Latest / TezTalks Radio - Tezos Ecosystem Podcast / 116: Latency, Instant Confirmation, and the Next Phase of Tezos

Transcript

116: Latency, Instant Confirmation, and the Next Phase of Tezos · TezTalks Radio - Tezos Ecosystem Podcast

• 0:13
  SPEAKER_00: Hey everyone, welcome back to Tez Talks Radio, where we sit down with the people building on Tezos.
• 0:20
  SPEAKER_00: Today I'm joined by Jan Vejigianis, head of engineering at Nomadic Labs.
• 0:26
  SPEAKER_00: We'll talk about Tezos X, what it is, how it differs from what came before, and where things stand today.
• 0:35
  SPEAKER_00: We'll also spend some time on latency and instant confirmation, what those mean in practice, and what kinds of applications they make possible.
• 0:45
  SPEAKER_00: Jan, thanks for being here.
• 0:47
  SPEAKER_00: Before we get into it, would you mind introducing yourself to folks who may not know your background?
• 0:55
  SPEAKER_01: Yes.
• 0:56
  SPEAKER_01: First of all, I'm really happy to be here today.
• 0:59
  SPEAKER_01: It's uh always great to uh to have an opportunity to talk about Tezos.
• 1:07
  SPEAKER_01: Um so who am I?
• 1:08
  SPEAKER_01: I'm head of engineering at uh nomadic labs.
• 1:12
  SPEAKER_01: I'm um leading, um helping with uh the development of uh the core engineering, especially the Tesos X uh roadmap, as well as the uh Tesos protocol in general.
• 1:29
  SPEAKER_01: Um my background well I uh joined uh the Tezos project uh almost uh yeah uh six a bit more than five years ago.
• 1:43
  SPEAKER_01: Uh first as a staff engineer, I worked on uh the scalability roadmap, I uh led the uh as an engineer the the smart road ups uh implementation as well as um some optimizations of the L1 protocol.
• 1:59
  SPEAKER_01: And uh at some point I started uh being more of a manager, but yeah, actually I still do a lot of engineering, so I don't know.
• 2:10
  SPEAKER_01: Uh I'm I'm just here to uh make sure that Tesla is a success.
• 2:16
  SPEAKER_01: And before that, I've been uh uh associate professor at the University of Paris for 15 years, mostly working on functional programming, verification of critical software system, um yeah, and many other stuff.
• 2:30
  SPEAKER_01: But that's a short version.
• 2:34
  SPEAKER_00: That's great, thank you.
• 2:35
  SPEAKER_00: For folks who might not be familiar, what is Tezos X?
• 2:41
  SPEAKER_01: Okay, so Tezos X is um um a modern uh blockchain.
• 2:51
  SPEAKER_01: That's uh how I I like to see it.
• 2:54
  SPEAKER_01: It's um a complete stack where you have uh a full integration of a layer one, the Tesos layer one protocol that we all know that has been there for a long time, has evolved a lot.
• 3:11
  SPEAKER_01: Um, a roll-up, a very large rollup, um that includes um several runtimes interfaces to uh to many ways of um in building in a blockchain, and um these whole things actually creates an opportunity for a very um um uh yeah blockchain that offers um an important throughput, um an important composability between applications, as well as um very uh low cost um uh for transactions.
• 3:57
  SPEAKER_01: So yeah, I can come back on uh on Tesos X on the many aspects, technical aspects of it, but the idea is to just make sure that when you deploy on Tezos, you have a way to uh feel at home.
• 4:15
  SPEAKER_01: If you come from the EVM world, you have an interface.
• 4:18
  SPEAKER_01: If you come from the more traditional Tezos environment with Mikkelsen, you have an interface, and all this run in a single rollup that um is extremely powerful from the performance, the security, uh latency uh side of things.
• 4:39
  SPEAKER_01: So that's more or less the idea of Tesos X and what we want to propose uh to the Web3 people.
• 4:46
  SPEAKER_00: What makes this direction different from previous upgrades?
• 4:51
  SPEAKER_00: Is it a shift in approach or more of a continuation?
• 4:55
  SPEAKER_01: Oh, it's a continuation.
• 4:57
  SPEAKER_01: We have been uh scaling Tezos uh for many years now, um, first at the level of the layer one by improving, improving, improving the block times, the execution uh of smart contracts, etc.
• 5:12
  SPEAKER_01: And um by introducing smart rollups, we allowed ourselves to um actually decouple certain aspects of blockchain.
