EB91 – Vitalik Buterin: Ethereum Frontier Launch, Scalability And The Road Ahead

Transcript

[Music] this episode of epicenter Bitcoin is brought to you by Ledger makers of The Ledger Nano Hardware wallet have peace of mind in knowing your private keys are protected by industry standard physical security go to ledger wallet.com to learn more and use the offer code eb09 at checkout to get 10% off your first order hi 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

K and my name is BR F CR we're here today with one of our recurring guests which is vitalic Ban of course he doesn't need much of an introduction and he's been on the show before in in December last year we talked about scalability no actually not so much we talked about proof of stake ethereum and it was a really great episode and many of you will have heard that eus just launched their Frontier release finally so we we have vitalic on and I'm super looking forward to it hugely so thanks so much for coming on vitalic yeah thank you glad to be on yeah so actually the frontier launch so we're recording this on on on Saturday August 1st it was just a two days ago uh and and I was right

where I am at this moment and and you you were there as well and it was It was kind of how did that actually feel for you finally finally reach the step cuz when when you guys started the idea was it was going to be like in a few months know the launch yeah I'm definitely relieved as you know this has been going on for quite a long time now and you know originally we thought we were going to launch back in uh well originally we thought we were going to launch back in February 2014 and then it get kept on getting pushed back and back a couple of months and you know there definitely were times where it seemed like it was just going to keep on going forever but you know finally it's happened I guess so we've moved past stage two and we're

on stage three now so that's great so you guys were both there at the ethereum office uh how did it feel to be there when uh when when it launched and when there was the Genesis block yeah it was great I we had a quite a lot of people maybe somewhere from a third to a half of the dev Team Al together in Berlin got to celebrate a bit cool yeah I mean it's of course hard to know right how big of an event this was perhaps it was big an event perhaps not you know history will tell but it was definitely it was definitely nice know cuz having you know knowing these people well who have worked on this and having seen them work on this for such a long time and it was always abstract thing and now it's actually life right it's

actually here and running so I think this is a huge change I'm sure it's going to be very motivating for people too now yeah I think so so let's uh get started by by talking about the frontier launch so could you explain what is the frontier launch in in in the different steps of of uh launching ethereum sure so the frontier launch is the first time that there is going to be the live ethereum Network and you know ether is going to be valuable um there's uh you know people are going to actually be able to develop applications on it and we they'll have con they'll have confidence that the that whatever state they those applications end up happening actually will end up carrying forward so it's not

just a test net that I'll get deleted in two or 3 months um it's a kind of very early developer focused launch so we've done is we've sort of split up launch into these multiple stages where the first stage is Frontier where we you know explicitly say you know this is the Herby dragons no safety guaranteed Network it's uh it could easily end up having bugs on it it's uh just some something for miners to help get the network started and for developers to start building on it you know the only interface is a command line so this is the sort of stage that we'd like to see the ethereum network going for for a

couple of months and if it goes for a couple of months without hitches then you know we'll move on to the next stage which which we're calling Homestead and I originally the plan was that Frontier would also be sort of quasi centralized so the ethereum foundation would to be like checkpointing blocks every day but we've uh decided there are a bunch of reasons why we can't really do that to that extent so we ended up uh delaying delaying Frontier a bit more and pushing it into the points where it's kind of halfway to where we wanted Homestead to be in the first place so differences between Frontier and home Homestead that remain are in to some extent marketing so it's kind of the difference between Alpha and beta um there are still are a

few kind of default kill switch features that are in in Frontier so the main one is this thing that we're calling canaries which is basically there are a few contracts on the blockchain and the core developers have the ability to send transactions through these contracts which basically Mark a particular chain as being invalid in some way so you know if say the blockchain forks and we decide that one fork is is valid and the other Fork is invalid and it's just happening because one of the clients has a bug in it then we would launch a transaction which is specifically targeted to the bad chain which would basically eventually get into the

blockchain and basically mark it as be as you know being wrong and and everyone will automatically stop mining on it so so so people the clients then will automatically stop mining exactly yes so that's something that's I mean it's centr to some degree it's there in Frontier it's a kind of training wheel that's going to come off in in Homestead um soide from that I mean yeah Homestead is just going to have a i the software is going to continue to improve obviously um it's the point where we start recommending that people actually start building you know more

serious and long-term applications on it instead of uh basically just sort of VI uh very early development and testing step three is Metropolis so this is where we release Mist which is you know the ethereum browser and ethereum is finally going to have a gooey yay um if you want you know you could actually download and use Mist right now so you know there is a very early early version of mist it works there's a wallet and a multisig wallet in that version of mist they work but they're still sort of very Alpha stage and we're and we're not claiming that they're kind of ready for anything close to mainstream use at this point but you know at some point we're going to sort

of formally release Miss then yay people will be able to use it is that also when the IDE comes out the mix IDE the mix ID is out already um it's continuing to evolve over time but it's uh I there is a version you can use it okay because there's some confusion around there at least for me is is this what what you're calling now LF zero or or is that something different so mix is a kind of piece of software that came out of Alf zero so LF zero itself you know is just sort of very simple gooey for the C++ client and the sort of main direction that the C++ client is going to take is being for development

and mix is a kind of guey that's designed for developing contracts mist is the thing that came out of go ethereum okay and so what other can you tell us about what what other components are being released with the frontier launch or what you know so there there's some other websites like stats Etc is that being released now okay so Frontier release includes number one the actual blockchain um number two it include it includes the uh development tools M miss mix uh number it has two clients uh so the go client is the one that's kind of the

most audited and and the most likely to be to be sort of free of serious security issues C++ client is undergoing an audit right now it should become more and more ready over the over the next month or so as the audit progresses the python client is not audited but you know it's there you can theoretically run it it SS it syns the blockchain so that's uh all in place um there's also a few of these sort of Statistics Pages there's a stats. e.com is the one that we're running it's this sort of view of the entire network you can see the world map you can see you know nodes mining on pretty much every continent of the world

map except for Antarctica but then those darn P Penguins I heard they like an xtm or something um the then there are a bunch of third party services that are up already there's I think etherchain do.org is already up so that's a third party service that's not being run by eev no it's not being run by eev it's third party and you know it's a block Explorer it's got all the blocks it's got the transactions and accounts it's uh showing all yeah so so I had a question uh why what are the what are the the utilities of having different

