On-chain upgrades for incremental evolution of public blockchains

Transcript

[Music] [Music] I can't be plus three years but it seemed like region of Carroll County to

be the inverse in next year's well I think this is the second time I've driven HD virtual so this is a fantastic moment in my opinion and we should try this for this let's get to the meat of the thing so it's quick introduction and background myself I'm Adrienne I'm founder of 50 laps and matter of state as you can see we already have two maps we don't get from out of state so Esther said you can find for our quick background here 30 laps is now a leader

that runs through state networks based in Switzerland we running out of the Swiss Alps and the purpose is to run highly secure infrastructure on technically interesting networks for example country life on tethers almost IRS Kusama and so a little bit better poverty neural networks to be interesting ideas that watch chemistry medicine on the other hand is a research interval in a very much focused on bringing novel research from across test together also hard problems in the country face to that end we have one of the main areas which this talk is also

rather than return mentioned some more which is cortical engineering so building protocol crates currently doing this with tenants the same one is two weeks of dependent island which to build transcriptions for contracts and third area is crypto cryptography zero knowledge proof and if you can find more information about office on the website or you can also appointment so with this brief introduction let's get into the evolution of public peace and first projects I wish I could change slides so the evolution of this talk is very much focused on looking what we have done so

far would be a during currently and what we can do in the future just to give you the context was launched eleven years ago nine and heaven changed much since same for theory theory was launched in 2016 I think 2007 sixteen I think of fifteen and hasn't changed much either arm but realistically speaking if we want the space to exist exist in the next thirty years and what these systems to exist in the next thirty years we need to come up with an answer for all these systems of all so let's jump to the next thing so why is constant

evolution essential blockchains that so that i don't right now uh in most cases prototypes there so early reset for the types and they're by no means for dirt romanian parts I know that teams will tell you otherwise in the loss of sales people in messes filtering otherwise but really from an engineering perspective they're just prototypes early research prototypes these functions are developed by a quantity and group of let's say 10 to 15 people they're not backed by sort of hundreds

of senior software engineers building these products assessing them so everything we currently have early research prototypes maybe this is not completely true for some Bitcoin not because pity panel itself sort of keeps up to date abroad because soccer is very stable a big con note that I launched five years ago I expect that though to work in fifteen years from now right so there's a lot of stability within the code base which to some extent may be good thing may not be we'll come back to this in a second but certainly all 16 and onwards researcher tribes and

they're far from being actually useful so first thing all trains by the way this most applies to public dock trains because if you deploy this in an enterprise setting like if you have a permission to blockchain it doesn't actually matter how happy you suffer is because if something goes wrong you can just hard for exit unilaterally this is a problem in the public setting then the permissions the other thing is we've seen so why is it important well in the last 11 years we've seen a tremendous run innovation around what useful flux could be out there that can be

implemented using block chains and due to that it is very unlikely the first prototype is going to be the ultimate problem right now so the thing that you launch is the failure like it should be the finalized version should be mined in 20 years to some extent existing black phantom take this into account and are running into real problems due to this like the thing that you launched today is pretty not the thing that you want to have in 20 years so you need to have a conference and evolution in order to remain at the edge of the use cases that I out there that use cases that block fins can solve and the last year is per

point everyone still trying to figure reported we don't know what the real amount of occasions I mean we started to see them we started to see the device pick up start to see that something payments could work but we were very far from the point where at several hundred since Ryan photography right just to consider this the entire industry probably today has less than a hundred thousand daily are like any shitty iPhone application starts beating this so will dog treats as a whole but they are far from being mainstream far being used actively by a majority of population and due to this as we

discover new use cases it's really important that large chains that existing systems can keep up with this evolution okay so this is sort of the introduction for y4y constant evolution is really important and I hope this gives you a little bit of context there next I'd like to talk briefly about the current approaches that projects are taking and you can sort of cleared eliminate them sort of 3 2016 2016 the kind approach is a really big on a very narrow like coin and of us so in that pattern that distinction we'll come

