A Modular Blockchain World - Nick Dodson

Transcript

and command f1 on a macbook with a touch bar uh okay displays okay okay i think that should do it yeah there we go great great

okay awesome okay can you guys uh this is good that's good all right uh awesome well thanks for having me eat denver and my name is nick and i'll be presenting um a presentation on modular blockchains and particularly where we're headed with them and also you know some challenges that i think we're facing and also some things we're doing horribly wrong currently and how you know the modular blockchain system will affect layer 2 particularly and kind of your conception of layer 2.

so in our current kind of state of blockchains we have a lot of monolithic blockchains so i don't know like raise of hand in the room how many of you heard of this modular blockchain idea and also monolithic blockchains yes okay kind of all right so essentially there's a movement in the blockchain space right now particularly with ethereum as well to move us closer to a reality of what we call modular blockchains but before you can understand that you have to understand what we call monolithic layer one block chains so monolithic layer one blockchain bundles data availability settlement and execution together and because it tightly bundles these

things you can't specialize and thus is not scalable and so in the monolithic model that we currently see now with chains you know things like uh well i mean ethereum layer one currently as it stands is monolithic but it's moving towards a different reality but many other blair ones are essentially in the monolithic model and as the previous presenter was bringing up this is you know constraining a lot of factors particularly on bandwidth so modular blockchains they separate execution from consensus and data availability and when you do this each layer can specialize and be and be more scalable and so it's the specialization that you

get from breaking up these pieces that allows you to produce far more scale that is secure and verifiable and doesn't affect decentralization so this is particularly the key for how we head into ethereum 2.0 and also other data availability layers that are going to come online and the particular direction is a modular one where you can expect an extremely high data availability throughput through things like sharding and you know we expect even maybe as soon as next year that we'll be able to have you know tools and the ability to leverage this kind of data and so

this brings me to my next point which is particularly particularly key and something that i think a lot of people should focus on um and that is that most layer twos as you can see them right now have a huge problem and it's right in front of you and it sort of is something that i think people don't realize when you're talking about layer two it's one particular thing and you can see it right on the screen which is that there's a huge fixation on layer twos to optimize for bandwidth and so this fixation is very very heavy there's a lot of teams that are focusing on trying to get

the literal bite size down of transactions with lair twos on ethereum but this is really dumb so this is not a good idea and this is a huge waste of engineering resources for the community and particularly it is really dumb because bandwidth is a monolithic issue not a modular issue the constraining bottleneck and the modular paradigm is particularly not bandwidth and if you focus all of your engineering efforts on producing transactions that are 10 bytes or like four bytes then you're basically

building a system that a we don't need and b is particularly not useful and is also just a waste of engineering resource in the modular blockchain setting it's not a problem again because the layers are broken up so a modular block chain can specialize in things like high data availability and when you can do that it opens up the amount of data you have such that that is absolutely not going to be a constraining factor for a very very long time at least probably for the next two to three years at best so the bigger problem in the space is that not only are we fixating uh layer twos you know particularly some of the

current uh you know high profile layer twos fixating on things like compression where they shouldn't be but we're thinking about modular block chains and layer twos completely wrong as complete nonsense in the sense of how this is and it's almost pass a in in many senses given what will happen and what is about to happen layer twos is a concept only exists from monolithic blockchains they do not exist for modular blockchains there's no concept of a layer 2 in a modular blockchain and this is something that will go out of date very very quickly as we step into a modular blockchain setting

and modular blockchains will eviscerate the monolithic models hands down because they can separate these things and retain decentralization and security so we have a massive incorrect focus in the layer 2 space it's kind of interesting to say the least but you see a considerable amount of fixation on bandwidth when layer 2s don't and shouldn't be fixated on this thing and also they're just optimizing for the wrong things in general so they're optimizing for other things as well that don't make a lot of sense to optimize for particularly and the real bottlenecks are something else