clients like why go client and a python client and a C++ client um there are a few reasons behind that and one point is that it's kind of there are kind of political benefits to having multiple clients because it makes it easier to say this is a system that's controlled by the where the sort of the protocol specification is actually defined by a formal specification first and implementation second whereas with a lot of these other crypto protocols they basically say you know while the implementation is the spec and if the implementation has a bug in it well that's just that's just a fun little feature that's that that was supposed to be there all along so that's basically the you know by having

multiple clients it kind of makes it harder for one of them to start sort of running away from all the others and uh adding on its own extensions and so forth so that's I mean this whole sort of governance issue is important to a LGE degree because you know we've seen how sort of things are evolving in the Bitcoin ecosystem with um Gavin and Mike pushing out Bitcoin XT and so forth so that so this kind of centralization of the ecosystem by means of one client is basically being able to define the spec is one of the things that we're trying to avoid another issue is just from from a security perspective is that you know by having all a very very large suite of

tests and by by seeing that you we have three separate implementations and Implement and pass all these tests perfectly that just makes sure that there's a much lower probability that there are going to be any kind of security issues uh between these clients um so you know basically you can think of one of the clients as being kind of like an automated generator of test of tests for the other now of course there are costs with a multiple client approach so you know one of the costs say that there's just a a much larger amount of development effort that's requ required so I mean opinions inside the team differ to some degree um I mean even though I was the one that was pushing this multi client approach much much

harder early on I and right now I do think that we over invested in that to some degree but it's you know there there are trade-offs between these things it's uh definitely still an interesting experiment that I'm sure a lot of future software projects will learn from cool so the launch as far as I know and asked you I think about this yesterday or something everything went perfectly fine right mhm yeah I think so okay that's right and and so one thing that's also notable right about right now in beginning is it's not actually possible to to do transactions because uh can you explain why that's the case uh yes yeah that was a fun one I remember after we launched

there was this really nice article on uh on the btcoin Reddit ethereum launch is incapable of doing transactions which was just perfect so you know basically what we did was we decided that we were going to set the gas limit in ethereum started starting off to 5,000 and 5,000 is only about a quarter of the 21,000 gas that you needs to actually send a transaction so you know this would be the equivalent of like Bitcoin having a block size limit of 20 bytes um the uh idea behind that is to just to let the network sort of start off and stabilize first and then only after the network is stable people can sort of

start bring setting transactions in so at this point it's looking likely that the sort of thawing process is going to happen in the next uh few days maybe Sunday or Monday um so the thawing process basically is that we ass set we release an update to the client which basically and that update basically tells miners to start voting the gas limit upwards so right now the way the gas limit is set in ethereum is it's basically this voting mechanism where each block has the ability to set a gas limit which is within about 0.05% of the previous gas limit and so over time you know if you think it's too low you can sort of push

it up if you think it's too high you can push it down so it's a sort of voting mechanism that should Target roughly the median of what the of what all the miners want in the long term So within that sort of protocol rule there are of course different strategies for how to vote so the current strategy is to Target just the 5,000 limit the strategy will update to is targeting a limit of 31 41592 which is the same as we had on Olympic then after that there are other strategies that we're thinking of so one strategy that I came up with is targeting a particular Uncle rate so Uncle for those who aren't aware is this idea in ethereum that if you mine a block and the block doesn't make it into the main chain you can still get

included after the fact and still receive most of your reward so the original idea behind this Uncle mechanism was that if uh basically to avoid penalizing miners that are too badly connected because you know with ethereums having a 15-second block time if you are badly connected that is a serious penalty and you know there might be sort of centralization issues with one particular cluster ending up having all the power and so forth so that's something that we wanted to kind of mitigate but the other nice benefit that this unor read mechanism ends up giving us is it ends up giving us a kind of in chain reading of uh basically how well the Network's doing so if the uncle rate

is 0.1 that means that you know 91% of the blocks are making it into the main chain and 9% aren't if the enle rate is 0.4 then that means that 71% of the blocks are making it into the main chain but 28% aren't and based off of that we have an idea of kind of how healthy the network is and the theory is we might be able to basically use that in order to sort of Target the gas limit so we could say we want a uncal rate that's not higher than 0.4 if it's higher than 0.4 then start adjusting the the gas limit down if it's lower than 0.4 and it looks like transactions are starting to fill up then push the gas limit upward so so so let me uh understand this properly is

is the reason that the gas limit is related to this Uncle rate because the smaller blocks propagate faster yes exactly okay so then basically if if you notice oh blocks are propagating too slowly there's a high Uncle weate lots of Orphan are happening you need to decrease the block size uh okay that's a that's a very interesting mechanism I think also interesting know if you think sort of Bitcoin right because similar discussions happened there a lot and it sounds like you guys yeah mean I think a lot of the sort of re theoretical research we ended up doing on Olympic has uh has Sim applicability to the Bitcoin block size debate because uh

What's happen with Bitcoin it's having a lot of the same concerns you know if the block size gets pushed upward then propagation time is going to increase and there are these centralization risks and so forth so you know in bitcoin's case I think currently the block propagation time is somewhere from from around like 20 seconds or so to get to the m to to the great majority of the nodes and if it goes up from a megabyte to 8 megabytes then they'll go up from that to 3 minutes and that and you know it'll start seeing larger Uncle rates so it's a lot of kind of the same math I mean my intuition is that with the way Bitcoin currently is now something on

the order of 8 to 12 megabytes is going to be safe anything higher I mean you know if they figure out how to make invertible blo lookup tables work well then that might be a way forward but we'll see well let's come back to this later we we actually wanted to ask you about that but the the one thing we didn't get to talk so much about the last time you had you on and it's topic we wanted to talk about is scalability and and you it's I think it's the one topic that you've been most interested in and I guess it's a topic and Bitcoin people interested in as well and and of course you guys have the advantage that you sort of can go from scratch really think through what way to go about it it's a it is quite

frankly I think a complicated topic for uh for us who don't spend that much time uh thinking about the sort of theep technical aspects of it but you've written a really nice uh a white paper which is a 50 page very comprehensive white paper so so I just want to briefly sort of read out the problem which I think will make sense to anybody so in the in the what do you call it the sort of in the abstract you WR so all blockchain consensus protocols have an important limitation and that's that every full participating node in the