back to in a little bit so what do they do in terms of old well because the try thing me very well be it's perfect waiters and it's not going to hold early I mentioned that I expect you know five years ago to still work 15 years from now that means is going to be incredibly little evolutional progress in the corporate control to some extent this may be what bitcoin is going for in terms of its positioning right in terms of the way the communities trying to position it you could definitely take the argument that it kind of sort of

money and you don't want change money all so much and all too creepy but what I can already tell you that Bitcoin based there is not going to be a global payment system right like no one is going to accept the US trade offs of having to wait and the very least ten minutes of transaction so if the design goal of Bitcoin is to be globally use payment system it's very unlikely that the base layers go to achieve this without evolving significant their total possibilities about interoperability where Bitcoin as an asset may exist in the next hundred years but I think the settlement of the state machine in which it will be transacted if it's three

money is not going to be the Bitcoin base their solutions but I come back like two sides as well cons are sort of this model in terms of there's really no evolution one of the huge advantages it's very stable right it is very safe very easy to a McDonald's it's they're very likely to change over time there's a very neutral path but something has been battle tested it has been updated 11 years at this point I'm actually fairly confident this is superb free you know all so far sir but the downside to

this is it doesn't think it is not very changing at all and it can't actually use straight if you whatever system you want some here in the states or from depending on where you want to be positioned as a project you want to say the for it could mean I wanna do this but it means that this it can't really take a new use cases as they come on which is what we've seen as define like defy doesn't work even though maybe you want maybe you don't deserve the way 31 horse is so fascinating to me if areum started off very much looking

like paper writing it's not going to change it very much especially here one third one that changes since no different than what's even though they have to be coupled up forth but sort of the changes here it was made to offer a lot of minimal at best and the transition relief one to youth to is like dog food is donated straight to space travel and this is all the complete opposite to bitcoins approach which is we want to not treat it all to you thirds approach we are going to go from Stone Age to space travel we figured out how to use a hammer now we're sort of flying aircraft amassed from an engineering perspective this is

also someone ethical to me mostly because this is not a common Georgica build frankly speaking it's like and also there you can more look at it like there is and each one is its own thing and then if - just happens to be built by pressing the same people but it's really not the same chain like in the distance between those two la try because right now you have a chain that can maybe handle 30 transactions a second and in like in this in the e2 you just have 2024 charts and so global scalability everything works everything is incredibly efficient you're thousands of values Billy and I mean the best

garbage I am very much doubt is that it's completely backwards matter what people buy we were shells you a hat due to those current approaches what has happened is that we have seen the development of layer 2 and I think they are to really comes you know because as I say good idea that because it's sort of to test our confidence because at the time a layer that you came there was really no no right you didn't have systems that evolve correctly over time and you knew that the existing system error was going to change police they are drastic so in order to come up with a reasonable story for how the future

for the ball we as a nurse you came up with layer to layer to effectively says well we don't have to change basically the base there's fine and we just build all the nice friends Egypt's that everyone needs only to this girl could be massive trader across layer generally hasn't seen adoption because it's incredibly hard to use incredibly difficult to make security build security such gaps that are secure it was really difficult and I think if we had started the systems that can evolve over time we would have never seen their today I think if we're to start with something tells us that can is designed to incremental all over time

we would have no senior out because the correct place to spend your effort is not designing some really complex narrative system but rather focusing on making the base layer better and we'll come to this in the next part of this talk which is normal approaches so under these normal approaches I really asked the next wave of networks so talking tell us tell us by far the earliest one that launched this gallery cosmos personage and like recently parted with a life event as well Nia Solana miserable and so just to define a calorie and now return we'll talk a little bit about the approaches

that like these novel approaches so how does tells evolution actually work well tell us very much taste the model that you will have constant incremental upgrades right you want to not go well today with donation to my reader turning in starships to Mars you rather I go well next one was shipping this and six months later looking back and telling me yourself ship upgrade every three to six month and in five years you'll have shipped ten upgrades and it will look very different from how it started but it sort of happen gradually over time and then allows you to take out appearance believe you with

