Brian G: Mining & the Role of Silicon (or why ASIC's are good) | ETHDenver

and he was obsessed with winter lager and the reason you wanted to win the lottery is because he wanted to fund the DA of the perfect yeah that wasn't going and so he was very devout and every day said god oh god please let him hit the lottery please let it win the lottery I'm gonna do this dao i'm going to change the world and friday came along and he didn't win but he was undeterred it was there about and he just redoubled his efforts and he said God God please let me win the lobby please I'm going to do this beautiful dao I'm going to solve world hunger I'm gonna fix world peace I'm going to solve everything in this dao never be happy friday men come around lost it was not her you just started praying in the morning Andy god don't please let me win the lottery I will name the DAO after you it will be the Yahweh here Dao it will never be happy would be the most secure dao ever it will be sports and forth and everything else Friday and he didn't win the lottery and he looks up and he goes oh god why do you not let me win the lottery big boys country heaven I miss yes Lord man's you got to meet me halfway by ticket ok the in the vacuum of the discussion good what I'm going to talk to you today is a little bit about hiding a little bit about the silicon Alex use and I hope that we'll walk away thinking that maybe the word ASIC in a 6 and minus look at that rising spot real quick I think get the great job dinner Doug do something and then in the city collector industry longer than I will buy tune that I am oh that was a known role when they the cell phone I've built chips and just about everything that you could use chips and also in my most recent job sold to some small companies that build Bitcoin asics so you know what i'll talk to you guys about today i have a little bit of all the world inside and then just by the way when i'm working on encouraging my current patches intersection to buy archean blockchain technology so if anybody has a similar passion please see me after the presentation so what's on tap I want to talk a little bit about silicon a little primer I'm going to talk about Moore's law I'll overview the mining proof of work that is used in Bitcoin and that is currently using a theory I'll talk a little bit about how various chips are used today from the CPUs that are in your computer to the GPUs that might be in your gaming computer field programmable gate arrays you might not be familiar with they were in room 10 Bitcoin and then finally a six I want to talk a little bit about what the basic companies look like and i'll talk about can we do something can we build in a sick for a theory or maybe more importantly is there something we could do in my profit in evolve morning first of all i want to get the nomenclature down a little bit I personally can use to Silicon Valley every other week so I'm in Denver and I'm in Silicon Valley and before this I was in Austin so sort of all the hot high-tech spots of places that I've been so silicon valley silicon is what you might hear silica is what you build semiconductors out of semiconductor industry so all ships are built out of sending you doctor you might hear the word chip I'll probably use to work chip but unfortunately sometimes I'll introduce myself and I'm say I'm in chips and they'll go buy salted frito-lay no sorry shipping if you might also heard the word IC integrated circuit asic we talked about basic technically stands for application-specific I see it's a little bit of a day to learn in the industry SOC is in other words you might hear which is system-on-a-chip so increasingly because you can put so many transistors on a chip things that used to be in multiple chips are on a board or entire system or now I chew especially if you look at things in storage if you look at some of the modern servers things that were all over the board or now in so if we talked a little bit about ncu is another term they're sort of a smaller micro processor microcontroller unit is no termina right here so Moore's Law any of you guys heard of Moore's law okay great so Laura's law technically says that every 18 months or two years you put twice as many transistors on a chip as you could but that also means it's of a hundred percent productivity every two years and that's been going on for over 50 years so that's huge I mean you can't think of another it's really hard to think of another discipline where a hundred percent productivity been going on for this long so Gordon war have sort of theorized this sort of popular culture thing in his 60s he was one of the founders of Intel if you think about 50 years that means the chipster of 32 billion times more powerful cheaper and lower power in many ways that's the foundation for everything we did I mean you could have built a computer without a really intelligent chip we could have built smart phones you could even networking you have really huge chips there at the heart of every router that bring in you know packets in 14 and out the other are able to look inside of it needed and figure out where he owes a lot of the games and automotive are really the result of silicon you know things that you can do in the engine now you can really take data you hear about things like huge day the data analytics there in the car it's pretty amazing and even high level software it would be impossible for us to write code at the level then we write out today if there were really powerful underlying chips no I think this is Gordon board about 1965 if it really rocks this look it could