network has to process every transaction right so it's pretty obvious that if that's the case then well then we're going to have the sort of discussion we have in Bitcoin now well the only way to scale is to make transactions huge or maybe do some sort of offchain thing yeah make the blocks huge so what is your what's your thinking how should what is an alternative design that does scale better I mean to start off there are two easy but bad ways of solving the scalability problem so the first way is to just say okay let's have ,000 blockchains there's two reasons why that's that's bad one reason is that

interoperability is harder and it becomes harder to sense transactions that go have effects between one blockchain and another blockchain the other much bigger reason is that each one of these blockchains is only going to have a thousandth of the security the second bad way of fixing scalability is to just say let's have big big big blocks and let's rely on all the consensus participants to having big big big servers and that's the solution so you know the first one compromises security the second one compromises decentralization and particularly I think some of these centralization effects might end up compounding with each other so for example you know with mining if we see the block size increased by factor of a thousands then

we could easily see just about everybody use two or three mining pools and then switch to using one mining pool so that's going to be a bit problematic now the approach that I'm outlining in that big long paper is basically that instead of trying to have every node process every transaction you sort of organize transactions into groups that have effects to to on Sim on sort of the same part of the state and then you take these big collections of transactions and you send them off to a random sample of validators to verify so let's say you have 10,000 transactions that affect like this sort of one corner of the map and you you might think of that corner of the map is say being all

addresses that begin with 35 and so one of the important properties of this kind of design is that you actually have to sort of think a bit about positioning and you have to kind of think about well okay I have a contract and I want to position this contract sort of close to other contracts that are going to be interacting with it a lot so you know this is sort of General theorem that it's impossible to parallelize arbitrary computation and therefore you know that means that if we want scale ility the abstraction is just going to have to be weaky to some degree so you know the idea is anyways that you have this big long set of trans of maybe a th 10,000 transactions with addresses beginning with 35 and this this collection gets uh

sent off to this random sample of 135 validators then that collection has a what we call a collection header and the collection header basically you know contains a hash of the collection it also contains the state route before applying the transactions and the state route applying after the transactions so the person who creates the collection has to has to figure out what the what the sort of very fine grain effect of all those transactions is and just summarize it with a Merle route then once that once that's created the person that creates the collection sends that collection with a bunch of Merle proofs with the Merle roots with the header along to these validators the validators all

check that it's and they all sign it then when you have the signatures you take the The Collection header and then you add the 135 or signatures to the collection header and then you put that into the main chain and the main chain basically follows a principle of well if if 2/3 of 135 random we selected validator say it's correct then it's correct so the header chain almost doesn't even know what's happening at the bottom level and so this is the kind of principle that it's uh that it's working on you know that you have all of these different groups that are creating these collections all you know hundreds of collections in parallel the collections all get signed in parallel then finally

the signatures make their way up into the top chain and the top chain kind of keeps track of the status of these moral Roots which are basically summaries of what goes on at the bottom so I mean that's of course a very a design that seems to make intuitive sense right so you have all this stuff you try to divide it up give it to some subset you know they process that and then somehow you integrate it back together but the the the obvious thing that jumps out here you mentioned contracts have to be close to each other I mean what what are the implications of that so like I come with my random wallet yeah so this to some degree this depends on what application you're using so I think with e with ethereum

particularly in some cases it's not going to matter so for example examp you could imagine augur existing in sort of one of these subspaces then you know Lazo existing in some other Subspace then some stable coin existing in some third Subspace and so forth and if people that use one application tends to just use one application then it's going to work mostly fine the problem comes from well what if I have some money in or some more General even what if there's some contract in say Subspace 35 and it wants to send off a message to something that's in Subspace 68 then basically what's going to happen is that well one of two things might happen there's actually sort of two strategies

for handling this there's the synchronous strategy and the asynchronous strategy so the synchronous strategy basically says that okay what we're going to do is we're going to wait until there is some collection of transactions and that collect that processes simultaneously subspaces 35 and 68 then once that collection happens in that collection everything can happen synchronously now and you know the theory is eventually that collection is going to happen the transaction is going to get in and it's going to have exactly the effect and it's all going to work basically the way it works now the problem with the synchronous model is that it's going to take a very very long time until all of these sort of possible

pairs of subspaces end up sort of appearing in a collection at the same time and if you imagine a big long chain of events that triggered by a transaction that might affect say five or 10 random different sub spaces then it's going to take a billion years for it to get in so the asynchronous approach this is the approach that uh I it works equally well it works pretty well under a kind of slightly modified synchronous model it's also one of the approaches that Dominique Williams has been working on is that it basically you say that if a transaction has an effect that goes into a different sub space then basically it doesn't have the effect immediately the transaction

just kind of creates a receipt and a receipt you can think of as being something that just sort of gets Merkel hashed in and it's kind of registered as saying Subspace 35 created this receipt then half an hour later the on Subspace 68 you would say you create a transaction that kind of pings the contract on the other end and then what that contract does is it checks okay are there any new rece are there any receipts in Subspace 35 that are not yet registered as having been sort of used up by Subspace 68 if if there are then that means that there are new receipts and off of these new receipts is sort of start started making a computation so the idea is that you know sort of

every and then when you do make a computation off of the off of these receipts then the receipts end up sort of in Subspace 68 the the the receipt ends up being marked as used that a can't be used anymore so the idea is a sort of asynchronous two-step process it's uh vaguely similar to Peter Todd's tree chains in some sense as well um so you know basically you have a kind it's a sort of debit credit system you have you know the contract inst STS Bas to be five grading in effect and then later on this the contract in sub space takes advantage of that effect let's take a short break so I can take you to Paris I dropped into lamei Bitcoin the house of Bitcoin in the heart of silicon saner home to many startups including Ledger and I spoke

with nikoda Baka Ledger's CTO about the level of security in their devices Ledger is based on a smart card chip that is exactly the same smart card chip that you have in banking chips and in passwort in passports today so that's something that has been validated for the industry for more than 20 years today's the favorite second factor is the phone um when using a phone you are transmitting an encrypted version of your transaction to the phone to the screen that can be displayed to you and validated to you so we leverage in two insecure devices your desktop and your phone to create a secure validation with Ledger we believe that Hardware wallet are very primitive today they are just used to confirm transactions and to display some validation and that's it uh

in the future as Bitcoin protocol moved toward smart contracts to a lot more use cases we believe that those devices will evolve into something that I would call consensus devices which basically will be able to confirm for specific specific rules how easy would it be to hack a ledger uh it wouldn't be very easy to be honest I think best would be for you to try physical attacks first like glitching like side Chanel attack if you want to really attack it uh you will need to get a bit more equipped with like a focus y beam one station or a scanning electronic scanning electron microscope so that's going to cost you probably a few Millions uh and as well you are going to be able to deap the

