Introducing the CAKE framework

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:08

   um I'm Stefan uh I was previously a co-founder and Chief Architect at flash Bots where I built something called the me supply chain and all the Jazz that comes with it um now I lead a research team called Frontier research and then I'll be joined with anit to talk about uh chain uh abstraction um it's a topic that we've been diving into for uh a few months now quite in depth but something

2. 0:31

   that's been passionate of mine for many many years um so I'll talk a little bit more about that if I can get the clier to go so my journey with um chain abstraction started in 2019 I built what I claim to be maybe it's not but is my claim to be the first chain abstracted application on top of etherum so I had uh basically no signs that you were using a blockchain despite you using uh ethereum behind the scenes was social login and all the inter actions were

3. 1:00

   basically using USD um and it was beautiful it took me a really long time to build it I really cared about it and it had zero users and it was really sad but um it was a really cool experience to build um I was really impressed with the usability that was possible to achieve on ethereum at that time um but what happened was gas fees went like this and then usage went like that and then we realized oh wow it's really

4. 1:26

   impossible to build anything that's web three on ethereum without ethereum scaling first um so between 2019 and now basically it's been somewhat useless to build anything L three weighted um and we've ended up in this situation where on ethereum if you want to build anything getting a transaction on chain goes through an entire mess of uh different actors um and so somehow we know we have all these different other layers now rollups uh that allow us to

5. 1:57

   have scalability of uh block Cas uh but we've lost the usability of the blockchain and we've gotten into a scenario where users have to manage balances across 10 different networks in order to get you know something done um so we flipped from usability to scalability um and basically my claim is that with chain abstraction um we can build new types of apps that actually achieve both on top of ethereum uh and that will happen this year um so we

6. 2:29

   introduced what we call the web3 cake um or chain abstraction key elements you might see this online uh we have nice merch over there and a nice cake uh that you can uh you can taste um but we presented this framework as an approach to building these applications in a way that you leverage all the middleware that's been developed in the past few years all the improvements over block

7. 2:52

   space capability uh while still having and providing a great user experience um to the users that are using app ation and a good developer experience um so we launched a working group with a bunch of different uh parties in the space uh near glor who helping Host this that we called the the chain abstraction working group uh or the cake working group um and uh we're hosting a workshop tomorrow

8. 3:17

   uh as well as one in London uh for E Denver um and we will aim to further the the cause of chain obstruction across the entire industry all right if we are successful it'll look something like this so just like web 2 had its own layers uh web 3 will have its own layers as well um and quickly we will realize that as we get bacing uh the cake is the last stage of uh of the process of of web 3 um I personally think it's the most interesting and exciting time to be building in web 3 like now that we have

9. 3:50

   the scalability and we have the middleware tooling to be able to build solid uh applications we will see a huge explosion of different uh uh you know applications being developed by uh developers that are coming from the web t space uh to to build with us um and uh yeah we can only see where this will go so I will pass this over to init now to go a little bit more in depth about the cake so let's uh let's first dive deep into figuring out what exactly is chain abstraction what we mean by Chain abstraction why is it

10. 4:22

    needed and then we look at what are some type of products that people are building explicitly and implicitly currently uh and how they fit together how they compos together and so uh let's go back 20 years uh when the web to Revolution was happening suppose you wanted to send an email from your Gmail to iCloud and you had this PDF like wedding wedding invitation and no SMTP no HTTP no standardization of how these different layers communicate with each other was

11. 4:54

    happening what you would do is you would go to an email aggregator you would somehow enter the two and from address paste the PDF and then you would get a bunch of options on how do you send that email from one person to another and once you do that basically once you sacrifice tens of minutes and uh tens of hours you would get an app version of PDF you won't even get the original version of PDF you get the date is missing somehow and so this is a very broken experience thankfully that is not

12. 5:24

    the case so thankfully that is not the case instead of doing this uh was tens of minutes and wasting tens of dollars what you have is a very clean uh very sophisticated drag drop and send type of experience and this is what chain extraction delivers basically um unfortunately that is not the case in crypto today like this is what we have and so you have 10 different rups you need you need to save balances in each of them to pay for fees every 10 seconds

13. 5:54

    you need to do this confirmation to switch networks and then you have these aggregators where you need to somehow bridge to actually do what you want to do and so if you want to give this kind of experience to web2 if you want to build the next set of applications you need to abstract this all the way from the users and it's our opinion that block space is abandoned by by now and

14. 6:17

    ux is the bottleneck to build the next set of applications and to build and to bring the next 1 million set of users um so the IDE ideal user flow should look something like the user just connects their to adapt signs what they want to do and then just wait for it to happen everything everything in between is just abstracted away from them and that is how you get the next billion users

15. 6:41

    in so um Stefan told you about these different layers of cake Let's uh look at what are the design tradeoffs in each each of these layers of cake and how all they compose together and if you are building like a chain abstraction uh infrastructure what are the trade-offs that you need to make to actually build a successful product so the first thing that you need to do is decide what type of message you're transferring across chains so are you transferring uh a normal message like

16. 7:13

    information or are you transferring B so a normal message all types of messages they need to be lossless and so if you are if you want if you're participating in a governance vote and you want to vote Yes into that governance vote you don't want that information to be converted into a maybe and so it has to be lossless and and for it needs to use all the secure Pathways that are and some examples are like government sping approvals for um message transfers the second type of

17. 7:45

    transfer that blockchains are meant to do are value transfers and this is what blockchains are made for so you when you transfer value from one place to another it has to be scarce you need to maintain the scarcity you need to maintain the atomicity of that trans action and um what you need to do is basically if you're sending from one person to another you need to deduct balance from the first person and Credit in the second um and it can be lossy like you can bring in third parties and say okay