be walking within four dead this is a lot of I eat the next day you know we might go number which is tacos she does classes like this it's a google oh so this is a curt ray kurzweil graph you know just kind of showing not only things ins in the IC age but all the way back to when you know we had a few computers that were built mechanically so you know what's interesting to kind of look you can look at some of the pinkie on some of the IBM chips some of these graphics chips from Nvidia and AMD slashing ati really powerful because oh-ho-ho or the grab but i thought two things were interesting it's grabs a little bit dated sometime last year we passed it we can build a chip which is more powerful than a rodent brain sometime in 2023 will be able to build a chip which is more powerful than the human brain and hopefully I live around for 220 45 will be able to build a single chip which is more powerful than all human brains on earth so a lot of really cool stuff up here you know a lot of this deep learning and artificial intelligence you know you're really going to need this kind of power but you know cut open seems pretty cool some pretty cool new things is and definitely this part of the curve you know from you know almost nothing really help adjusting things that we you know appreciate it 2045 is mom's garden Skynet out the skin so you can think of it like that I don't you just got to keep the brain away from the hand that's the cute you know I babysat for my granddaughter this morning she's sister she's three months old and you know she needs her brain moved away from her hand she's just keeps banging herself with her rings so that's we got to do we can't let the computers have the nukes so lastly just how do we do this you know how do you have you keep up with a hundred percent productivity every year and basically the shrink so you shrink the feature sucks so you're able to put more transistors on a way for each time today or a main extreme is about 16 an inator and just beginning to work on se7en anime so that means a transistor is seven nanometers wide so you can appreciate that a human hair is about 90,000 a light wave is 40 minute 400 there meters UV light is closer to 10 this is important as the way you make semiconductors as you shine light on you shine light through a maps onto the wafer and you take off the part that has a lot of work vice versa to finding out there so you have to really be aware of the wavelengths of light you have to do a lot of collection for Feud diffusions and things like that and the amount of technology here is just staggering a DNA molecules diameter is 2.5 so we're really small now and you did that today to build a new way for about 10 billion dollars which is a lot putting a lot of really good NFL stadiums maybe we can even name ours mile high field instead of whatever it will dance in tens of billion dollars of our d and I think we use now the entire periodic anymore or maybe the entire periodic table except one help so the amount of scientists involved in Sydney conductors it's just it's mind-blowing and even to build a single chip okay now could take 50 million dollars maybe a hundred of the chip does he have enough to build something that's competitive with somebody else you know if you can put this mini this much stuff on a chip if you don't do it your competitor will so even things that you think to yourself or your marketing department says well these two things don't belong together well you got to make them belong together or else should be out so the tools are staggeringly expensive the amount of IP or little pieces i mean the IP business is big and then you know the amount of engineering to design that much to verify that's one of the big things look good it just had a quick question for what you're describing for these types of semiconductors are these what underwire barman computing they don't get so quantum computing is still kind of a new idea it would be something different from a transistor so there's there's probably a regime that's not silicon that will take over you know when i started in nineteen eighty one silicon was on is over the new regimes gonna take over you know we keep delayed but yeah that part of the curve that we showed where the computers will have on the even brain it's going to involve something different from this something smaller than silicon something smaller than DNA or quantum where you can have multiple states and IBM starting to tilt to do the debut quantum so it seemed a long way often it doesn't seem that long enough to be now ya think about the 3d architecture instead of building up building smaller in a horizontal direction but actually stacking chips to work so I'm going to show some of that light come well hold off we'll talk about really good what business are you I'm future nice nice digital 3d stuff so I get the pictures I'll show there so now we switch over to talk about mining and we've sort of done the info Marco so let's talk about proof of work I'll try to keep this simple if I can basically what Ruth of work is is there's a minor that has a candidate block they want to add to the block chain and they've got to win that right in the way that when that right is they they create they come up with a random knots sort of a very large random number they run some kind of hashing function that we're going to talk about and they get another really big number and if that really big number is below some difficulty threshold which means lots of zeros in the beginning that they win they're able to put that block on now they're going to change not one