Ethereum and Starknet in 2024

Share this post:


Let’s discuss Starknet, Ethereum, and the road ahead for 2024! Join Eric Wall for a special discussion with Eli Ben Sasson, Avihu Levy, and Liron Hayman to discuss Starknet being unaligned, what’s next for Cairo, and which milestones will shape 2024.

Transcript

Eric Wall:
I think we’re live. Yes, we are live. I choose the wall as my background because my name is Eric Wall. Happy to have you guys here. Liron, head of business development… Correct me if I’m wrong, but Liron Hayman is the head of business development. Avihu, the head of product and of course Eli the CEO of, now quite recently, the CEO and co-founder of Starkware and also board member of Starknet. How are you guys doing today?

Avihu Levy:
Thank you for having us.

Eli Ben-Sasson:
Pretty good, how are Eric?

Eric Wall:
I am in vacation in Costa Rica. So I am exceptionally well. I cannot… Yeah, would be strange for me to complain.

Eli Ben-Sasson:
Are the bricks of tungsten?

Eric Wall:
One day, not yet, this is a virtual background. So, I would like to, unless I got to anyone’s title wrong, I would like to start with a topic that has been circulating for the last couple of months about Starknet, and I want each of your guys’ opinion about this. Is Starknet unaligned And you can interpret that question as you wish, but do you feel that Starknet is somehow unaligned? And when I say unaligned, I’m talking about Ethereum, the Ethereum ecosystem, the Ethereum technology. So, I don’t know. Eli, do you want to go first?

Eli Ben-Sasson:
Yeah, I definitely want to go first. So, I’m right now in the middle of this amazing biography of Einstein by, I forgot, it’s like a recent one. The same guy who’s now writing, he wrote the biographies of Steve Jobs and recently of Elon Musk, so he also wrote one on Einstein.

Liron Hayman:
Walter Isaacson.

Eli Ben-Sasson:
Einstein was deeply unaligned with physics and everything and so on. And I certainly, as a researcher was unaligned with a whole bunch of things. I think unaligned means thinking about what is actually important and where are you heading and saying this is where we need to go, or this is what makes sense and to the extent that the world is aligned with you, then you’re going to be aligned. I don’t know if this is a good answer. So we are aligned with the long-term vision on what integrity webs are, what Starknet is, what integrity means through our technology. I think to a pretty large degree we are aligned with what Ethereum aspires to be and stands for. But it’s far more important to be aligned with the vision and goals than to answer the question of what exactly Ethereum is aligned with and are we aligned with that?

Eric Wall:
Avihu?

Avihu Levy:
Yeah. I’d say that we are not, we don’t love or we don’t like to change things, but when we feel it’s needed, we are not afraid to change things. So, we’ll go with everything that is, we will definitely participate in all the discussions, we will lead all the innovation and development and then, when we recognize there is an important goal and it requires a change, whether it’s proving capabilities for scalability, account obstruction, I can name probably 10 different things, we will go the unaligned way and we will change them, but it doesn’t mean really unaligned, it just means that we are taking the direction of getting a 100% of our targets and we do that while improving Ethereum.

Eric Wall:
Liron, do you feel like you want to weigh in here?

Liron Hayman:
I’ll tell you a funny anecdote. Jeff Bezos says he always has to speak last at meetings, because otherwise, people will just repeat what he said. So if you told me to speak after Eli and Avihu, I’m a bit scared. But I think that there’s a misconception in the community that alignment means copying the same stuff that already exists. So, something I often say is, one of the boldest decisions at Starkware was to develop Cairo and not use Solidity to develop the Stark proofs. And that’s done precisely because we want to scale Ethereum.
Because of the importance of getting 100x, the throughput of Ethereum at 100x, the decrease in the costs. You have to make bold decisions like developing a new smart contract language and that, at first glance looks like unalignment, but it’s actually the most aligned thing. In contrast, some teams are so worried about alignment that they are so committed to EVM and Solidity that they then delay stuff like fraud proof implementation, which is a bit of a shame. So alignment, yes. Does that mean that doing exactly the same as everybody else? No.

