Layer 2's Cambrian Explosion by Ben Jones of Plasma Group

so I guess what I wanted to talk about a little bit is I think some of the trends but the big trend that I think I see having a plenitude space right now and like most simply but i would say that that trend is diversification so what we're seeing there specialist the past year and half two years is that the number of scaling solutions out there and particularly layer to scaling solutions are exploding we're getting huge huge numbers I want to talk about sort of like first of all the history like what gave rise to that how does how does that progressed how does that change over time and maybe also what we should be doing about it sound cool excellent ok so I see a lot of familiar Christian women so hopefully you guys have an understanding of what else who does I like to think of it as like literally a layer cake where we have things like just assumptions and centers and particularly these two guys actually computation and networking to build something more hard then the black chain itself allows you to do right even be sort of stuff that all on top of the main chain somehow and this gives us some nice properties a different system than the my team did and particularly the thing that I think is most important that it's very very helpful in our understanding of what the heck we use there are two things are is it's all about the speeds like this analogy you don't go to court to cash a check you go to court when check bounces this is the analogy that usually allows is like the intuition for why our two scales right we don't scale by allowing ourselves to process more transactions we scale by processing disputes without transactions instead of the transactions themselves and this seems to fit our that concludes models from one of theirs you know like how the world was today and it seems to be a very good paradigm for us to be doing in Malaya - right and then the goal is you know usually to go fast so please throw the speedometer up there okay so what kind of things can you out on top of your blockchain more than uh watching well the simplest one is to comfort black shade right so I barely on later to research went towards side chains and side chains are simply we lock some money up from one chain and we sort of open it up on another chain and now the second chain is where we're processing transactions and because of this we have to block chains and we have double the scale or maybe this this second side chain is like you know less secure and has even more scaling for selection to be more than double their right so this is this is like some of the early the early explorations of where to I would say where it gets really interesting right state channels I would say that state channels were like the big aha moment of layer two development where we finally realized hey we have this thing that's not a blockchain but does transactions that we know sort of are going to happen on a black Channel and related to the blockchain and the idea is we exchange io u--'s right so we have this we have these great tools of cryptography allows to sign things in the non-portable way and so if we sign a bunch of things between each other desire use and we give a period to go onto the watch and say hey this guy's I'm going to have you and he's not talking about the bucks then we built a state channel so a state channel is really cool I don't think oh I can Ark was the first channel construction but it's obviously like the big one that we care about today in public it was like you know two-lane lots of great things about the Lightning Network and then I think this is more contentious but I think another big moment in their disability was understanding plasma at the time of this paper was put out that was a very ill-defined concept but it is what I spend a lot of time working on and it opened the doors for a very large design space because plasma sits somewhere really in the middle of state channels and side chains in that plasma does have this notion of another chain and others sort of chained of even blocks you could say however it does not just act independently like a side chain does it's sort of tied in to the route chain which is the watching the main chain and because of this time you can use disputes that that have to do in these ties to get some very nice security properties maybe even divided without a little bit the plasma landscape itself is massive so this is where I spend a lot of my time researching so I'd like know firsthand that the rest of you can't even track all the things that are going on in Cosmo we started with minimal viable plasma you know mr. mr. Kudrin came in and saved the day with a plasma that actually worked but then immediately we got another plasma which was a lot of other people coming together and saying hey there's something called plasma cash that's a lot better but very very quickly write the number these things type increase and what we have today I think this is actually almost a year old now every six months okay we have a massive number of plasma instructions right just a huge number so this is really cool it's really exciting to me as a plasma researcher because each one of these has a cool little nugget of information that I can you know break down and explore but realistically it's crazy these things are blowing up and to step back to later - I knew we have dozens and dozens and dozens of these counting solutions we have true good arbiter of no cost are both right this list is massive right this is just a random selection things like pulled off the top of my head there's a lot more of that came from so that that's a bit of like history on lay a tube and I think it's relevant to see that what is clearly happening is we have a trend we're way more layer 2 solutions are being created and you have anticipated three years ago and the eater making trade-offs and doing different things and there are many of them are very very very interesting okay so that's the first point questions so far I said I wanted to be reactive I've got no questions yet so I'm gonna have to make you guys feel awkward and stand here for a minute hey contemplated yes we've questioned about the shine of Kings we do that there ah so I would absolutely include this so shadow chains also being couple recently optimistic rollout gone by a few different names over the years is a sort of plasma ee sort of it's a very strange so make up I'll make an aside here which would be to say that one of the things that's happened as this design space has exploded is that the line between layer 1 and layer 2 is being increasingly blurred so there are lots and lots of like a sort of hybrid or combinatory protocols which take some elements from some parts of layer to south even some outlets for parts of layer 1 and so what like very