chip so that's going to be a bit risk for you as well uh but you are free to try of course and let us know if you manage to do anything Ledger is building an infrastructure which will provide the best level of security for the Bitcoin industry you two can benefit from this technology and get an affordable secure setup for storing your Bitcoins with The Ledger Nano so go to Ledger wallet.com and use the offer code eb09 to get 10% off your first order that offer code is valid until September 30th of 2015 we'd like to thank Ledger for their continued support of epicenter Bitcoin are there some security implications because is a received as trustworthy as receiving a

transaction so the idea here so the amplifications are so the thing that he wants to avoid is he wants to avoid a situation where let's say Subspace 35 sends something sense a transaction the transaction then that ends up having an effect in Subspace X8 but then as it turns out Subspace 3 that that particular Block in Subspace 35 ends up getting reverted for some reason and then you have an effect Without a Cause so there are two ways of avoiding that one way is the synchronicity approach where you basically end up saying that if Subspace 35 ends up getting deleted then pretty much or reverted then all of it all of the sort of children in what I

call the dependency cone also get reverted so the dependency cone is this idea that you know you have a uh basically the block and then every block that's dependent on the Block every block that's dependent on the block that's dependent on that block and so forth and so you can think of you know a change in one Subspace as sort of eventually propagating to every other Subspace in a kind of butterfly effect so theoretically you can calculate the dependency code and you can sort of revert the entire code the second effect or the second idea is that you basically say that in order for this receipt to be claimed you have to wait until until it gets finalized so there's some point at which you have this concept of finality

where you know in Subspace 35 at some points you agree that this block is never going to get reverted and at that point only after that point will you be able to claim this receipt in another Subspace so look these receipts what they basically are is they're kind of sort of cryptographic documents basically saying here is an event that happened in this one Subspace and then that event might be something like contract 35 some contract agreed to send 10,000 units of uh I don't know e dollar to 68154 then you know later on if let's say you are a user

68154 then what you can do is you can create a transaction that basically pointing hey guys over here in Subspace 35 I have a receipt addressed to me so it's a sort of internal protocol mechanism for like keeping track of these sort of cross substate events that get created I mean it almost reminds a bit as well of the the side chains thata you know of having these proofs in one chain and then using that in the other chain yeah you mean you definitely think of it as being a being a side chain type protocol today's magic word is Frontier f r o n t i e r head over to let's talk bitcoin.com to sign in enter the magic word and claim your part of a listener [Music]

reward [Music] so what are some of the other uh ways to solve scalability so there you wrote a blog post recently about state tree pruning can you explain what that is so state tree pruning is I mean it's not the ultimate solution because you know every node still processes every transaction in stage Tre pruning but the idea is that you know right now one of the problems with with the sort of Merry approach that we took is that you know basically ethereum clients store the state at a full copy of the state after every block at any point in history so you know they store the

entire history they store this you know the balance of every account every storage item of every account at every Point ever so this is so that also sorry to cat you up but that also means like the code of every contract yes okay so that's um I mean it's not as bad as it seems because you know most of the time from one block to the next block this state is almost the same and there's not much extra information that has to be stored but even still it's inefficient because what it means is that because of the way that the the Merle trees are constructed every time even one storage key is updated that's another maybe thousand bytes worth of worth of tree

nodes that have to be stored in the hard drive so the idea with state tree pruning is that well if let's say at some point there there are some nodes that were part of the Merkel tree and then the block a new block happens and those nodes aren't part of the merel tree anymore then at some point after let's say 5,000 blocks those those old merel tree nodes just get deleted so they're not stored anymore so the idea is that you know yes we'll be storing the current state and will'll be storing 5,000 blocks of historical state but we're not going to be going beyond that so that has the benefit that you know it's going to be much more sort of efficient in terms of State size so it

should be able to reduce the state by quite a lot but on the other hand it's also a bit uh I it does well the one issue that introduces is that clients are not going to be able to revert more than 5,000 blocks so if you revert more than 5,000 blocks then basically you're screwed and you have to redownload the whole chain so this is the sort of one vulnerability that you get as a trade-off but you know I think it's an acceptable tradeoff in exchange for cutting down storage cost by factor of 10 to 100 so you know without this on the Olympic test net we got up to somewhere between 10 to 40 gigabytes of sort of total storage State pruning should be able to knock that down by a factor of quite a lot you know

more than definitely more than 10 um the one thing about Olympic though is that there were a lot of people that were kind of deliberately spamming the network with transactions that tried to sort of take up as much storage as possible but that's something that is you know it's not going to happen on on Frontier because in Frontier you know we specifically economically incentivize being sort of economical with your storage and even clearing storage as much as you can so you know with frer there should be almost as much clearing as there is adding and so but I mean so in general stage Tre pruning is just this optimization to sign to save on hard drive space so you know instead of storing 100 gbes you'll

able to store 5 gbt and that's great the things that it doesn't solve are first of all you know the cost of processing transactions it doesn't solve bandwidth issues it's uh and those two things are ultimately going to be what defines these limits of scalability for ethereum okay and I mean but any you know given the potential that ethereum has and sort of the long-term uh you know the long-term Outlook of of eum you know you want to try to keep that down as much as possible and Implement every scalability solution possible so that it is scalable over the long term so one of the things that has been talked about and U which we can go into and now is to

in you know in a subsequent release of ethereum we'll be moving from proof of work to proof of stake um so can you talk about that and what is the relationship there with scalability I don't think proof of stake is a scalability solution um I mean it can improve things slightly because it reduces uh there are ways that you can sort of reduce some of these centralization issues and and say make the system work when say you know every Noe is humming full node is humming along at say 50 or 100% CPU utilization instead of 10% but it's ultimately not going to solve it the thing that proof of stake will do is it'll allow us to provide much more security much more cheaply so you know there's people that sort of always keep on saying that oh

you know you can't use you should only Ed the Bitcoin blockchain because it has 100 times more hash power that makes it the most secure well with proof of stake you theoretically have I what we we can calculate as all being basically the same security margin as Bitcoin and and even faster so the way that the these Proto the way that our current thinking in terms of proof of stake works is it's a sort of multi-round protocol where every single round people basic or the validators basically make bets on which blocks they think are going to end up being in the final chain and people have the ability to make these bets at kind of whatever ratio they want 3:1 9:1 50 to1 100 to1 and you know the theory is that the the higher you bet the the