18. 8:11

    like if I'm sending $100 I am willing to take a loss of $1 and just receive 99 on the other end um and so you can bring in these sophisticated third parties which give you a better us and 99 to 95 to 99% of transactions today on blockchains are value transfer there like transfers swaps payment for gas all these are types of value the next set of decisions that you would need to make are in these uh different layers of the cake and so um when you are talking about the

19. 8:44

    permission layer the four tradeoff over there is between ux versus agency so where in the spectrum of ux versus agency does the user fall uh if the user wants to have all the control then they would prefer something like an EA wallet but but if they want better ux they would switch to a more they would give control to the user or a solver in between so if things like account abstraction or policy based wallets which are telegram BS what near this building are examples of these

20. 9:14

    permission layer uh uh infrastructure uh options which basically uh give more uh ux Improvement to us the second thing uh that we would want to do is on the solver L is figure out this trade-off between optimal routing which is finding the best path to make sure that the user's transaction get executed versus actually making sure that the transaction gets executed so imagine in a cross chain world where there are a

21. 9:45

    transaction needs to take 30 hops to actually satisfy what the user wants to do there are M failure cases in between and there are multiple possibilities of actually making sure that the transaction gets executed and so if you want to find the best price if you want to find the best route what you would do is you would bring in a lot of solvers you would say hey give me the best price compete against each other uh but if you

22. 10:08

    do that then other solvers can front run that transaction and reduce execution guarantee and so you need to do this tradeoff between how much information I should reveal to uh to resolve this tradeoff the third tradeoff is in the cment layer which is between low fees and execution speed and so uh um when you are trying to settle these transactions you need to somehow communicate between these two chains and you need to wait for finality if the

23. 10:38

    finality is is on bitcoin you need to wait a certain number of blocks and that can take something like an hour and so there is this if you want uh low fees you would basically use the uh ecosystem aligned finality but if you are okay with um if you want more if you want better speed then you can bring in more solvers you can bring in okay the solvers will take on some finality risk and give you what you want at a at a better SC B so these

24. 11:07

    are these are the four tradeoffs that you need to take into account if you're building like a cross uh chain abstraction framework uh what you want to transfer is it value your message uh how much comfortable your user is and do you want to basically optimize an execution speed uh execution guarantee or uh fees so that's what brings us to basically the cross chain tremma so there this trilemma between these three things execution guarantee low fees and

25. 11:39

    execution speed and depending on what you try to optimize for you would fall into one of these three categories and depending on how comfortable you're using users you would go more into into this direction of this trial and looking at these four trade-offs that an application needs to make and this cross chain trima you get six pieces of cake or six different designs that basically emerge um where

26. 12:09

    uh you can the user can choose whatever they want depending on what they prefer and these six different designs compos very nicely with with each other and they have their own product Market f u so suppose if you want to do a value transfer and you want to sacrifice for Speed but have very high execution guarantee or uh and pay very low fees what you would do is you would use a token anointed Bridge so something like

27. 12:34

    USD C where what happens is the token team basically says okay this is the canonical token address on each of these chains and this is the canonical bridge on each of these chains which can Min tokens on the target chain and so uh this is how you have token anointed Bridge it has it has to wait for finality and all the chains where it is transferring value but because it is Anointed by the bridge there is there's

28. 12:57

    very little the second thing that you would do uh if you want to transfer a message and not value is you would use any custom aligned Bridge so this is like a shelling point in an ecosystem where the ecosystem appoints some sort of a bridge which is used to transfer arbitrary messages across uh the ecosystem so these are things like IBC Adler super chin where these are ecosystem align message Pathways to transer messages ACR different um parts of the ch

29. 13:30

    the second type of uh chain abstracted products that you can build are called as solver competitions so if you want to get a good price if you want to find the optimal route what you would do is you would let solvers compete with each other and so the first type is if you want to transfer value or if you want to do swaps you would make solvers compete for Price amongst each other and so uh

30. 13:53

    these are things like chumper bu Unis swex where there are a bunch of solvers uh the looks like aggregators and a bunch of solvers they give them the they give them some Cort and based on the best price uh the user gets what it is on the message transfer side you have these wallet coordinated messaging systems so things like Alfred avocado near where the wallet acts like a super

31. 14:17

    so the wallet is an aggregator of an aggregator this was an aggregator the wallet is an aggregator of aggregator and it can pick and choose whatever the user wants and executes it on the third is uh this execution competition where you sacrifice on basically price Improvement or optimal price but you are trying to optimize for execution guarante so you are trying to optimize for uh if I can actually make

32. 14:43

    sure that my transaction goes through if I can actually make sure that of those 30 things that I need to do to actually satisfy users and Dent they actually end up so uh you get things like solver speed competition the solvers are competing not on price but on speed whoever whoever gets to execute that intend at first and then uh finally to transfer messages you have things like P off for Robin Hood where you just

33. 15:07

    there's no competition between solvers you just give give whole flow to the solver one single solver there's no information and then they can p they can batch things together and make sure that the users in saf at a high execution so these are the six pieces of cake um to just recap I think uh chain abstraction ideology is the ideology that users should not care what network they are on users caring what network

34. 15:36

    they are on in web3 is very similar to users caring on the internet ER what what cloud providers they all so like AWS or GC they should they don't care about that um and we introduced the cake framework to think about these chain abstraction key elements and how they compose with each other how these different layers composed with each other and what are the design tradeoffs and um what are the six designs that in t and I think in order to successfully

35. 16:01

    make the Deep of chain abstraction uh let developers build the next set of application bring the next magnitude of users of crypto we need some sort of standardization on how these different layers of cake communicate with each other and that s thanks a lot