• 5:23
  SPEAKER_01: So typically the security, the settlement layer, is in the layer one, the execution of smart contracts is in the rollup, and you also have the data publication aspect that is in a layer called the data availability layer, the DAO.
• 5:43
  SPEAKER_01: And by decoupling all of this, we have been able to evolve uh progressively Tesos into a very um powerful, a very large uh environment where block space is um very um uh stable in the sense that you have bandwidth, you can build everything you want uh in uh uh by using any environment that is available to you.
• 6:14
  SPEAKER_01: So that's more or less the the provo the promise of uh Tezos X, and uh it has been gradually deployed, and actually it is still under development because our goal is to ship uh Tesos X by the end of the of the year, but as usual with Tezos, it's not uh you know uh a promise that will come in uh many months, it's uh continuous um deliveries that we are we are doing on the Tezos project.
• 6:44
  SPEAKER_01: So yeah, that's uh that's what Tezos X is about.
• 6:51
  SPEAKER_00: Where would you say things currently stand with Tezos X right now?
• 6:56
  SPEAKER_01: So today you have Ethelink.
• 6:58
  SPEAKER_01: So Ethalink is this uh EVM interface to uh to Tezos, but it's a way um to come with your uh Solidity code base and deploy on Tezos um with um the usual tools, tooling, all the services that you are expecting.
• 7:23
  SPEAKER_01: It's really uh similar to uh in terms of experience to what you could find with uh let's say Arbitrum One or uh Ethereum or um any um EVM compatible chain, except that you are on Tesos and especially on Tesos X.
• 7:42
  SPEAKER_01: And on Tesos X, you will have from the EVM interface the capacity to interact atomically with other runtimes.
• 7:51
  SPEAKER_01: So typically we are developing uh um another um interface uh called Teslink, or uh it's really about uh allowing people that have been used to deploy on the layer one to deploy on the on this rollup, and they will immediately uh have um uh a benefit from that.
• 8:16
  SPEAKER_01: Um the capacity of the system is uh larger, so the the cost of transactions is is uh cheaper.
• 8:24
  SPEAKER_01: The um pace at which blocks are created is also uh far higher than the layer one.
• 8:32
  SPEAKER_01: So today the layer one produces blocks in six seconds, while Etherlink is producing one uh every uh 500 milliseconds.
• 8:44
  SPEAKER_01: Um so with Teslink you will get uh the same uh the same um uh benefit and an additional one being in the same rollup as uh Etherlink, the EVM compatible layer of Tesos, you will have the possibility to atomically use other smart contracts that are deployed on EVM, on EtherLink.
• 9:09
  SPEAKER_01: So it means, for instance, that you can have access to the TVR uh that is currently deployed on Etherlink from uh a Tesos uh smart contract.
• 9:20
  SPEAKER_01: Um so all this actually creates um a completely new um environment both for uh EVM um uh people from the EVM world and people from the Tezos world.
• 9:39
  SPEAKER_00: What are some of the more interesting challenges or unknowns you and the team are navigating right now?
• 9:46
  SPEAKER_01: Oh there are many, many, many uh uncertainties because we are building something that never existed.
• 9:53
  SPEAKER_01: So uh you need to uh to be very flexible in the way you uh enter this uh this work.
• 10:01
  SPEAKER_01: Um maybe one thing uh that um um we need to reinvent is the way um um environment uh to execute smart contracts interact with each other.
• 10:18
  SPEAKER_01: So today if um you are in um let's say in an EVM chain and you want to interact with another chain, you have uh several possibilities, um, but in the end it's going through a bridge.
• 10:36
  SPEAKER_01: Okay, so it means there is something external of chain that will orchestrate an interaction between two chains and more precisely between two smart contracts running on two chains, two different chains.
• 10:50
  SPEAKER_01: And you're not really in control, you can have some expectations, you you can but but it the control is really outside the chain.
• 10:58
  SPEAKER_01: Okay, uh with Tesos X, we are um building um an environment where you are completely um as a as a smart contract in control of this interaction.
• 11:13
  SPEAKER_01: So the interaction itself is on chain, it's an atomic interaction.
• 11:17
  SPEAKER_01: So to be very concrete, it means that uh if you are uh an NFT marketplace on um on the Tesos runtime, you can um in the same transaction go inside Etherlink, um I don't know, uh do something with uh some DeFi protocol, or uh you can uh also uh interact with another uh NFT on on Etherlink and come back to your smart contract inside the same transaction.