not two knots in the reality is n is going to be a number on the order of a hundred thousand trillion and so that's proof of work obviously the faster you can do this the more likely are you're going to win the right at that block and typically in most crypto currencies the right at that block will have a with the associated some money reward of newly minted coins or transaction piece together I will talk about a lot about why there's a proof-of-work part of the consensus process it helps control the block time so if you increase the biggest one more difficult they've all been altered by that if you make it more difficult by reducing the output you just think it would difficult in receive us so you know you have a table in a haystack human vacation exerciser denver then you make the haystack the size that the atoms in the universe and hunting these numbers all of these numbers how do you need a Bitcoin addresses the unit 256 bits are probably larger than there are atoms in the universe by many many many times so that's how if the blocks code to quickly then you put another zero on you and they space back out makes sense so that controls the block I would also make sure that there's only to be about one or two people that are ready to put block together at once you can imagine if you didn't have something like this a hundred thousand people would all say I'm ready for a block and would be really hard to negotiate who gets the next block this kind of makes it spaced out in time a lot of time for a Bitcoin the less contrary theory this is one of the things that worry about Lee creates a little more chaos if you know everybody set the teacher's desk at the same time it also means there's an element of chance so everyone could theoretically participate obviously if you're a line here that has 100,000 square feet in Shanghai your chances are higher than if you guys you in your basement but theoretically everyone has a chance and that you know allows some amount of democracy or less to game the system it also means that the miners have to expend money so they have a real interest they have to spin electricity after by and then theoretically there's a whole bunch of things and as you guys well know in the blockchain world there's a lot of people to have a interesting game theory in it's fun to talk about game here in I'm not really sure how much how'd you make this so let's talk about hash functions because that's all about mine obviously if you can do this hash function with a small chip thousands of times faster than anyone else you're going to make me look like the hash function that bitcoins unit uses as a short 256 and then Empyrean has a colony a sure they also call up the decoration I will talk about that so what is hashing at the high level ashing it's a one-way function so what that means is you can't go that I can take something any size and a hash it to something that's 256 and you have absolutely no way of having any idea looking at that hash but it came from it in fact if you were to send me you know the novel morning geeks and we hashed it to 256 bit 6 bit number if you changed one letter from a small egg to a capital pay the number would be completely and utterly different you would not have any idea that's true so that's the that's the creek that's best attract ography in a lot of the you know cryptography behind what we do in all the crypto currencies really is related to things like hashing and also things like in friction there roughly you know kind of related so again any size data becomes 256 so it doesn't really matter how what the blog looks like it's always going to have to 256 to UM and interestingly it involves operations at the bit level basically if you guys understand boo anyway logic exclusive oars and a modulo edition which means you you throw away unique carry and also shifts so if you would look at this that's what these are these are modulo adds these risk loosing Wars these are shifts this kind of explains at a higher level how you create how you hash and it turns out these operations or what ships are good at but that's the base of every chip you know if I was to write RTL write code basically at the bottom line would be gauge the new hands or shifts that's the whole thing so you take these really low level things and build these really high level c CPUs but unfortunately when you run the high-level code it doesn't really take advantage to grab a those look like things if that makes sense if we looked at the etherium hash of dagger Hashimoto I think Hashimoto had some idea and I think the colic added that dagger because this is a directed acyclic graph which I don't know why you don't have to put fancy words behind simple concepts but there's a whole lot of people it's like every time you understand look at the Chiefs and words oh we call those you know and ons before and how their catches so basically what that does is it moves unpredictably new number so what they're trying to do you know what they wanted to do is try to build something that may be of CPU is good at or possibly a GPU but maybe an ASIC wasn't as good because the bitcoin thing basics are like money that it's really easy to build Bitcoin a sit this will be hard so again basically it jumps you know don't try to get lucky you feel you're about a gigs worth gigabytes worth of memory and then as you do the hash you're moving unpredictably through memory so you grab in different memory places and you know that's harder for an easy to do harder to do anything weird and then as you go through here you're hatching today and you're mixing it together