larger the reward is but the higher you bet you know the much much more you'll lose if a block ends up not being in the main chain so the general approach is that you know there are multiple sort of cidate blocks people end up betting one of them ends up getting more bets than others eventually kind of converges People BET let's say three to one for a block once once you see that every other value bets bet 3 to1 for a block you bet 9 to1 for that block once you see everyone else betting 9 to1 you bet 27 to1 and so forth and eventually one two3 of the people are comfortable betting 10,000 to one that's when you call the block finalized so one of the sort of economic arguments that I've made is just jump in I think that is a that is

absolutely fascinating way of arriving at consensus I have to say that that sounds that sounds brilliant yeah and so one of the uh interesting things as that proof of work to some degree is also a betting game because you know with proof of work basically when you mine a block you are saying you know I am making a bet where I agree that if I mine this block and it gets in then I am going to win 25 Bitcoins but if I money block and it doesn't get in then I am going to lose some number let's say 10 or 15 bitcoins worth of electricity costs so to some degree you know when you're actually when you're actually downloading the longest chain you

actually are literally checking for you know which is the chain that that the majority of people are willing to bet on so you know the advantage of proof of stake here is basically that instead of the sort of amount of money at stake converging linearly it kind of converges exponentially just let me jump in here and ask a little bit more about this so so if if the if the point at which the block is like Chosen and says that's the one it gets in is is when the ratio is 10,000 to one I mean isn't the implication that somebody if they have a lot of financial power could essentially just put that money up and and then get in the idea isn't that anyone has the ability to bet the idea

is that there are these validators and you know these are like validators that put in a whole bunch of money in the proof of stake system and you need and each of them has the rights to vote a particular amount so theoretically you know if you have a lot of money then yes you could take you know you could throw it all in and you could sort of make whatever block you want win but that's basically equivalent to a 51% attack so the amount that that would cost is basically you know more ether than there than there would be deposited in the Stak system in general period okay no no that makes sense so I wanted to ask about 51% tax so it what does a 51% attack look like in in

ethereum in in proof of ethereum or proof of stake ethereum well that's both proof I mean proof of work ethereum is just like Bitcoin you know you just mine your go 100 blocks back mine your own chain your chain is longer and Bam you win 100 blocks to get reverted that's not different at all in proof of stake it's uh so there's two possibilities I mean one possibility is a sort of attack before finality and the other possibilities s of attack after finality so before finality you know theoretic it might be possible for you to sort of slightly bribe all of the voters to sort of switch over from one particular block to another and you know you might be able to have some some amount of

influence there but a you know we don't see we don't sort of treat that as being too much of a problem because well the block is not supposed to be finalized at that point anyway after finality what that basically means is that V these validators basically Stakes their entire security deposits on that particular block being finalized so one thing that could happen is if 2third of the validators end up uh being uh end up basically wanting to create a new Fork then they can say okay well we're going to we already voted on that block but we're also going to vote on this new block and you know when that happens that means that all these validators are going to have to lose their security deposits on the but you know they'll be able to sort of make two chains and

those two chains will look indistinguishable from each other and at that point basically the system goes into this mode with where it kind of uses subjective resolution so you know users would individually look at which shame they saw first and you know eventually sort of a consensus would form around one of them but this is the kind of emergency mode you know the cost of it should be basically economically equivalent to you know buying more as6 than the rest of the Bitcoin Network combined and at that point basically having power the power to sort of screw up with it to whatever extent whatever extent you want in any case and so with double spending uh so in ethereum we have two types of transactions right so we have the monetary transaction of financial nature and then we have uh

contracts uh what what is the risk of double spending on a contract does that look like you can rerun the contract Bic the so the idea basically is that You' be able to revert blocks and when you rever blocks that means that you know if there is some instance where a happens before B then in in the new BL auction you created you might have a situation where B happens before a so one example of a 51 % attack might be is that let's say I register on the ethereum name registry vital. e then you decide o I want I'm an evil Advertiser I want to I want to have that page for myself and I want to create create ads about pornography on on that site then basically what you do is you uh say you

do a 51% attack where you sort of revert all the blocks at the before the after the points where I registered vital. eth then in your own private chain you would register vital. eth for yourself and you know in your chain you would have you would own that domain and you would be able to point the domain to to where whatever you want so you know that's another example would be a lottery so for example or just gambling in general so if you're participating in some kind of gambling application and you lost a big bet you could try to reverse the a few blocks and sort of reach rany if you if you win the next time um so another example might be a market so for example if people are placing orders

then let's say some big event happens and because of that price is suddenly moved by a lot then you might want to sort of revert the blockchain by a bit and be able to claim all of these sort of historical orders that are up prices that are really much much much much higher than they currently should be so like basically every different application has sort of its own different set of things that you could potentially do if you had the ability to magically reverse the watchings time okay but the I guess the idea is that with proof of stake that will be more expensive I I I wanted to I wanted to to Circle back very briefly at one point there so you said in the case of 51% attack uh the attackers so these 2third of the validators that that would

execute that attack would lose their security deposit but how is that possible if the chain is forked wouldn't then they be able at least on that fork that they control to still have that security policit or to roll back to they had it so actually so basically this uh it depends so theoretically they control the chain and they you know they can refuse to accept transactions but I wrote this uh blog post on censorship a few weeks back where basically pointed out that you know if even a sort of a monopoly Coalition of users controls a particular chain and they want to sort of make it make that chain available to

everyone then there are going to be ways to kind of sneak transactions into the chain that ends up having the effect of uh basically you know doing just about anything and one of those things could be destroying security deposits so it's uh so you know they basic the theoretically would be able to but at the cost of like basically you know destroying the functionality of the entire network and at that point the value of their security deposits will be knocked down to zero in any case you're listening to this show in parts thanks to all the support we get from our great sponsors we have some some fantastic companies we work with and we're very proud of what they do and of their products and their companies we happy to stand behind every week we

reach a large audience of thousands and thousands of people deeply involved in the Bitcoin cryptocurrency space people running companies people having a lot of influence and people who are just enthusiastic about technology about progress and about the future so if you're involved in a project or startup and think that reaching our audience could help you email show at epicenter bitcoin.com and let's talk so just very briefly before we move on you know besides the main client right some of the things you guys are working on is whisper and swarm yes what's their role particularly in the context of scalability are are those relative relevant for scalability as well um so and maybe mean whisper I mean