• 11:54
  SPEAKER_01: And you are in full control, you don't have to trust a bridge that will work or not, or maybe uh take some time to get the finality on on both sides.
• 12:02
  SPEAKER_01: No, you are inside the same uh environment.
• 12:07
  SPEAKER_01: And to us, it means um um actually uh that we are solving a problem of interoperability between environment and also a problem of composability uh between smart contracts.
• 12:26
  SPEAKER_01: So um at the end of the of the day, we believe that um Web3 developers will have a lot of opportunities um seeing how to get this to world interacts.
• 12:42
  SPEAKER_01: And what's more is that we are building something that is generic.
• 12:46
  SPEAKER_01: So it means that when we will add more runtimes, let's say a runtime that run on JavaScript, Python, or uh that is capable of running Solana smart contracts, the same uh protocol will be used to have this in atomic interactions between smart contracts.
• 13:07
  SPEAKER_01: So we believe it it could it could solve a lot of problems that uh are today not uh solved in a satisfactory way by bridges.
• 13:21
  SPEAKER_00: When you think about what Tezos X unlocks, what do you think or who do you think benefits the most from it?
• 13:31
  SPEAKER_01: Um yeah, it's a very good question because uh you know today um we can see that um a lot of um WebScript developers are more or less interested in multi-chain uh deployment, in um also um um I mean the the the the ability the ability to uh to interact with uh defined applications uh in very smooth way.
• 14:08
  SPEAKER_01: So we believe that uh uh these people will be uh will find with the sex a very um um comfortable environment where the assumptions about security are uh very clear and simple, so that they don't have to ask ourselves to too many things.
• 14:27
  SPEAKER_01: And if we believe, that's probably what the future is about, uh, that more and more of the code running on blockchains will be coming from AI agents, but putting in place um an environment that is simple in the ways that it works, in the way uh its security assumptions are formulated, could allow us uh to um validate what this AI agent uh will deploy on blockchain.
• 15:01
  SPEAKER_01: And that's that's extremely uh important because you know that one of the challenges that we will have now is not really um the ability to generate code for uh smart for WebScript environment, but more the capacity to put in place sufficient uh validation processes for this uh for this code.
• 15:28
  SPEAKER_01: Uh so producing code is not anymore a problem.
• 15:32
  SPEAKER_01: And you may uh remember that uh one of the value propositions of Tezos from the very beginning is formal verification.
• 15:41
  SPEAKER_01: It's the ability to take a smart contract written in Mikkelson and get a mathematical proof of the smart contracts.
• 15:51
  SPEAKER_01: What does that mean?
• 15:52
  SPEAKER_01: It means that the smart contract of its code, you have its specification expressed in very specific way on machine with uh a lot of mathematical definitions, and you say, okay, these two are compatible.
• 16:08
  SPEAKER_01: Okay, and it's not just because you look at it, it's because you have another program that checks that and will tell you yes, this is a bug-free program because if you look at this code, if you look at the specification, this code is uh implementing this specification.
• 16:30
  SPEAKER_01: And I have a theorem that gives me that.
• 16:34
  SPEAKER_01: So I think one potential uh um value that we may underestimate today is will be the ability on um uh on Tesos X to reuse all this machinery to get uh the highest possible level of uh uh certification on uh code produced by AI agent and running on a blockchain.
• 17:02
  SPEAKER_01: So that's one thing that I'm really thinking about these days, uh, looking at the AI uh revolution of code agents and say, okay, maybe we have something to propose here that uh uh will be far more difficult to get in uh, for instance, EVM uh environment, because from the beginning Mikkelson has been sought to facilitate mathematical proof.
• 17:28
  SPEAKER_01: Well, of course, you can do some mathematical proof about EVN programs, but yeah it is it has not been designed to do that.
• 17:35
  SPEAKER_01: So it's far more painful.
• 17:40
  SPEAKER_00: Let's talk latency.
• 17:43
  SPEAKER_00: What does that word mean in this context and why does it matter?
• 17:49
  SPEAKER_01: So latency is um the time it takes uh uh for a hand user to submit a transaction and get a receipt that its transaction has been um considered inside the box.
• 18:05
  SPEAKER_01: And I'm using here a word that is a bit uh uh a bit blurry for uh for good reasons because um you you you want to know if uh your transaction has been included.
• 18:21
  SPEAKER_01: So it has it will be in the ledger, I mean in the in the blocks somewhere.
• 18:28
  SPEAKER_01: You it it can be enough for you.
• 18:30
  SPEAKER_01: Okay, I know that it is included, and uh given uh what I know about the state of the blockchain, it's fine to me.