Eric Wall:
Okay. So what I’m hearing is that your view collectively is that you are aligned with the goals of scaling this massive world computer that is Ethereum and making sure that the world, the billions of users across the world, are able to access the system, do the computation, run their transactions, and that is how you definition alignment. A goal aligned with the long-term goal of making this technology, realizing its potential. But I do want to ask some more nitty-gritty specific questions there. I completely agree and understand what you said Liron on about, alignment should definitely not be measured to the extent that you’re using Solidity or how EVM, how much of the EVM you have copy pasted and how much of it is compatible or equivalent. I think Vitalik himself has even said in conversations that, if he could start over from scratch, it’s not necessarily that he would use Solidity or the EVM.
Again, of course there’s been lots of learnings and if you have the ability to create and execution environment today, then you would make it different. So, copy pasting the EVM when you’re Layer 2 and you have this amazing possibility to do things like account abstraction and build a virtual machine that is more easy to see or knowledge proof. I think that he has said that he’d even consider strongly using something like Cairo as the native language of Ethereum itself. So, being aligned with Ethereum shouldn’t mean to inherit the technological debt that Ethereum has from many years back. I don’t think that even Ethereum is aligned with Ethereum today from a technology perspective, but they made some choices and they’re now to stick with it.
However, there is one thing though that I think also touches on the heart of the alignment question. There are many different Layer 2 systems out there on Ethereum and Starknet is one of these Layer 2s that [inaudible 00:07:53]. Nearly all of them have their own token, right? Some of them use it as a governance token, but in Starknet, the Stark token is also used as the fee token. So, when a sequencer sequences a transaction, it requests payment in the form of a Stark token. Is this a problem for alignment or are there some strengths with having your own fee token that you guys want to [inaudible 00:08:22]. What was the core rationale behind this argument of making also the fee token be Stark instead of being ETH and whoever wants to start can start.

Eli Ben-Sasson:
I suggest Avihu start.

Avihu Levy:
I have a very nuanced thing to say about it. I mean I’ll leave for Eli. The larger…

Eli Ben-Sasson:
The less nuanced.

Avihu Levy:
No, the big picture about how an ecosystem should look like and so on. I just want to comment on a very specific point, maybe two things. First, we’ll make sure that the user experience in Starknet stays very good even with ETH, and in particular, that users will be capable of paying not just with Stark but also with ETH and very soon, not just with STRK and ETH, but generally speaking with every token they’d be interested paying the fees. ETH, what will happen behind the scene is that what sequencer will get, what operators will receive will be eventually the STRK token. Now one very nuanced thing about that is that we have this program for developers, we call it Devonomics, where basically fees, at least for now, we are running this some sort of experiment. I mentioned earlier that we like to test all kinds of new things, where fees are not going just for payment to the network operators, but also we want to take some part of the fees and give it to developers. Which developers? Developers of the most used applications and developers of the best and most important infrastructure.
And when I say we want to, in fact you already did one round, we gave several millions of dollars for dApp developers on top of Starknet. And I think that one small but important thing you get from paying fees in STRK token, and this is a small thing, but it reflects on everything, is that those developers, if they receive dollars with they receive STRK token, there is a different value in this meaning, they can now affect what is happening in the ecosystem much more if what goes to them is eventually the token of the ecosystem.

Eric Wall:
Right. I have another question. I don’t know how techy we want to get, but I do have unrelated to the subject, so I want to cover it now before we move on. Is there something about the specific cryptography choices that Starknet makes, that makes it more difficult to be aligned with Ethereum because I know that there’s this big difference between hash based zero-knowledge and curve based zero-knowledge or just hash based cryptography and curve based or KZG based cryptography. And it seems like in the Ethereum domain when they’re moving to Proto-Danksharding, they’re leveraging a lot of this KZG type cryptography, which is very different from the STARK and hash-based cryptography. And I believe that it’s also the case, and please correct me if I’m wrong here, that it seems like there are more pre-compiles that Ethereum favors that are in favor of this KZG commitment. So is there something on the cryptography level where Starknet has to fight a little bit and have you thought about, for example, doing something like verifying the STARK proofs using some type of KZG verifier? So that question is directed to Eli.

