitcoin pre fork and Ethereum and it was a longitudinal study expanding multiple years So there were some funny funny results from there And there are too many to do that would be a separate presentation on its own. Well, I'll give you one hint Ethereum actually has less resources the average Ethereum node has less bandwidth available to it than Bitcoin nodes and Yet Bitcoin actually Ethereum processes more transactions per second than Bitcoin So they're operating at a higher point in throughput with less resources. So it's quite possible Another thing to notice I mean there were like a lots of interesting Outcomes in that space. They I mean we could go on and on they they suffer from more orphans They have a slightly different protocol designed to compensate for that They definitely need something like what I alluded to at the network layer to try to reduce that And I can just go on like this for a long time But the paper actually makes a cogent case of those two systems But sharding specifically have you looked at sharding at all? Yeah, no, we are we've thought about sharding quite a bit It's it's definitely one of the things that is being actively worked upon by my group and by my colleagues at I see three It would be great to see I mean I think the the framework that you put with with measurements and matrix is incredible because that will make these Different approaches comparable side-by-side and that's that's the really exciting bit. Thank you. Thank you. Thank you I mean for awesome presentation awesome question