• 18:38
  SPEAKER_01: I can I can I can uh compute uh what what it gets to me and uh what what it means to me on on my assets, on other assets, on my interactions with some smart contracts.
• 18:53
  SPEAKER_01: So that could be enough for you.
• 18:56
  SPEAKER_01: Or you could be looking for something a bit stronger and say, okay, uh, I not only want to uh to to know that my transaction has been considered by the by the block producer, but I want to make sure that there is a settlement from the network that this uh block is uh is final, it is written in stone.
• 19:20
  SPEAKER_01: Not only my transaction has been considered, but it has been uh globally accepted, globally finalized.
• 19:28
  SPEAKER_01: So latency uh first, if you want to talk about latency, you need to decide which what what kind of uh um guarantee uh you want on your transaction.
• 19:42
  SPEAKER_01: Um and um today uh for a long time with layer one first for for a long time for layer one protocol, we um really uh used a consensus algorithm to wait for finality and make sure that the transaction was finalized to consider it uh to be good for us.
• 20:07
  SPEAKER_01: And with the advent of uh sequencer, so you know this uh centralized um uh point where you submit your transaction and you know the the sequencer has a primary role in the process.
• 20:24
  SPEAKER_01: It is the block producer, it is the one that will actually create the chain.
• 20:29
  SPEAKER_01: Uh since we have moved to this model, now you have this possibility to um, in some sense, accept this centralization aspect first um and and get the latency that you would expect from a centralized exchange.
• 20:50
  SPEAKER_01: So something that can come back immediately to you and say, no, no, we don't have to wait for the consensus.
• 20:56
  SPEAKER_01: I'm as a coincidence, I promise I'm doing nothing wrong.
• 21:00
  SPEAKER_01: I'm just putting your uh your transaction in the sequence and I will produce a block.
• 21:06
  SPEAKER_01: And this block will be uh checked by many um other nodes, but I can immediately tell you that it will be injected, and this will be the result of this injection.
• 21:22
  SPEAKER_01: And if you accept that, which is not um um too too complicated in the case of uh Tesos, because the Tesos X sequencer is something that is very open, that is governed, that um is uh behaving very well from the beginning, so people accept that, you get an answer that is far uh quicker, that comes very uh fast to you, um than what you had with a layer one and a consensus protocol.
• 21:58
  SPEAKER_01: So typically what that means is.
• 22:00
  SPEAKER_01: Essentially, you move from several seconds to several milliseconds.
• 22:06
  SPEAKER_01: And that's a game changer.
• 22:08
  SPEAKER_01: Of course, that's a game changer.
• 22:10
  SPEAKER_01: Because in terms of UX, it means that some activities, so typically trading is one of them, require you know very fast decisions and a very fast interaction, retroaction loop, okay, to be in place.
• 22:31
  SPEAKER_01: So typically you will need to be very close to the council, have your own node, do some actions, see immediately in a few milliseconds what the result of the actions, and immediately have another actions after that.
• 22:47
  SPEAKER_01: So that's the kind of interaction that you can now have with the blockchain, and that is unlocked by um this uh instant uh confirmation.
• 22:57
  SPEAKER_01: So latency um to summarize because I'm talking uh very for very long, sorry, today.
• 23:07
  SPEAKER_01: Um yeah, uh I I hope so.
• 23:10
  SPEAKER_01: I hope it's very clear.
• 23:12
  SPEAKER_01: Um so to summarize today for payment, it's it's fine to use um um an environment that uh just um um you know, just like etherling produces a block every 500 milliseconds, it's fine.
• 23:29
  SPEAKER_01: You get your answer in less than a second.
• 23:31
  SPEAKER_01: It's it's it's it's enough.
• 23:34
  SPEAKER_01: But now that we have implemented on Etherlink this notion of instant confirmation, you can get your answer in uh few dozen of milliseconds or um maybe less in the future.
• 23:47
  SPEAKER_01: Um this unlocks um some form of very uh high, I won't say high frequency, but very fast uh trading uh opportunities for uh for users.
• 24:02
  SPEAKER_00: How much of the work on latency and confirmation ties back into the bigger Tezos X picture?
• 24:10
  SPEAKER_01: It's interesting because it was not planned uh from the beginning.
• 24:14
  SPEAKER_01: Yeah, we we didn't uh uh plan to work on that, mostly because um we didn't see any demand.