Eli Ben-Sasson:
Yeah, terrific question and I think it goes back to the previous question of what are we aligned with. So, on this thing I would say our first and foremost alignment is with math, integrity and efficiency. And on all of these fronts, as I’ve said many, many times across many forums, and by the way on Thursday in two days there’s going to be a STARK @ Home with a bunch of KZG experts, Zac Williamson, Mami Ratsin from Tyco, and I mean Zach of course of Aztec and Jim from Binius and Shahar Papini and I’ll be there speaking about FRI. I’ve spoken about this many times, but let me say it once again. There is one advantage of pairing-based systems in our context and that is that the communication complexity, the length of the argument, the number of bytes is smaller by roughly a factor of one to two orders of magnitude.
The length in bytes is shorter, but on every other front, when you look at math or engineering or numbers, hash-based STARKs over small fields dominate. Much faster proving, much faster verification, much safer, fewer cryptographic assumptions, much greater scale, everything is better. So again, what do we want to align with? I want to align with long-term future-proof techniques. I want the safest math, I want our users to have the safest math. Having a trusted setup, even if it’s vetted by a very worthy set of people, is not as good as having something that requires no trusted setup. Now, it is true that I think mostly for historical reasons because SNARKs in Zcash became popular under groth16 with a curve and I think Ethereum led the way on that front in allowing something that was completely inefficient to verify on the EVM, which is a pairing in a SNARK, adding a pre-compile for that. And for historical reasons, that particular curve and setup got a subsidy.
I think it would be very unfortunate if this thing becomes enshrined and you need to align with this thing because it just leads for systems that are less safe. There’s a single point of failure, it’s not quantum resistant, it enshrines not just a very particular proof system, but even a very particular setup and stuff like that. So it’s not a good idea. And, just based on, it’s not about alignment, it’s just based on math, security, stuff like that. But I’m very optimistic that Ethereum led by visionaries like Vitalik and Justin and others, they understand that they want Ethereum to be the place where all proof systems have a level playing field and that you don’t have a single point of failure or one, you don’t have to fit the squares through circles or vice versa. So, I’m very confident that Ethereum will once again lead the way and find a way to allow multiple proof systems to play on a level field and not force them all through one system that just leads to worse security.

Eric Wall:
Avihu, did you want to [inaudible 00:16:32] in?

Avihu Levy:
Yeah, just to repeat what Eli said, at the end Ethereum wants to be rollup aligned, and that means that Ethereum wants to be data availability aligned, but also proof verification aligned. So that means no to very specific set of parameters and cryptography, yes to enabling all proof systems to be very efficiently verified on Ethereum, and I’m sure we’ll get there.

Eric Wall:
Okay, let’s move on. Or actually let’s take a retrospective look on the year that has been 2023. What were some of the most significant moments in your view about the year that it has been, it has been an eventful year. A lots of things has happened, but when you look back on that year, are there some significant moments that stick out to you that feel like milestones or surprises or something that maybe you want to remember people of, something in particular? And maybe we’ll start with Liron this time and we’ll work ourselves down and you’ll get to speak last Eli.

Liron Hayman:
I should have prepared answers. Maybe not a specific event, but something, a trend that definitely changed throughout, 2024 or 2023? I think 2023 you’re referring to, is that when I joined Starkware in mid 2021, and back then it wasn’t at all clear that what we call zk-rollups, validity rollups were production-ready or going to overtake optimistic rollups in terms of adoption and production readiness in the near future. And I think what we saw last year is a flipping of that. Even Vitalik made a few comments that he thought that optimistic rollups would be further ahead of zk-rollups, and that flipped.
Maybe I’m forgetting teams, so sorry, but risk zero Polygon with all the different ZK efforts, ZKSync, Starknet, it has become clear that the end game is significantly closer than we thought it was. And that’s through a bunch of different events with all these milestones that these teams reached in 2023. I don’t want to cause a fight by saying this, but you do look at some of the biggest optimistic rollups not even having fraud proofs in production, let alone decentralizing the ability to submit those fraud proofs, and then you’ve got multiple different teams running zk-rollups at scale submitting proofs. That’s a huge achievement of 2023, all of that happening while there’s a broader macro bear market.

Eric Wall:
Avihu?

Avihu Levy:
Many things, but I’ll just talk about one. I think in terms of scale throughput for L2s and Starknet in particular, we started the year somewhere with demand of a friction of a TPS and capability of let’s say one TPS. And we ended up the year with days where we went over 10 TPS and the ability to do a 100 TPS. Which is roughly two order of magnitudes in terms of both demand and supply and I sure hope that we see another one or two orders of magnitude demand this year, but we’re definitely going to see improvement of at least, I don’t want to commit too much, but I am sure pretty sure that we’re likely to see even close to an order of magnitude again in throughput this year. So if we did up to a 100 TPS by the end of 2023, I hope that you’ll see few hundreds at the very least sometime this year.