so you're going to end up with something that's a fixed-width as you go through it and he go through it about 256 times did you use an exclusive alors about multiplication by prime things like that so that's what's there today um it's interesting to me you know I'm not hardcore coveteur to haven't spent a lot of time but when I look at what I think the code is in there for for doing a theory minor it doesn't seem to be very optimized it doesn't seem like people have spent anywhere near the amount of time trying to do my knee better and that's been done in Bitcoin so probably that's intuitive because you know only six months ago a theory mather was only worth dollar or less there wasn't as much money in mining and we've always had the promise that proof of snake was going to obsolete proof of work in a short amount of time but now that neither is a 10 or even 20 there's probably ten billion dollars a year in mining and just like ken says there's a good chance to proof of stake is likely to get delayed by kind of years there might be 10 or 20 million dollars here you know you wouldn't get all of it but you know something to put their mind to it could get a couple of years and I think that's kind of interesting and importing into the industry to build that mining software so I'm going to talk a little bit about how chips are you so CPUs you know what those are there in a computer most people that would a graphics card is there's something called programmable logic which is in the middle so I seen you use going to be something like a pentium it's probably going to have a few chords on it more force a force whatever on sometimes in cell phones they put they put eight cores and cell phones because aids a very lucky number of Chinese culture things my stereo so they race to get to eight cores even though it's probably only to do anything ever time because it's good and then the guys that do the course normally called are more British company they actually built big and small code so that way you can get your extra four cores without spending a lot of time and money and give for that lucky hey there's cash is their system memory so there's a whole lot of interesting things to to you know as you're running a program and you realize that my god this program is doing something that's not in cash or not in the first one cash I can get to assistant memory more quickly a lot of effort is put into that it doesn't help either the two minds at all so a whole bunch of the magic that's put in cpu-z the GPUs to make memory more efficient just completely unused that you might be counter productive so again you run general general purpose computing you run things at a high level so that means you're not really taking advantage of the low level stuff like you should you're running a harlot abstraction if you're smart enough to write the software you might be able to run into you it here the course especially Bitcoin so if you're clever and you multiply the software which isn't very easy to do you might be able to build a better fill deadline you strap if you go to the GPUs so this is Anna graphics card your latest graphics processor from Nvidia or AMD to the smaller pti probably has over a thousand groups now they're really specialized kors made you to do graphics but they're also pretty good at Bitcoin remind me ok atty theoretical so you know you have really hot band with memory that might help you a little bit in Bitcoin you can run hashes in parallel that helps you off so this is going to be a better way of doing my certainly was a better way to do not eat big coin and it probably is in the theory but i'm not positive it's up optimize one of the things I've done in the previous life is a soul that ending the design the controllers that control the memory so I have a lot of insight and how memory works in some cases I know what you'd like to do with the low-level it's not clear to me whether you know how you make that work than I 3d so the latest to generate the latest generation of chips from Nvidia and AMD have what you call 3d so they actually take the chip the graphics chip and they hook it to four different midst accent memories and then these stacks and memories have a bunch of different memory chip center so this memory is a la carte man a lot like amazingly more it's also 100,000 and 24 bits wide instead of 32 bits and that's one of the things you can do is you stack them they're stacked together like this they actually bore holes through the silicon to create what they call up tsv SB is to connect all the way down the stack so it's another it's a enormous amount of technology it still probably pales in comparison to what we need to do but it does a lot of it like you said dulha a lot of apartments in it princess but I don't think any of that's been taking advantage of in theory I don't even think they are aware of the fact that the width of the memory is now tension force of the page sizes bigger so some of the specs and some of the code looks like it's stuck in time but this is cool stuff you know it's definitely there working in fact this whole thing right here is actually another silicon chip a much cheaper version I called an inner publisher and in a hill connect everything together so this technology is being used in graphics it'll probably be it is and will continue I think to be used in cell phone so some things that are super high volume where you can afford to spend billions of dollars or at least hundreds and millions of dollars or tooling costs or you know be an artery or song costs you'll start to see that nil allow you to build you