things so simple thing that you can do that we've realized is you can use only one block chain don't have to block chains but have a sort of shadow training sidechain a roll of chain in which they bunch of trans actually did it is posted and we paid for it to go on the watch chain but we don't pay to execute the transactions so we have this optimistic notion if someone says hey here's a transaction I claim that the results will do this and instead of checking on the watch chain in this port system that that would be that we simply allow someone to say ah it was actually something else this person lied and so even in doing that we get a very large nice scalability increase that makes pretty few sacrifices but it's kind of weird cuz it's all happening a layer one right so and you can run that system almost entirely through layer one that's very low sort of architectural over in terms of complexity and payment channel routing and Plasma confirmations of all these things some very very simple construction and it really somewhere in between right so again I think that's a great example of yes I would put that in here and I would say that really what we're seeing is not only an explosion of layer twos but an explosion of layer tools like the wards later not like this we have two dimensional sort of explosion of designs yeah great question other questions all right well that's fine by me because we got one in so far so next thing I want to talk about this sort of like with that framing the history of l2 and like developments in l2 what's going on what should people do and I think there's like two real framings to ask especially one is as a developer as someone trying to make blotching applications what should I be doing about this explosion ability protocols and the other one is as an l2 protocol developer what should I be doing about this I deserve to pretty important lenses under which we should be addressing addressing this explosion so as a dad I'll start there I think the big one is you gotta understand the trade-offs right so not all of these layers of applications are built the same and we have to understand that one one really big caveat here is like scalability metrics can be extremely misleading I don't think there's ever been a word other than maybe like ICO or Cloyd in the crypto space that is as over news and under specified as transactions per second like GPS is this magical thing which I've own claims that would make a high number of but there's a lot more that goes into transaction than the number then you can do per second write a very very large number so for example one of the things is transactions per second is usually specified on a an average basis but there are many schemes in which you can get a million transactions per second but really that's because a billion transactions get finalized every week right and so if you do that then that has like protocol as a sorry excuse me as a app developer that has very different properties than selling of this instant transactions right this is a very very important thing understand it's not it's not transactions per second is not do one metric there are lots of metrics there this is a really fun quiz that I found off the Bitcoin cache subreddit I will admit but it's very good example to trade-offs that later to design spanking so here we have a payment channel network and the question is how many coins can Alice sent to Frank now there are six coins I believe locked up for average user of this payment challenge and these six people but in fact Alice to only sends two coins to Frank and actually in one leg in transaction Sheila's in luncheon f72 but regardless of that that detail it shows that these trade-offs are very real because one of the things that we would observe about this is six coins locked up the person in the world of layer two is going to be higher fees because there's an interest rate that Carol and Ella and Bob and David could be seeking elsewhere in return for the money that they have locked up and so for them to be willing to facilitate a transaction between house and Frank they're gonna want to get people and so these fees may be non-trivial in Campania channels and as the demon channel of space becomes more robust it explodes we'll see what that trade-off actually looks like so because there are so many of these guys I can't really go into the specifics of all the trade-offs but there's like three general three purest sixth at I think I'm pretty good as a application builder that I think are relevant so the first one is we can scale the sacrifice in security I think this is a land dominated by side chains so because you can make a side chain across this very large number of transactions in return for hidden things like blog propagation time and verification costs we could get a large number of transactions through a side chain but it's actually its own security another option is we can sacrifice the cost so we may be able to get a lot of transactions but they're not as cheap as another scaling solutions this is what scheana's did to probably probably do and this we'll just talk about my side another example is in plasma which is where I started working in the layer 2 space we sacrifice functionality so we have a nursery generalized plasma and the ability to build more complex contracts on Tom plasma well the truth is that's very hard to do it is not like an ethereal contract writing a generalized plasma contract to do selenium is a non trivial undertaking it requires the reason about the incentives and the ownership and details of that particular contractor trying to code it's not like writing a salinity contract so these are some nice heuristics that I think are relevant if you're building on a way to application questions how is concerned since the next question boss just a highlight be like plasma versus channels are fundamentally kind of open contracts versus better parties yes that's another business so in this craft notably not only can Alice only send two points depend but she can only in this graph assuming this is the program said coins to Alice Bob Carol developer Bank there are no other people who could say they received a coin in the payment channel based on this craft and so you're absolutely right this is in contrast to plasma in which someone who does not have any sort of channel on the mid chain they might not even have an address on the main chain yet can receive some money so that's a very notable difference another great example of trade offs yeah absolutely other questions comment [Music] absolutely yeah so this is like my antennas framing this as someone building an application is to say exactly that depending on what the application is you need to change how you want to apply there to to your application and some of them don't make sense okay other questions