Eli Ben-Sasson:
I want to start with the challenges and then go with the things that I found surprising in a challenging way and the stuff that I found then surprising in a very positive way. So, along the challenges, first I’d say that there are many competitors now. When we started six years ago and we said we’re going to use STARKs and validity proofs to scale Ethereum, no one understood what we’re talking about. Now everyone is building their own validity proof systems and they’re doing tremendous job in the context of SNARKs and STARKs and KZGs and whatnot, it should be noted that all of the leading validity rollups actually use STARKs and FRI some of them again because of this subsidy given to a very specific groth16, squeeze it at the very end through that but, all of the scaling is done basically by the technology that we spearheaded and predicted.
But on the challenging side, projects like ZKSync and the Polygon systems, they’re very good and they’re using EVM and allowing very easy deployment of existing solidity code, and it’s surprising, I thought it would be more challenging to do that, but they’re doing a very great job there. So, on the positively surprising side, I would say two things that are external but are very heartwarming. The first is the realization of many developers that Cairo is just a fucking amazing language to write smart contracts in. So, it’s not just that it’s a good language if you want to scale your code with validity rollups, it’s actually a next generation programming language for smart contracts, that’s very surprising, positively surprising. Related to that is the fact that we are the ecosystem with the largest increase in developers. It’s not something trivial, it’s something that actually many folks were saying, you’re trying with Cairo to boil the ocean and developers are not going to come, it’s too hard and the language isn’t ready and everything should be EVM based.
I think the developer ecosystem is proven the world wrong and this is very, very positively surprising. And the last aspect, which is related to what is called unalignment, but I call it being aligned with truth or where we should head. Things like native account abstraction, multi-calls, the whole notion of Cairo and Starknet. They have empowered developers to build amazing UX for end users. So, only on Starknet or all wallets basically with an interface that is Web2 like, and this is a fact that is not well enough known by people but when the masses are going to onboard onto blockchain, one of the questions will be, where is it easy and yet safe to onboard onto. On Starknet, all of the leading wallets, Braavos, Argent have very Web2, using biometrics, things like that. And this is because of being unaligned or rather being aligned with future-proof technology. So again, that’s another very positive surprise that is converted into very good UX.

Eric Wall:
Got you. So I’m hearing the narrative that zk-rollups would be completely outrun by optimistic rollup, that has slipped. I’m hearing that we are seeing orders of magnitude, multiple orders of magnitude increase throughput that the system capable of handling, and I’m hearing that there’s also been this dramatic change in the perspective of Cairo. Cairo used to be this language that was seen as something that was clunky, but it has changed. There was an upgrade, the Cairo language was upgraded. So, it’s not just that Cairo all of a sudden has become awesome, it technologically has changed. So it used to be Cairo Zero, then there was Cairo one, now it’s just Cairo I think. And apparently it’s one of the fastest growing and one of the largest developer ecosystems. So, we have three great things, zk-rollups really gained a lot of surprising strength, and this is surprising even to me, strength against optimistic rollups.
Maybe it’s even possible that many of these optimistic rollups will become zk-rollups to catch up or something like that. And we have throughput and we have a language that developers actually love instead of absolutely hate. That sounds like absolutely incredible year, amazing progress. But what about some of the things that I think Starknet has gotten some criticism for during the year? I remember vaguely, and I wasn’t hands-on during that time myself, so I don’t have full perspective on what it was that happened, but there was something around a regenesis moment when there was something that, I don’t know if it didn’t work smoothly or if there was some small amount of funds that was locked into a contract.
Can you talk a little bit about what was this event? Why were some people angry? Was there just confusion on Twitter or what was the reality of the situation? And is there some lesson from this that you guys feel that you had to do better on in terms of communication or is there something just process wise that needs to change? And that question goes to anyone who feels like they want to answer.