they are real they aren't to blockchain scalability but they definitely are relevance to application scalability because applications definitely should be doing a whole bunch of things off the blockchain so that you know if you don't need consensus on them and just for listeners who aren't aware of that so whisper is the file St no whisper is the a messaging system and decentralized messaging and swarm is a file Source system or the other way around yeah mhm correct um okay so you've briefly mentioned before the topic of of governance and that's actually something we wanted to talk about I think it's something that's really important that we we start covering more because it's a it's a

tricky topic that's sort of under I think under discussed um and and we we're going to have Gavin and on soon to talk about this topic as well so in in this context you guys just added a new uh Foundation board yes can you tell a bit about that and and what's going to be the role of the foundation board in the development of ethereum so I we brought in a new executive director Ming Chan and also this found three new members to the foundation board so and the foundation board you know is going to be the board of the foundation you know the sort of governing group that sort of ultimately makes DEC makes

decisions um they are in the I I mean they basically we are doing the same sort of things that a board of directors would do in any kind of regular and nonprofit organization um and uh I mean there are also other sort of specific focuses you know things like fundraising things like sort of managing institutional Partnerships that are also important okay so so first of all when you talk about them doing something that any other nonprofit Foundation Bo would do well I mean I think normally because we have a sort of a weird tension here now like a foundation

is a is a centralized institution and then it's running this decentralized protocol right so I mean the foundation isn't really running the protocol so to speak I mean right now you know it is the one that's developed the software but it's uh the uh and you know the foundation is going to have sort of a lot of I guess moral authority over it but you know moral Authority is basically all that there is so you know if the foundation ends up acting in some way that's objectionable then people can create sort of a different thing and call it ethereum and you know the theorum ecosystem will ultimately kind of end up following that so and so there's one thing to point out here most of the people on the board are not I mean apart from yourself or not from the crypto

space was that a choice uh or did that just sort of happen because of the people that and what why why did you choose to do that um in part because just because we wanted to sort of help expand ethereum Beyond just the existing crypto space and try to push hard toward you know sort of mainstream institutional adoption um it's you know basically just part of this this whole idea of expanding you know that I've been caring about quite a lot over the over the past half year is really just expanding beyond the sort of small crowd that that's sort of congregating around things like Bitcoin and altcoins and moving it

toward you know opportunities to actually sort of help people on a large scale and doing that necessarily involves having sort of input from people that are from these uh more traditional Industries yeah I mean that definitely does make sense uh now the foundation will have the task of managing the funds that were raised on the crowd sale and so essentially how it works for people who don't know is there's a foundation who who has the funds from the crowd sell and they Pay E Dev was a for-profit company developing the software uh I'm curious like what will that look like in the future like how much money is left from that crowd sale and first of all importantly this idea of eth Def being a for-profit company I

mean technically it's registered as a company but you know it is a subsidiary of the foundation so this isn't this isn't like you know myself Gavin and Jeff have shares and and we're uh and and we're sort of Contracting with the foundation right now it's uh e Dev is basically an arm of the foundation um now in terms of how things are going to go in the future there definitely are sort of a lot of people inside ethereum that are heavily interested in doing their own thing and that could that that'll entail starting a company I think we'll see multiple companies um there's uh you know people are interested in going toward applications some people are interested in sort of staying in staying staying and working

on the core so I mean I think it's not it's going to be impossible to avoid sort of some kind of diaspora effect in that sense but I think ultimately you know which organizations were involved it might not necessarily matter too much because you know we're ultimately all going to be in the same Skype groups we're all going to be developing the same sort of stuff and uh in any case um and in the long term like it's one of the ideas was for the foundation to take on sort of a more I guess advocacy uh and communication focused role but I think we're still deciding the extents to which it'll be that and the extents to which the foundation will sort of directly be running and managing development so then

is the idea that eev continues to be the core developer of the software or or that all these smaller companies spin-off companies then I think the spin-off companies are going to end up doing in increasingly more over time um I think like a good outcome would be something similar to Linux where you know 75% of the code is just written by by companies in the space is uh you know with the foundation right now you know in terms of how much money it has left the foundation's current capital is I believe somewhere on the order of1 and a half to $2 million or so not including the ether now including the ether of course there

is quite a bit more and you know right now nobody knows really what the exact price of ether is um if I check on coinmarketcap.com for ether coin that's uh that's obviously sort of one of the Bell weathers that I guess some some people are looking at but it's uh I think somewhere around currently yeah currently at 280 um there there are exchanges where you have uh bid orders already popping up some of them are going like somewhere around the order of $1 to2 so you know it's uh very hard right now it's hard to say once this sort of tha process happens in 3 or 4 days and we actually start seeing

a large amount of volume then I guess we'll see just how rich the foundation is I mean it could end up sort of surviving for any amount of time between six um between six to8 months and something like and uh I mean if ethereum really goes up then you know forever and but then but then it importants to note that you know ether isn't the only route for the foundation to get funding for continued development so you know there are just sort of plain old donations and sort of bigcoin foundation style memberships you know there's a lot of possibilities and that's one of the important points that you know the board and the directors are actively working toward okay so that that's very

interesting one of the things that I was talking with uh with Taylor Garing at some point and he mentioned the topic of private blockchains and he mentioned that uh that ethereum was specifically set up to support that and and I also saw on I think Sebastian you saw it on one of the pages it mentioned basically how you can say an identifier and basically run sort of your own private ethereum Network what role do you see for that do you think that's going to become a frequent usage or or do you think most yeah I think there's going to be lots of private networks um there I know a lot of groups that are very interested in that concept already and for what app do you think

that these private networks make sense and for what applications do you think people would would choose the main ethereum right Network so I actually have a blog post on this that I'm going to publish in the next couple of days but the general idea is I mean both categories have different advantages so with private networks for example you know it's sort of easier for for whoever is creating the application to sort of have control over it just to have the ability to quickly update things if they needs to reverse transactions and so forth um things are also much you can also have much lower latency faster transaction confirmations you can also have uh potentially uh uh cheaper transactions higher