• 24:26
  SPEAKER_01: And um we were uh with uh Thomas Leton uh one of the tech lead of uh of Tezos uh at the test dev, you know, this um uh event uh where we hit uh the the community and builders and people that are interested in Tezos.
• 24:44
  SPEAKER_01: And um I remember we were talking about uh with uh the the people of uh Andrew protocol, and they told us, oh, do you think that we could have this um instant confirmation on Etherlink, like uh 10 milliseconds or even five milliseconds?
• 25:01
  SPEAKER_01: Uh because it could open up a lot of uh use cases on our side, and we were you know uh just discovering this need, and I think we immediately started discussing it and understood that it was perfectly compatible with our architecture and the way the node is implemented, etc.
• 25:23
  SPEAKER_01: So, well, gave it a try, and actually it worked uh very um very well uh almost uh at first try.
• 25:34
  SPEAKER_01: So, yeah, that's uh you know, the ability to be agile.
• 25:38
  SPEAKER_01: You of course our roadmap has been there for two years, so you could say, Oh, they have a lot, it's very ambitious, they they they have to focus on what they need to deliver.
• 25:47
  SPEAKER_01: Yes, that's true.
• 25:49
  SPEAKER_01: If you want to deliver, you need to focus, but sometimes longing fruit are not that far from you, so just have to grab them.
• 25:57
  SPEAKER_00: Very true.
• 25:58
  SPEAKER_00: As you look ahead, what are you most curious to see play out once these pieces are in place?
• 26:07
  SPEAKER_01: Um many things.
• 26:10
  SPEAKER_01: Um I will uh be very interested in in uh looking at um um how you can take, for instance, a solidity code base, so something deployed on uh on Ethan, and say, okay, now we want to be serious at some point and uh and certified uh this piece of work.
• 26:41
  SPEAKER_01: And uh I think we will have opportunity to just uh take the existing contracts, translate them to Mikkelson, prove them uh correct, and gradually uh um uh not translate, migrate uh an existing smart contract system into a certified uh smart contract system with uh mathematical proof to ensure that there is absolutely no bug.
• 27:12
  SPEAKER_01: And I'm super curious about that because I think for a long time uh all these uh formal methods and verifications were you know very costly because you have to you had to put a lot of brains working on the mathematical proof, and uh well, it takes time to write proof.
• 27:30
  SPEAKER_01: Now that uh we have coded agents, it's uh it's a game changer, and I think that's something that um uh we we will see on on Tesos X, uh, because the system is simply uh, in my opinion, the only one today to uh to allow for such uh such uh gradual certification of smart contract system.
• 27:52
  SPEAKER_01: So that's one thing.
• 27:53
  SPEAKER_01: Another will be um for the Tesos people, Tesos community to um um exploit the opportunities uh that they will find on on EtherLink, because on EtherLink, you know, there are many uh very um interesting and uh popular uh uh platform, uh DeFi platforms that have been uh deployed there.
• 28:18
  SPEAKER_01: Uh so it means a lot of liquidity.
• 28:21
  SPEAKER_01: We also have uh our own answer to uh our propositions in terms of uh as a as a as an ecosystem in terms of DeFi.
• 28:30
  SPEAKER_01: And uh yeah, I'm curious to see how these two worlds will interact.
• 28:35
  SPEAKER_01: So that's uh at least two things.
• 28:38
  SPEAKER_01: Um yeah, there are many others, like for instance, uh the future of Tesos X, because you know we are focusing on these two runtimes right now, so the FLINK uh EVM compatible runtime, the Teslink, uh Tesos uh Mikelson compatible runtime, their atomic interactions.
• 28:58
  SPEAKER_01: But we will add more uh runtimes in the future depending on what people need.
• 29:04
  SPEAKER_01: And um, yeah, I'm curious to see if uh people will ask for a solar runtime, let's say, to stay inside the blockchain world, or more general on time, like a Python or JavaScript runtime.
• 29:20
  SPEAKER_01: So that's that's typically uh uh yeah what I'm curious about uh these days.
• 29:27
  SPEAKER_00: Well, Jan, thank you so much for taking the time and for walking through all of this with us in a way we can actually follow.
• 29:35
  SPEAKER_00: So thank you.
• 29:37
  SPEAKER_00: If you're watching and you want to keep up with Tezos X and the work we've talked about today, keep an eye on nomadic labs updates and the usual Tezos channels.
• 29:47
  SPEAKER_00: That's it for this episode of Tez Talks Radio.
• 29:50
  SPEAKER_00: Thanks for watching.
• 29:52
  SPEAKER_01: Thank you,