Liron Hayman:
I can try and tackle this. When you go to Mainnet as a cutting edge technology, there are risks involved. It was very clear to the community, that is true even with something like this, with the specific thing that you’re talking about with the regenesis and the fact that V-Zero transactions would no longer be supported by the protocol. Firstly, it’s important to understand the technical reason for this. Basically with account abstraction, now every single transaction goes through two processes. One process is validating the transaction, the second process is executing the transaction. And those V-Zero ones didn’t have that breakdown between validate and execute.
So you go live, you realize you have to have the structure. Why do you need the structure? This structure basically facilitates account abstraction because now with account abstraction, you can have absolutely whatever you want in that validate function. It can be the face ID from your phone, assign some hardware in your phone, and then the signature from the hardware gets sent to the validate function. At the protocol level of Starknet not to some off-chain verification.
So, there’s this question now as a protocol, you have some wallets that are deployed with V-Zero, but you want looking to the long term as this thing gets out of alpha to be generally available on working at scale with account abstraction and you make a calculated decision that there’ll be a few months now where people will be able to transact with those wallets, but eventually there’ll have to be a cutoff date and then close monitoring of the exact damage. The unfortunate thing with Twitter, especially with anonymous accounts, is that the people can make a lot of noise when there’s actually not real source for concerns.
So here I think you have a few hundred thousand dollars across roughly the same order of magnitude of accounts. I can pull up the exact numbers after here, but let’s say it’s like $400,000 across 300,000 accounts. So you’re looking at the average of a $1.30 per account, which is unfortunate, and I don’t want to say it’s not a lot of money, but there’s this decision here, you have to make this calculated decision of, V-Zero transactions will no longer be supported, there’s more than enough time for you to move across and then people have to deal with those consequences.

Eric Wall:
All right.

Eli Ben-Sasson:
Maybe I’ll add one thing to that, which is just that we have focused a lot of attention on building on developers. And as a general theme, we’ve always done this, because we’re building infrastructures and then the developers are needed in order for servicing the users. But in the end it is really about offering something that users will use. Generally speaking, we’ve become much more attentive to the needs of users and it’s something that definitely we’ve adapted to and we’ve learned and over the past year, and I think users are seeing this and we will definitely see it in 2024, this increased attentiveness to what end users need and want.

Eric Wall:
All right. So, moving our heads from the past and looking inside, into the future, you already said, Eli that developers are starting to change or have changed their opinions about Cairo, that it’s a great language to build on, but what are some of the things that are cooking for Cairo and I mean both the language and the VM, what’s in store basically for the year that’s ahead of us that people should learn about or be excited about, what are some of the highlights for the coming year? And anyone who feels like they want to talk first can do so.

Avihu Levy:
Yeah, I can start. So not everything is public yet, but let’s set the direction. I think in the first half of the year we are going to focus on two things. One is continue to push forward the performance. That means in particular, we’re going to roll out again parallelization in execution of Cairo and later, also faster execution by creating a separation in what the sequencer is actually performing and what is later being proven. So you can still use Cairo for proving, but in fact the execution of transaction would become much faster.
And the second, so that’s one performance. The second one is cost. First there is the famous for it for forthcoming, which will likely dramatically decrease state availability cost, which is the major cost per transaction on Starknet right now. And we have a couple of surprises coming there as well. So I’d expect three months from now to be in a even higher performance and a much lower cost. So that’s talking about three to six first months of the year.

Eli Ben-Sasson:
One thing that I just want to add that one thing that I don’t know the answer to, but I hope that will give some answers to is that, because we realize that some of the early brave or unaligned choices as to the core of Cairo have paid off very nicely in terms of UX. We’ll be exploring similar such things and themes. The analog of going full-blown account abstraction, so looking at or multi-calls or stuff like that. We’ll be looking for adding more such thing to the core base of Cairo so that again, developers can work their magic and have an even more amazing and unseen elsewhere experience for users.

Eric Wall:
And about improving cost and increasing performance, I did see some, just in the last couple of days I saw some tweets from the Starknet account about modularity, embracing modularity. Was I dreaming this or is there actually something happening here about making the Starknet stack compatible with using Celestia as a data availability layer? Is that a front that you guys are exploring? Is that something that you have some information you want to share? Because it seemed like there were just some small tweets, not that much information I could find. Is there anything that you’re ready to mention at all? Is Starknet going to open up the possibility to use other systems for DA?