scalability and higher transaction throughput um now that particular part of the calculation I think is going to change a lot once uh scalability comes in but you know this is the world we live in right now public blockchains they have two benefits one of them is the no trust aspect so you can create applications where even you don't have full control and you know even the developer of the application either has no control or partial control um the second uh thing that's kind of important as a sort of network effects argument so the idea here is let's say take one particular blockchain application let's say domain name transfer escrow so the idea here is that you know there's a in general if I have

a domain name I want to sell the domain name to you then the issue that we would have is that you know either it's a standard counterparty risk problem if I send the money first then you might not send the domain name if you send domain name first then I might not send the money and in general we have entities on the internet that work to sort of to solve this counterparty risk problem that are just escore providers but they charge 36% now blockchain based solution is if you have domain names on a blockchain and if you have money on a blockchain then you can send your domain name into a contract which automatically hands it over to the first person that provides the money so the trade becomes Atomic it either happens or it doesn't happen and

you know with that you can basically cut the ESC cost down to 3 cents and that's a really cool application and it's wonderful but one of the important point is that in order for that application to actually work you need to have domain names and uh currency on the same public blockchain or on the same block but you know ultimately it's going to be very hard to sort of I think for people to agree on a sort of private blockchain for everything because that would just involve so many Industries so I think the public approach is going to win out for that sort of stuff yeah that makes perfect sense I think that's going to be very interesting to see how that goes and I'm sure there will be a lots of uh sort of interoperability as well where maybe uh private blockchains will be

interoperable with with the public blockchain um I I wanted to Circle a little bit back to this governance topic especially because it's been such a you know because of the block size Tob mainly I'm sure you've been watching that as well and you know before ethereum you were deeply involved in in Bitcoin for a long time uh what are the main takeaways that you've learned from watching the block size debate when it comes to ethereum right so there's two issues there one of them is a sort of object level issue of well you know well how do you regulate the block size and the thing is that you know with Bitcoin has always been the choice always been between sort of one megabyte and 8

megabytes and it hasn't really Advanced much beyond that but you know with e theum we've always sort of been not thinking not about numbers but about mechanisms so you know originally the thought was fixed value then the thought switched over to uh this uh sort of exponential moving average approach now I'm switching over to this uh uncal rate targeting approach and I think that the sort of choosing institutions instead of choosing results in general is a kind of more robust way of of doing governance but the sort of metal level concern is that you know in general how do these decisions get chosen right so let me just jump in because you guys you can maybe have those discussions and say oh

we we're going to talk about institutions and mechanism not about end result because you have much more control over that whole discussion in Bitcoin nobody has that control I mean you can go in there and say hey we need to talk about the process not the result but that doesn't mean anybody's going to listen to you and I agree I think that yeah I mean ethereum does but does benefit from from a sort of Greater degree of development centralization and sort of just uh I guess moral centralization to some degree and you know that does have benefits and and do you think that's something that because of exactly that you would want to maintain in some extent um I think that in the long term something you know that's sort of highly decentralized is necessary but I think

like these are this is the kind of phase where having some group of people that sort of has a sort of the sort of shelling points of Greater moral Authority definitely can help sort of move the community forward in One Direction I think I mean I think it's kind of like the frontier canaries I think it's a training wheel it's useful now and it's it's going to get less and less useful over time and at some points you know it'll switch over to the points where making it as distributed as possible is the correct thing yeah of course like one of our listeners pointed this out you could have that built into the Genesis block you could have the governance model built in yeah why was that not uh done for ethereum was that something was success it was just something that that we for a water

of those things we just didn't have enough time to think of them and we wanted to just keep Frontier simple and is that something that could be implemented in the future or is it too late now that it's been released um I mean something it could be done so one of our ideas for 2.0 is basically to have this idea where people can vote on changing the code and even having like as much of the code as possible be inside of the virtual machine language so all the changes could literally be sort of processed just automatically on the consensus level okay cool um so before we come to an end here we're sort of at the end of the technical discussion um so looking back on the last two years and you've you've come a long way from

writing the white paper to now releasing uh ethereum you know can you give us sort of an an idea of what you've been through and if you would have done anything differently what would that that have been uh there's a lot of things that I would have done differently I mean first of all I think just the way the project formed was just incredibly non-standard like most sort of startup like things tends to be oh hey you know here are a few of us we're friends and we have a good idea let's work together and make it whereas ethereum was hey I have a white paper I'm going to send it to 20 of the people that I just happens to know in the last one or two years and you know the first 20 people to respond back on an email or Skype just ended up being co-founders so

that was a sort of approach that's highly not conventional ended up having its costs and I mean there definitely were sort of moments internally where the thing came it looks like it was coming close close to pulling apart on a couple of instances but you know ultimately we went we got through can can you elaborate a little bit like what were some of the instances that that this approach almost made this project fail um there are think a lot of sort of cultural misalignments um especially sort of between people that are kind of more like developers people who are more like not developers um I think that the thing is that in general you know just about just about anybody anyone can be

can sort of theoretically be a business development person and that there's a it's whereas you know with developers it's so you know with coding it's something where a lot of people just it's obvious that if someone can't do it at all and if someone can do it that means they can do it at least decently so figuring out sort of really how to structure the sort of non non-technical side of the team I think is important because like non-technical people I think are extremely crucial but you know the difference between people that can really push push a project for forward and people that just sort of just sort of hang around is uh I think a big one um there's uh I just on the cultural side I think all people had all

all these sort of different Visions but I mean that's inevitable to some degree but it's uh something that really sort of needs to be handled well um and there are sort of on the technical side I think we sort of ended up doing one of the problem that we kept on having over and over again is we kept on have thinking that the project is going to launch in uh two or three months and so we ended up sort of TR continuing to try to keep all these sort of clients up and lock step with each other and that ended up slowing development down by a lot and you know we ended up trying really really hard to get consensus between C++ go and python on features that don't even exist anymore so that

was something that you know if we had thought carefully on that we could have easily think cut the development down by at least 3 or six months um in terms of just costs I mean there were a lot of costs that we had early on that we didn't need to have you know the sort of $1.7 million that we spent even before the crowd sale I think obviously could have been greatly optimized um I mean to some and to some degree this is all kind of firsttime major business experience both for myself and for a lot of other people on the project but I think that you know from you know from here we've learned a lot of lessons and we're and foundation's going to be quite a bit more more sort of efficient and focused