comments concerns how are we doing on time when you make sure you're being respectful bow as you get so next I think we're okay on time okay so the next perspective I'm going to talk about it is sort of the perspective that I have as a protocol Cheniere developing later to applications than there's some other wonderful folks after that kind of group so that's exciting to be able to speak to some like lines on that what are the big things that I think about okay let me back up and explain why I have this icon right here I think that the name of the game must be unification so I do not think that we are going to be able to get by in a world where a thousand different scaling solutions have a thousand different api's I'm a thousand different sets of terminology to describe the disputes that are used and so on and so forth there's a lot of thousands that you can name there right well the neighborhood game is that for people to even be able to to to understand the trade-offs and to choose which or not they want to participate in there has to be a unified way for them to to to understand where two protocols because protocol developers are always gonna understand layer two protocols better than how patient offers are that's how it should work that's the point of the distinction in the first place so I think that's very good for you so I am in the game use unification so one great way in which we have our signification to mix a weak side out to the state channels folks in the rooms is that state channels are coming together and unifying this concept and this is a very very big deal we've got a bunch of wonderful people on this list that are participating in that I think this is really really really good it's crazy to think that multiple state channels especially because they serve a very similar purpose across different state channel permutations the API should be the same crazy to have two different states all applications in which the API is which they communicate and ten percent money are different this is very very unhealthy for the ecosystem and for a developer trying to build adapt on top so the unification of state channels is coming along it's very exciting I'm super jazzed to competing more on this it's gonna be great and now it's time for me to show my own stuff obviously why else would I be here I don't like about that ovm so what we've seen is that beyond state channels will show the forest that there's a very large design space going on and this is especially true in plasma what we were working on and this led us to say that we call the OVF org optimistic you're a machine so basically the idea is that even though many of these men even though there is a large number of these layer two new protocols and they're all making slightly different trade-offs and design decisions and they all function a little bit differently at the core they do really all work very very very similarly and the way that they work is you dispute things on main chain and the way that you guarantee that you have some money or that some state is in a certain state is that you know you can dispute it to be as such on the main chain if required but optimistically you might not have to right and this is what coming at the scale of blockchains synovium is an attempt to break all of the liner to scaling solutions down and understand them as individual components and based on those components use just one wallet and just one contract so this is the sort of far-reaching appeal of the OPM there's definitely a lot of research that needs to be done to really make that universally true for the broadest number and possible layer two applications but for things like cosmic channels we have a we have a pretty good start at doing it so if you're interested let's talk about how that works a little bit you can kind of think of the OPM as a set of dispute Legos and actually based on this fun thing which I pulled a great Wikipedia screenshot from a dialogic logic so this was really great dialogic logic and I can't think of a better name my goodness biology logic is the idea of framing truths about the world as arguments between two people maybe a player and an opponent that ought to call and they may per state may challenge response protocol and it turns out that for like 100 plus years philosophers have been talking about what if we thought of logic as dialogues between people well this turns out to be a great analogy to watch a dispute games because effectively blotchiness begins our dialogues between two people in the state channel for Alice and Bob to be in the state channel when they have a dispute they say well I have this I use and says well no Alice I actually the more recent IOU from you and then I'll says well Bob I took the idea back and so on so forth right so this ends up being a very good training so for example and this actually has been very closely logical concepts such as and so leave it or not there's an end game okay and so you express an end game as game one engage you with an anger outlet and being able to win an endgame corresponds to being able to win both Game one and Game two another example of this is an app game so if to win the not G one game you have to be able to beat someone else at G one game right and then maybe another example it's a little more complex example is the for all game the universal quantification game which says for all G 1 of X anyone else can choose some particular axis we'll call that T and you have to be win be able to win G 1 of T for any such X any such - yes absolutely any such X is substituted in for the free variable of X [Music] so it turns out that we can actually build channels and plasma using these constructions so these are a crazy large pieces of code flash Jason that I won't go into too much detail right now but we can see we've got a knot here we've got an N here we've got it there exists which is the dual of Universal quantification existential quantification not for all not is exists and that exists not as for all we also have things like like a for also exists so we have an aversive modification here so we can express channels and we to express plasma under the spring so there's much to be done with the OBM but we think that it's important effort to be able to do this because in the future just like we've seen in the past the number of l2 solutions out there are going to get much larger and we want to be able to explore the design space in a negative manner and be able to incorporate the designs into applications without having to reinvent the wheel of API is and networking and all that as we go so that is what I work together it's a good question absolutely the format is quite lovely when the language is online uh yes absolutely so this is so actually I think this wonderful plasma example from Crypt economics loud shout out