Avihu Levy:
I can take that. We like to experiment and I think modularity is not just about data availability, it’s also very much about the ability to execute and to prove, and I think that if you want great clients that can publish data availability in high throughput layers, but at the same time also verify execution very fast, you have to rely on proofs and Starknet is the tech for it with Cairo inside. So, we are definitely exploring integration to many other data availability layers. Like I said earlier, the first one will be 4844 in public Starknet whenever it comes out.
In fact, we already tested goerli for that. We did announce that we also experimenting on the side with the Starknet and Madara. With Celestia’s data availability and we’ll push more the innovation on those fronts as well. So yeah, that’s probably another direction we’re exploring innovation and we are definitely looking into all kinds of things. I can mention Celestia, but I can also mention that we have been talking and still thinking of a mode that we call Volition that enables different data availability to be available for developers in the same system. So that’s in general something that we plan to do in Starknet as well.

Eric Wall:
Yes. Liron, you want to say something?

Liron Hayman:
Yeah, I think the whole concept in that you have one unit of measurement for all computational resources, whether it’s execution or storage or bandwidth, doesn’t make so much sense. I think even with 4844 with having a data gas and a regular gas, we acknowledge that and modularity, one of the benefits that it gives you that. Also, another comment worth saying just for the benefit of the newer listeners, the biggest bottleneck, bottleneck is a scary word, but the biggest challenge is data availability more than the actual scaling of the proofs. The proofs, we know what to do. Right now, roughly 90% of gas costs go to DA as pre 4844, it’s pre Volition, there’s a lot of stuff to do that, but a lot of people are surprised by that, they would think that data availability is a minor thing relative to proof verification.
But even looking at StarkEx app chains that we were running previously through stuff like Validium, by the way, Validium was the word invented at Starknet, I did not know that. Or more specifically by Avihu. That’s one of the biggest gas reductions comes from beyond just the… The proof verification comes from using Volidium DA rather than putting everything on chain. And there’s a lot of other stuff like there’s this whole debate do can zk-rollups just send state diffs onto L1 or send a representative of every single transaction on L1? And Starknet has opted just to do state diffs because it saves gas. So maybe it’s a point about, I think modularity helps you optimize each component separately and it’s a good thing.

Eric Wall:
Right so, Volition and EIP4844, is it a high probability that we’ll see both of these things affecting Starknet simultaneously or one after the other, but in total during the year 2024 or are you guys planning to wait and see what happens after EIP4844, or is Volition happening anyway?

Liron Hayman:
The second one.

Eric Wall:
What’s the plan? The second one.

Avihu Levy:
The logical one.

Eric Wall:
The logical one.

Liron Hayman:
One reason to explain that is just maybe repeating myself, DA is one component, but there’s a lot of other stuff that Starknet doesn’t have yet like fee markets and just general performance improvements that you want to have instead of just optimizing only for DA. So 4844 is a big enough lift. Volition’s another lift, but they’re both probably solving the same thing, you want to solve other stuff in parallel.

Eric Wall:
So, another thing that I’ve seen talked a lot about, and I want to give each of you maybe a chance to talk about this, is there seems in Starknet that there’s a lot of focus on this gaming use case. And I want to know to anyone who feels like they can explain it, what is the case for gaming being a good use case for blockchains and perhaps why is it good for Starknet in particular? Are there many teams that are building something gaming related on Starknet, and do you think this is something that is going to shape the ecosystem over the year or why is there so much talk about gaming in particular? Why gaming specifically?

Eli Ben-Sasson:
Who wants to go first?

Liron Hayman:
I can start. Okay, gaming. So a few comments, Vitalik’s whole life post about why we need Ethereum, his first one ever was describing exactly this use case of having self custody of your assets in games. And the industry evolved to this situation where you had the assets on chain, but actual game logic was off chain, and it’s this good step in the right direction, but it’s neither here nor there. A lot of the teams acknowledge that, it’s just a step towards the ultimate goal of on chain proof of actual the game logic. So, you basically have two camps in the gaming world, meaning equal, but in terms of the philosophies, one is we don’t need blockchains at all, but one is if we do need blockchains, we need everything, not just the assets but also the game logic to be running on chain.
And I think the only way you can get actual game logic running on chain at scale for the end users is through zk-rollups because of the scalability benefits and account abstraction, because otherwise you’re just going to have end users losing private keys. You’re going to have to… Issue like, you need session keys in order to be able to transact openly without signing each individual transaction as you make moves in the game. The last comment I’ll say is it needs to be proven on chain, it doesn’t necessarily need to be run on a blockchain and there’s a subtlety to that.
So basically you can attest to the validity of the game in batches. It could be a daily batch, it could be a weekly batch, and you just know that every single player played by the rules and didn’t get special treatments, didn’t get any special cheat codes, etc. And that doesn’t need to be run in real time with these blocks. Blocks can add overhead that is bad for user experience, but having some sort of a cadence of sending proofs that attest to validity, that is something that is strong part of the gaming community does want. Maybe I’ll pause then others can add stuff and actually talk about actual stuff [inaudible 00:42:18].