know your your watch there's no camel activity in the future can be a computer of the cell phone with you know who knows what it will really allow another level of miniaturization if we want that now Phil programmable Gator wedge array is another thing you probably don't know a lot of that you heard of the company's alterra which is bought by intel recently in silence they build these things now what they do is they have three different components that logic blocks they have interconnect they have ahead of us so what you can do there is you can take those low level operations those exclusive or those hands and you can actually program them into the chip and now the chip will actually do hashing it are really low level and allow it to be much much or faster in much lower power so you can program the boolean operations directly in there it's a lot more efficient than either a CPU or GPU once its program that's all I can do it's not going to be like your computer to only be able to do one thing and that's it but you can reprogram it and more importantly you can go down to the store and buy all the silence you know development kit for fifty dollars or five hundred dollars nothing like several million dollars it takes to build a nation and I think butterfly labs is the one that got a pretty badly think they were one of the first people to do this this is an obvious thing to do trouble is the second you realize this is a good idea you realize the chip is away so whatever you can program in gates at an fpga level you can put at least 10 times that much into an actual chip if you're willing to spend the money they got spend a lot more upfront money to do so you can put a lot more in there a lot denser you can also run these things a lot way below step so these guys would call us up and go hey dude we're we're your PLL running like 26 bolts we go you realize the spec is 29 so you're like way below this and they're like hey you know if we get a two-thirds yield we're super happy so you know if you go from point nine to somewhere in the point six that's fifty percent less time these powers is where when you mean you look what is that in the other text the only trouble with milder the chip is every chip doesn't work and then you don't need a lot of funding you don't need a whole lot of funding because again you need a lot of engineers and then I peas are extra things you need to add to the gist so in my case you happen to be able to talk that chip a gig or two years in company I'm with right now so far so we would sell that piece they would look at usana children that we give me fifty thousand dollars so i get the feeling that probably what it was as I yeah that's one of those super computing it does mean intellectual property but it what it really means for some reason and goes way back in time it means you're building a chip you don't want to build everything off the chip office I don't want to have your engineers building only the things that they're good at you know that this hash you think does need a clock you know you might run unless there's a clock that's telling hash hash hash hash hash in the clock you need to get from the third party then on the third part now you want we call it intellectual properties you want to have certain IP rights to be able to use it to be able to move connection so it's one of those horrible terrible words thats wanted my industry you know and it's like oh my god so instantly we have to have a much longer legal contract because we have to have all these IP rights so that gave I'm just selling you a piece of the gym you know it's just like you want to go if you're a Tesla you need to buy tires for somebody you do need to have some kind of right to use that time there's some intellectual property or contractual rights have to come along if you don't call so it just means the trouble is on a lot of these chips you put together you've gotta buy so much stuff from so many third parties you might spend ten million dollars off that stuff here as it turns out it'll be harder than anything maybe like one piece of IP you need a standard cell library 50s senior good now the dude one thing you do have to do is you have to build it on the bleeding edge and the trouble is if you go in your competitor will so these guys are like they were using 16 nanometer technology two years ago and then we're going to the fabs and saying you know what the tabs are saying we're not ready the technology is not completely ready we can't guarantee your yields we can't guarantee anything and they go we don't care we're in a higher risk deal will about 25 watts up front we'll spend 2 million dollars on master 1 billion dollars on the mast or maybe you got a good deal in will order you know 2 million dollars of parts and if it doesn't work we understand now that's actually good for the facts because I got they got pipe clean this technology so it's always good for a factory to have someone who's willing to say hey I'll go in your factory first and I know there might be some problems you know like the first person that comes into theory I know I might lose 50 million dollars but we got a pipe clean and so it's good to have that are willing to take the risk is if it works they make several million dollars all people of coin carrot one year they didn't have the IPO cash yeah that's it sperry competitive if you're the second-best guy you're nobody right i mean if there's another chip that comes out and the difficulty goes up and now it costs more money than electricity to run your your mana creek you throw it out right you're not you can't use it