you to think people hug so this helps and to also incorporate effectively data you get from lock in terms of what use cases actually building for right because in the build ones launch now a change system you are effectively assuming that we know all the use cases that will be that will be required instead of saying well let's see what use cases are actually valid and which use cases appear and then start improving on those toothpastes so heavens to me is really synchronous automatic haunting upgrades ksama and do roughly the same thing you know in terms of engineering which is really

impressive so the way it works engineer expected in service brief cited over them with menacing we've been building so the last three upgrades at esos in conjunction with somatic labs and seduced that I know a lot of a lot how the system evolves so we want to record a lot so whatever you happen to practice you have a restricted subset of you compiler and the protocol itself is written on account and so then anyone can also as an ID injective protocol see if the community likes it governance sites do we want this to not want this and if they decide that they wanted and

then they get vote on it and when the protocol sees that a supermajority wants to have this new protocol no to automatically download the source compiler link it in Atlanta so and because someone does the same thing I do the same thing with webassembly but it's effectively the same idea it's at some point and some block my legs instead of running and vertically we not survive but it will be and that's how migration between you but it's a continuous chain where everyone that follows this maintain but also there's no use and attacks require one of the downsides whose approach though is that it is sharing risk right because

so significant is the else so there's so but it's difficult yes which determines

stop the roads export the state engineer Genesis file install the next evolution of with the Genesis file which is effective est export from the previous King so tell us the way updates book as he start on zero zero one you then add not ten equal to zero zero to a hundred use your three and so what happens is 0.10 you do not great and then he don't state and start new chain at zero this makes it incredibly difficult to sync the history genesis and to some extent also lost validators to have much lower incomes over the

governance process because they easily go and say oh we love upgrading our notes whereas in the tells case even if they wanted to as long as it total as want to do one two hopper to specifical anyone can just bring in Jack same protocol and be kind of in another vote and there's very little the validator so a person pillows can do to prevent this yes okay I also like to quickly talk about future approaches so we've seen what happens in the old systems a point in theory of an arrow so unlike coin may be very classic then we've talked about what happens in newer models tell us Carlos and I think

those are really two main things we will see we will see synchronous on China automatic launching upgrades and synchronous manual operates and now the future purchased how do we do ask how do we mitigate the risks that are apparent in the previous to Africa well we can do something that leverages maybe speed I know that maybe not all you know I briefly explain it I will see sense for interpreting communication it's a protocol very much in a sense of tcp/ip justify it's run by the same validators and let's say I don't tell us - sorry we can't work on knockin tells us of 0:07

and in this future model what could happen is you know on just a zero centigrade as tell us - it doesn't duplicate the asset doesn't duplicate ACKs disease but it allows people and it's secure tell us to run by the same power later set as fellows one and then use us instead of having to do a governance both they can asynchronously start moving coins or what it tells to in to test it out to see how it works and if they're like in the end state and then assuming that enough people move for toes want to tell us - you can have migration in let's say you sell bitcoins I've moved from one to two at that point you can just force migrate everyone else

the remaining 20% so you still maintain the property that you're on poking into crazy many systems is still making coherence but do you dear iskele all these operate because instead of having to hope that everything works at a low-stakes way in which tester operates and the seven real benefit to this is be kind of many upgrades existing at the same time so one of the problems in the current government's approaches is that you have let's say twenty competing proposals and like many versions of some just slight modifications of each other but as agreed you have to converge on a single one and this is sort of in a government function but maybe other

incentives there but without reason why we see with connecting reference you have proposals and you see where the market wants yes so this is so very briefly some thoughts and ideas and what's happening in the industry right now the ability of rockets so he spoke different strategies what was effectively built why we are noble and in theory we jump from the Stone Age to space travel with incremental synchronous evolution Devin Frost in the future connect Akkad why we see incremental so thank you very much I'm a driven

system again we'll just for a fee if you have any questions please feel free or feel free to attack me and thank you very much to listen and I guess we can take questions over here [Music] if you have some question no apparently and nobody in that case thank you very much and have a nice day you