Eli Ben-Sasson:
Avihu, You want to go?

Avihu Levy:
Let’s just talk about games on Starknet. The bottom line is exactly the summary of what Liron said. You can execute tons of computation, very, very inexpensive. Games is a very natural use case where you talk about a lot of computation. It’s something that is very hard to run today on let’s call it traditional blockchain, and it’s very easy and is getting better every day, every month on Starknet and we definitely see a future not in five or 10 years, but very soon where tons of computation will still be very inexpensive to run and therefore games.

Eli Ben-Sasson:
I’d say that since you wanted all of us to answer my answer would be that you exactly see that the benefits of being something that is novel and next generation. Developers wanting to have on-chain experience, they understand that you can’t do it on Ethereum and you can’t do it on the optimistic [inaudible 00:43:31] and you can’t do it on the EVM, it won’t scale, long term, and if you’re going to invest so much time in building a game, building a proper game is like two years of work just on the front end or something like that of a large studio. So you need something where the engine can actually handle it. And the most plausible candidate is actually Cairo and Starknet. So, a lot of wise developers are saying since we believe on-chain gaming is the next thing, not just creating the NFTs and putting all of the logic off-chain, and I think they’re right.
They’re asking, okay, which is the platform where we can deliver this both now and also to scale to global scale? Because I mean, here’s the thing, all projects, they start small. So you’re building something, but your aspiration, your hope is that everyone uses this, that this is a game that will go ballistic. Now, those of you who’ve been long enough in blockchain may remember the CryptoKitties craze, which is, or then it happened with Axie Infinity. What happens if your system actually succeeds? Well, if you don’t have the right technology that can actually scale, it just won’t work.
And that’s what the successful projects have found out. So, all of the Starknet customers that came to us came to us because they succeeded and they were bleeding gas and they couldn’t service their customers. And the ones coming to us now are the ones with enough of a vision to see that if you want to succeed and you plan to succeed and you go past whatever, an airdrop or funding or a grant, you want something that will actually dominate gaming, you need to put it on an engine that can scale and Starknet is it.

Liron Hayman:
One thing we missed is just naming teams. I am always scared to do this because I always end up forgetting a team. So, if anybody’s listening, if you’re not listening, I apologize, my goal here isn’t to pick favorites just to whatever’s the top of my head. There’s an influence team that has actually a live alpha you should go and try influenceth.io I think. There is also cartridge.gg they are also running on-chain games. There’s a loot realms which sort of evolved from those, what was that called, those eight words that were worth hundreds of thousands of dollars in the last bull market. They’ve also done some on-chain gaming stuff. Briq the game, B-R-I-Q is basically Minecraft on-chain.
They’ve also formed a community called Dojo, D-O-J-O, which is building the tooling for on-chain gaming. They’ve got the sequencer called Katana, which uses the Blockifier sequencer that public Starknet uses, but it has some wrappers around that that are optimized for gaming, and also just has a strong community around that. That’s again using on-chain games, not just on-chain assets. And then, Topology is another team that’s built the Shoshin, S-H-O-S-H-I-N.GG that you can actually, I encourage people to just go to that website and just try and play the game, see what it’s like. Again, I might be missing some teams, that’s not my intention, apologies.

Eric Wall:
All right, one other thing that I wanted to talk to you about is, do you think, maybe not in the next year, but at some point, will there be some level of privacy inside of Starknet? I think a lot of people sometimes when they hear zk-rollup, some of them automatically assume that, oh, it must also be private because it’s zero-knowledge. It sounds like the system must have zero-knowledge of what kinds of transactions people are doing, but that’s actually not really the case.
There is information revealed about which account has which balance, which account is transacting with which account. The only thing that we have zero-knowledge about is how are those transactions being proven to be valid, and that is what is aggregated into this zero-knowledge proof. But, are you aware of any internal work that has been done to look into, like okay so, how do we go beyond just using zero-knowledge to scale but also make it encompass the privacy characteristics of the system? Is that something that you guys feel passionately about? Is there something that you want to see in the future or how do you think about this question of privacy in Starknet?