so it's super competitive but that it did breathe a fair amount of entrepreneurial life into our industry that you can imagine something that's so capital intensive it's really hard to get funding and unfortunately some of the exits haven't been good with pets and consolidation so you go to disease and you say hey can we get like 75 million dollars and I'm going to try to put together like the best engineers in the world if possible and I'm going to build this thing and you know with the company successful exits room 200 million their life that ends in that's three million dollars an app this week for next Friday what I feel and Robbie so you know this things come and go but this was good this really grieve some entrepreneurial I so that's one of the things i want to leave you with from the industry is that you know some engineers that took a risk made some money the fats got a chance to to know pipe clean some cutting-edge technology we saw some startups which NC especially real 2008-2009 years eve about this so that's cool and then you know if we don't have an asic so I'm not a finance guy but what I do remember from AP code 101 is that basically capitalism is going to operate it about a net zero so if mining produces X amount of dollars miners are going to spin on a grid exit that amount of dollars on that it's just a matter what they're going to spend so without a six the quest is just going to be where can I get the cheapest electricity where can I get the lowest cost harder we're going to get the cheapest labor and so you're going to end up with some sweatshop mining deal in China next to a dam or something in Iceland work electricity is free at least here we're spending money on brain power on to the newer ship or pipe quick cleanings and cutting-edge manufacturing and one other good thing is you got some people that are working on really good silicon encryption or hatching stuff if some of that can be used it on a places right so you know this I can call you guys know that one of the things about cryptos and really spurred a resurgence and cryptographic research because it was sort of a kind of it that feel righteous not the kind of thing if you were professor and said on the probably you get away from you how they like holy cow you're like a rock star and so that's good because we're really going to need that everywhere right we're gonna if we do the idea of things and we have technology that's really close to home direct your personal security we're going to need you know cryptographic stuff that goes into it was like mind-blowing only be put on the big computer for 50 years ago now it's gotta go into your your toaster some sensor before for your fitted right so that's good so I think as long it needs it happen there I'm proud to why this down the last one so if anybody's interested in you know trying to make money in theory and I think it's about a 10 million dollar a year in mining positive about that sounds really understand exactly the gas and fees and chemicals and everything since I'm about 10 billion dollars doesn't look to me like there's nearly with people that have been spending time in it like I said maybe because either as a recent and because goodness sake is plan I think GPUs could have a big advantage they do have very high bandwidth memory the trouble with the algorithm that the talk and the guys come up with is it really doesn't favor high bandwidth memory fingers high late so not to get you to board but the difference between latency and bandwidth latency is like I want a piece of data from this address okay here it comes Texican in unfortunately latency hasn't increased to New out of 15 years what then with needs is I need a graph extreme starting now and it keeps on coming out and so it's really hot band so as long as you know where you want to go next you can have data coming out of the pipe and it's super enormous rate and that's what you care about in graphics the graphics you really don't care about lady driving your turn your you know when you turn the TV or you go to another channel if it thinks another 10 milliseconds to start you don't care but if it's not hot band with enough then you're going to get all it pixley stuff that you get on TV set so it's not clear that will help but doesn't look like to me there's probably one it can be done there it doesn't look like people to put a lot of investment in that cannon ACV bill yes I promise you no matter what you can think of I can get a team together they can deal with a chance that's what we do for a living can it be done cost-effectively that's always the question you would need to build someone really on your footprint you want to build some cool memory controllers with you things apparently you want to look for the lowest memory lane so you might want to do the 3d things that you talked about which is skin lily and a half one but it could be done not as a russian blonde so we're the one that highly touted features or benefits of a theory Michael to you my name is it's a secret system to your deadline easy going to do with this idea that it's almost as I'm being worth of democratic way of my anything it was sort of design with the intent about having these centralized server farms in China but you're saying that the asics actually can't be built oh yeah special I'm from me it will use about memory card type aspect instead of sure hashing power cpu power of it you're saying the same thing could be applied there's nothing you can think of that someone can build an ASIC to do I stink split insane as it was going to be a he have a pulse of fact right