going forward um in terms of yeah I mean from a technical standpoint going back to this multi client issue I think and once again opinions differ but the approach that I would have preferred go is to have maybe stick to two clients and have one client not even bother with networking and just sort of interpret the blockchain at least right up until maybe two months before launch and only at the sort of two months before launch Point actually sort of start bringing the networking up into consensus with each other um proof of work I think we personally think we should have just stuck with sha three because as cool as the E6 resistance Pro work algorithm is it's the and you know

as much as it's uh I think it's nice that we have it now we the time that we lost I don't think was worth it especially given that proof of work is only going to exist for about a year so you know knowing what I know now I would have just sort of stuck with SH for the proof of work and went straight to proof of stake after that um cool there are there are a few other protocol changes that were kind of like that as well you know these these are a lot of things there's another topic we we briefly want to come to so ethereum of course you guys are using this notion now or this term of a world computer which which I really like I think it's a good metaphor and um one of

the characteristics or features of such a world computer is that it's enormously powerful know one could assuming everything works well you know one could do great positive things and this is right right now this is the world smartphone computer in two or three years when we have scalability it's going to be the world super right right now it's a very very shitty sort of world computer that can't do any transactions but uh you know presum assuming you know it works out it will it can be very powerful right and of course people will presumably do many great things with it but also uh bad things with it so how do you feel about that is there something you worry about uh bad things yeah I mean people

are it's definitely going to happen and at this point we kind of don't really don't really know what's what's going to happen to some degree just because it's uh it's not even La yet we haven't even honestly we haven't even seen any activities in the sort of community like I've seen absolutely no nobody even suggests the idea of know even I mean even even something that's sort of on the sort of borderline controversial like so grode on ethereum like no one's even even brought that up and that's that's kind of I think from me sort of PR standpoint it's definitely a positive sign um yeah I mean I think more talking also

the long term right recently yeah in the long term definitely I mean I think if if you think about it's sort of inevitable like any you can't you can't blame the software uh you know you can definitely do harm with C++ or Linux or any type of software operating system what have you um but there is potential there because it is quite gamechanging since there's no central point of failure Etc like you know if we look at this on a more high level know like recently this is sort of uh topical because recently Stephen Hawkings Elon Musk I think Steve wsnc wrote this open letter um calling for the ban of autonomous weapons is you know in the context of ethereum is is

this something that you can see is potentially in the future you know powering really really bad technology or harmful technology for Humanity um I mean if a theum gets any that I'm sure it'll end up end up powering really really bad technology just because you know things that get really big all are going to always end up having some good and some bad in them it's uh not to give the sort of corly cliched standard response but it's just like the internet in that sense um and and what's your view in this context just on AI in general like strong AI this is something you think about oh AI is definitely a problem I mean to some degree it's orthogonal to the blockchain

issue just because you know here we don't even have sort of anything close to Smart AIS yet but it's uh I mean this sort of idea that at some point when AI becomes slightly more intelligent than humans is going to become really good at optimizing itself and become even more intelligent even more and more intelligent become and become this sort of Singularity thing that you know depending on whether or not we program it well is either going to make our lives SP really nice or it's just going to fill the universe with paper clips and choke everyone that's uh yeah I mean that's defin that's definitely a concern and I definitely applaud everyone who's so working on solving that problem to what extent that ties together with the what with walk hand issues I'm I'm I'm not sure because I mean the thing is it's almost a sort

of completely different category of stuff like the sort of decentralized technology is they enable all these sort of applications that are unstoppable and that sort of allow people to create these kind of complex patterns of interaction that cannot be prevented and you know yes there definitely are sort of these patterns of interaction that are going to going to end up leading to bad things but ultimately you know at the bottom of it all it's still humans doing things to doing things with humans to humans so it's to some degree a bit more mundane in that sense so where do you see your role in ethereum going forward I know you don't plan very far ahead but is it something

that you see work you see yourself working on for you know uh next months or 10 years or are there other areas you're interested if the E if the ethereum ecosystem Contin you know continues being a being the sort of viable viable and growing thing which you know all indications point out that that it really is then I mean I'm defin I'm definitely going to be in the ecosystem for quite a long time um in the ecosystem in what capacity I'm not sure yet I mean that's still something I have to decide for myself and are there any other areas uh maybe not directly related to you know decentralized technologies that you're interested in or the places you'd see

yourself going towards um hm I in cryptography economics um all this sort of I mean this whole sort of rationalist community is always sort of interesting and worth following at this point you know yeah I mean I follow a bunch of different things but you know as i' as I've said I don't really plan too far ahead okay well uh this has been a really interesting conversation um I I want I want to say like I installed the the client and I encourage every anybody who uh who who wants to you know get their hands dirty and give it a try to

go to ethereum.org I have no IDE who runs aum.com uh certainly someone who's trying to make a lot of money on domain reselling uh and uh so one thing that you you need to know is when you install the client you know you have I did it on Mac it's it's it's quite simple to do but um you know you do have to have either Mac or PC or Linux to do it uh and yeah no free BSD installations yet what's that no free BSD installations yet I think and you have to generate the Genesis block um and then you can launch a node and start mining Etc and in a few days hopefully start writing contracts

mhm yeah so vitalic thanks so much for coming on and I think it's it's super exciting that etherum is finally launch and hopefully we can revisit the topic in you know a few months six months or so uh when we we'll actually see some application see some traction have some more like proof and and the testing and in sort of a real life context of what's possible with this how it works yeah definitely I'm really looking forward to following that process I'm definitely looking forward to it myself yeah absolutely so we at the end of our show so uh one last thing before we wrap up actually two last things but one is that we've started doing this

this H contest which is bribing people for leaving reviews on iTunes now uh we do a draw once a week and one person wins uh a t-shirt regardless what they write it can be very negative or it can be very very wonderful and full of Praises both are are fully acceptable so if you want that please do that and send us an email with show ad AB send a Bitcoin so we can know what the review was and also one last thing we're redesigning our website and redoing our website it's been on our on our plate for a long time and uh now we finally need we really need to do it and so we're looking for a WordPress developer

so anybody uh who knows WordPress who knows how to do HTML CSS you know who likes to use gulp uh all these web Technologies uh should get in touch with us and um and hopefully help us build our new website uh thanks so much for joining us so we put out uh new episodes of ENT of Bitcoin every Monday you can subscribe to show on iTunes or get it through your podcast app on Android uh or iOS and of course you can also get it on you YouTube watch the videos on youtube.com/ Bitcoin and thank you so much we look forward to seeing you next [Music] week