and something that highlights this very clearly is a conversation from from peter who's the head of the go ethereum team and he was talking with avalanche uh on twitter an avalanche uh you know because this is an ethereum conference we can we can riff a little bit uh but you know avalanche problematic to say the least was essentially saying well if we can just switch out the consensus algorithm uh we can produce enormous amounts of scale uh and uh peter was like no you can't because consensus is not a bottlenecking factor it's actually something else which is the evm and state growth and

particularly the entire processing model of the ethereum system and so this is a very clear and known thing we are tremendously bottlenecked by these processing systems and if you just open up the bandwidth even slightly on evm based systems and it doesn't matter whether it's a roll-up or a layer one you're going to be heavily heavily constrained by these processing factors and so the solution to these constraints is to get your mind away from layer twos and into something else which is called modular execution layers modular execution layers are different from layer twos in the sense that they

are a verifiable execution system which can be its own layer 1 layer 2 layer 3 side chain state channel etc the key thing here is verifiable and it's what makes it different from just another layer 1 or you know any particular system a ver a modular execution layer encompasses the idea of a layer 2 but it's a little more than that it's a verifiable execution system and verifiable meaning fraud provable and also focused on something that i think as a system isn't targeted towards a layer two uh for a monolithic blockchain but is actually optimized and this is

the key thing for modular blockchains you can forget about the concept of layer twos and think about verifiable execution systems in essentially the modular execution layer idea this brings me to the shameless plug of the presentation which is my project called fuel and fuel is planning for this particular modular future and fuel is the fastest modular execution layer that exists ever in the world at least right now what does fuel bring to the table for modular blockchains it brings maximum security and the

highest flexible throughput and the focus on flexible meaning general purpose turing complete smart contracts in the highest throughput setting this is really taking advantage of the modular blockchain paradigm and how data will not be the bottleneck and this is a particular focus for the project in general and how we achieve the throughput is by de-bottlenecking the existing massive processing bottlenecks that we have by using a different paradigm on the execution layer and so this is focused on particularly parallel transaction execution which is something we see heavily with the solana design but secondarily

it's also a superior virtual machine to talk to the parallel execution point parallel execution is a tremendous scaling factor it allows you to open up more cores and more threads which means more ops and more compute which means you have a wider consensus zone over the same data that everyone else will have so this gives you uh huge amounts of processing potential that you're not going to get on a single constrained thread or system and currently the evm as it's currently designed is effectively single threaded it's effectively only leveraging one thread so chains that can use more threads and use them more effectively will

they'll they'll give you so much more compute and so much more throughput you won't ever want to use that system again and it will be highly antiquated so this is the first thing the second thing is the fuel vm and in this we focused heavily on all of the design you know shortcomings that the evm has particularly around how it handles compute its bid architecture the way that it handles calls the way that it handles memory the way that it handles almost everything in a way that can be redesigned for a better execution layer and one that has a much wider throughput that it's going to be handling for its node and the last thing is a superior

developer experience and this is what we get with sway and fork and basically to sum that up we've spent a tremendous amount of time learning from all the mistakes and the shortcomings of architectures that we've seen with the solidity ecosystem and we built upon that but we built upon that also knowing that rust is a thing and rust is a beautiful thing and that we can learn a lot from that ecosystem as well and other ecosystems in the space if you want to see more about sway and fork you can look at emily herbert's talk at 120 on the wave stage and she'll be kind of going through you know sway in in more depth and as well doing some comparisons between

solidity and sway and lastly to bring us to this future and also to bring us the highly sophisticated machine that we have and also the throughput potential you know i think it's good to highlight that we have more than one or two engineers on the team so uh you know a thank you to the 28 contributors and engineers who are on the fuel engineering side that are kind of bringing this to light in every way possible and if you want to follow us and what we do with these high throughput systems particularly that will run on things like ethereum 2.0 you can follow fuel labs you can also

follow the language and tooling side as well which is swaylang and you can check out fuel network and our github and i'd highly encourage you to read our specifications because they are thorough they're very clear and there should be no edge cases so essentially you should read the specs and we're also hiring um so that's my presentation um hope that that makes sense uh thank you awesome thank you uh we do have about five minutes if anyone has any questions [Applause] let me bring you the mic