so ASMR here is 10 million dollars a year if your hero developing basic and it's a fake resistant it's you're going to spend more than 10 million dollars trying to get out on a sec even he's done just what's the profit there's no competent yeah you might be able to i would say if i was gonna spend my money i would probably do something here but you know you're naive to say that it's going to be democratic and everyone's going to have their own little computer at home it's going to be good i mean there's money in it people are going to figure out how to do it you know i send it yeah they're going to go with electricity cheap they'll go to Iceland go go where labor is cheap go seen those mining operations and the outskirts of the outskirts of the outskirts of the outskirts of Shanghai they look like a freaking labor King you know and they'll do that I mean we were crazy to think just I think the trouble is six months ago as a million dollars here in pairs now it's 10 million but you know like you said if proved a steak is just over the horizon no one's going to know but if poop estate goes out two more years the etherium rises from 10 to 40 is 40 million dollars a year did you know I think people want a piece of 40 million dollars unless it's easier to hide if you can hack it down at 50 million yeah what do you want this is how I work you broke that it to program GPUs there are really tricky there's this there's an open CL which is not probably good enough now if anybody wants to take a run at this I'll help I have really good contacts in nvidia name so i can get to anybody and those guys would probably want to help with you maybe they think money but at least it's a good it's good press x right when we help you know do mining so currently just funding currently there's 105 million dollars per year in for my rules not counting transaction fees hundred drive love your God as a today by you I sites are so guys one of ten bucks for you through it slow motion yes that's it i also used in five ether every 15 seconds ok I'll I think so maybe that's what maybe that's why they keep talking about purpose take leave them at you planning for it it's almost like this big ugly like you're gonna fest in a second you're never going to be able to use it I mean I think I think Rufus take is just an idea ideology you know so but you know obviously real quickly you're going to say well doesn't that mean the ritual control things you know I mean group of sake is one of those things it's not ideologically weekly so I'm not sure I sure have a proof of state works is it too busy too difficult to explain right now I mean I know I don't think that that one's right improvements take that hint for snake and very good God level it's just that you do have to put something at risk to make a bet on which blog is right and the idea we be there people also have to pre commit to do it you can't just come into mine for 10 minutes out an online form of that sort of on Adwords and certain ideas that you're going to be interested in finding the right blog for building the blockchain right but but when you send me typically any proof of stake says the more you put up at stake mathematically have much higher probability of being the Gothic forms of law no I'm not when you could join us a house now if there's actually like you're like there's only be like a thousand group of state of notes and it's going to randomly selected which ones actually get it doesn't matter whether use like one either or thousand there's an average of minimum is that there's a bit of a Mad Max not affect the way you can't you can't so everybody has exactly the same body probability and that's what I'm how the other improve it's not exactly the same but there is a there's a discouraging there's discouragement for driving is it sometime thought about how they group of state oops take mine in and in a different type of coin let me let me see if I've got that got it master to what you're talking about you click down group of state state and then you might have block and then your next block is your next block more difficult until a certain number blocks have gone by then your difficulty decreases as a number box go by making it easier for you the next block for Jupiter is that about right now other than the work algorithm goes out the window it's what would with the advent of paper and I won't say i'm fully cognizant ball but it read it once lightly the the argument is that there's a certain kind of a thing you get with proof of work requires your resources you can gain a better use electricity you've got to do certain capital equipment and abuses okay and that kids a certain kind of behavior if you have that behavior mathematically that that tends to encourage the argument that the etherion guys are making is that proof of stake gives that same kind of behavior but the steak is a bet your you basically went a little if you have the right block because you're betting which much ripen and if you would do it but you lose a whole lot for betting on the wrong block so therefore if three of us here are malicious actors and we're trying to get this phony block that gives us a lot of benefits to become the right block in the blockchain we can go ahead and play that game we can bet on our phony block to be the right block but if the network goes against us we lose our entire state she lose a lot of game I don't want to take a lot of time I'll go out to help deliver papers papers worth of me okay because I explain it yes it's too complex 2nd finger cream yep so done I just know i'll go up there you're an overly 95