to CBL oh yeah I'm ties with stuff so this is this is because they write in rust because they're crazy in the great catch code and in general what we have seen is that there are like several ways to express mostly attitude applications so for example G 1 and G 2 is the same game as G 2 and G 1 so one of the open questions and interesting questions we'll be seeing is that there are several linked logical interpretations of layer 2 systems that look different but in fact they're the same and so that's absolutely right but that would be an example with that but you could be seen them in any case that is the end so if you want to follow positive group or any of these nerves and turn handles but thank you all for being here and listening and ask more questions please [Applause] that's a great question so that is a that is something that we alluded to before and I in fact tried to emphasize it on this slide here sacrifice functionality evening out on plasma has been a long-standing need just like cosmos the long-standing theme and truthfully we are only coming to understand what that really means and what cannot be possible as we explore this design space so right now there are no secure plasma implementations that allow you to write a salute II contract and it may be the case that this is a fundamental result that without some further restrictions it's not possible however with the oh yeah we found that there's a larger design space so there something called optimistic roll-up we talked about this shower chains before which does allow you to add some layer to scalability to police smart contract systems the intuition for why this is the case is that almost always layer 2 solutions handle disputes based on intent so Alice and Bob are the relevant ones to dispute a state channel because they are the ones who some money but a jealous salinity contract is unclear who if anyone wants to dispute this state until evidence is making it very tricky to build a very very scalable off chain application so it may not be possible as a matter of fact a great question yeah so from a research perspective I definitely see the deal like having kind of a very abstract just kind of description of having spread a false work but kind of from a practical perspective so like a channel example like this is a very practical need of we have some standards and we need to make a new standard to add that they are so essentially like how do you see kind of the like what do you see as going to be practical outcomes of kind of applying this or a model great question so I am definitely a researcher and so you're right that I could totally be going down a research rabbit hole of unnecessary stuff however in this case I think that it's very much not the case and I'll give you I'll give you maybe two reasons why one is because I think that this explosion of layer 2 solutions many of these designs are going to develop and we really don't want to rewrite the API so the beauty of this all beyond designs right is that you have just one wallet and just one smart contract that goes all the disputes for these different applications so in general this is very useful because it means that this code right here these sort of jason'll objects are attracted by a generic smart contract and a generic client that automatically understands that it has safety on some off chain property based on these constructions and so that means that you essentially cut the work in half for alt development you no longer have to write it client and write a contract and make a pump each other you have one client and one contract that speak OPM and you write things on the end that they understand as a more concrete example what is valuable the all of the hard problems of channels are solved by plasma so you can put channels on top of plasma and it's both of them much much much more powerful however it is very very very tricky to integrate a state channel client and a plasma client if they not happens same language we tried to do it and it was very very very challenging so that's I think the most concrete example is that I think that the future of plasma and channels are that channels are always on how the plasma hive has always under channels I mean that's very valuable yes one more question yeah I'll go again is this kind of like on changes like work separately different let's say a plasma operators who an interesting question so it's hard to understand what you mean by a plasma operate there's a dispute to mechanism so the thing that a possible operator is mentally doing is providing a optimistic state transition scheme in which disputes don't have to happen so the role of the plasma operator really is not always to dispute things it's to follow the rules so that users don't have to dispute the plasma operator so it's a little it's a little tricky to understand what you mean by that but maybe that's helpful answer other questions well make sure we could give no type of links were always asking the question so I my my intuition is that it might not be as efficient like to just happen right you never specialized market very very high optimized and sometimes may not be comfortable all that culturally I'm very very serious about language you know kind of how can we combine the or have some kind of a way to optimize it so that it's straight off the college now right that that you never might be done at the factory that's it that's a great plan so certainly Jenna generality versus specificity often comes at performance cost definitely the case so the long tail answer that I would get would be that eventually both of these objects should be compiled so we see these structures these OPM structures as in fact that not things not nothing's to be interpreted by one smart object but in fact to be compiled to a smart contract but there's only you know you can't boil the ocean and starting with a compiler from this dispute language to the dispute contract it's very very difficult in the long term I think that that's absolutely the right approach is to be able to compile these things into contracts and you remove the this generality sort of extra hop that is definitely going to give you a performance but hopefully also because we don't want to speech down in any way we're like optimistic that they won't have us hopefully the downside and the slower dispute resolution process is matter people of dispute things because why would you just good things are gonna lose all the time and that's the point yes absolutely so it's important understand what that means so you could build the decks on plasma you can build the decks on a state channel and you can go both both plasma and a state channel on the OPM so you absolutely can there's a larger question asked there which is what kind of decks do you want to build but you must definitely can absolutely