Liron Hayman:
I’ll take this first. The first thing I want to point out this, Eli will take this to his grave, that’s why he wants to call these things validity proofs and validity rollups and not zk-rollups, because they’re not zero-knowledge in the current state. Now the primary benefits of ZK of actual zero-knowledge, it’s two things. It’s the scalability because of this asymmetry between the verification of the proof and the actual size of the transactions, but also the privacy. The main pain point for the wider community right now is scalability. So you want to solve that first before you get to the next stage about privacy.
I want to make one comment about not so much privacy but more confidentiality. So, this is an important distinction and I think it finds a sweet spot where it makes more realistic applications for the blockchain whilst not being, I don’t know, I personally just get too worried about how the regulators and the incumbent bodies view privacy supported apps without getting to the philosophy arguments about that. So I want to talk about confidentiality for a second. If you run an app chain with a Validium data availability or Volition, but the user opts for Validium, the base layer doesn’t see the individual transactions and it doesn’t even see your individual balances. Those individual balances are stored with the Validium data committee, whoever that might be or however that’s stored. So, you can run a Starknet app chain with Validium data availability mode and go to Ether scan once approved and submitted to Ethereum.
Doesn’t expose that Eric now has 100 USDC or that Liron transferred money to Eric. But, if you go to the committee on that running that app chain DA, that committee knows the balances of the individual users and presumably has historical transactions to generate those balances. Now that’s a good sweet… Others may disagree, and this is my personal opinion, that’s a good sweet spot between privacy and full-blown transparency of the blockchain where certain permissioned entities have insight and visibility but the rest of the world does not.
So when you look about onboarding banks or just running a game, but I don’t need people in the game or trading platform. I don’t need random people who don’t know me to see my balance, my trades, but other people who do know me or who should see those trades do see those trades, it’s a good balance there. So, maybe I took it to the other direction, but I think it is an important benefit ability and that’s often not really highlighted, that’s confidentiality.

Eric Wall:
[inaudible 00:51:08] okay.

Eli Ben-Sasson:
I agree with what Liron said.

Avihu Levy:
I have one comment which is, we want to support also a bunch of cryptography. Maybe it relates to your previous question about what Ethereum supports. So in Starknet you have this flexibility, inside accounts to verify whatever signatures you want, but also we want to have developers and then users enjoy greater flexibility in terms of what are the cryptographic primitives they can use and how cheap they are. And we want to add later this year very, very inexpensive pairing verification so that all kind of known privacy schemes and privacy apps will walk out of the box and very inexpensive on Starknet. So I guess that will support privacy very much.

Eric Wall:
And before we start to wrap things up, Eli, you tweeted publicly on Twitter that you’re looking forward to getting grilled by Eric on a Twitter space. So I feel like to some extent I have to live up to this world and I’m going to try to ask you…

Eli Ben-Sasson:
Go ahead.

Eric Wall:
A question that tries to cut at the heart of some of the things that maybe it’s difficult for you to answer. The question that I have, and it’s been something that I don’t know myself the answer to. Sorry, I’m going to stop beating around the bush and get straight to it. So, sometimes I see news about stuff like Ulvetana, you mentioned that you were having a workshop. These are the same people behind the Binius system that seem to be making some advancements towards making the EVM a lot easier to zero knowledge proof or validity proof to use the terminology that you prefer.
And maybe so, at the surprising level, and we’ve also seen in the past, but happening in a more real sense now with the risk zero team allowing people to prove native Rust code very easily and provide validity proofs for that. Considering these things happening that we didn’t know about many years in the past and we’re seeing these advancements on making the EVM more easy to validity proven, making even Rust validity provable. How do you feel today about this big direction of Cairo? Basically my grilling question to you is, how confident are you today that it’s still worth it with Cairo given the opportunities that we sometimes see on tweets and threads and some small experiments?

Eli Ben-Sasson:
Yeah, so I’ll answer honestly, it’s along the way, we made a large number of bets and whenever you make a bold and courageous technological bet, it takes time to see whether you were right or wrong or somewhere in between and it’s too early. So, I could go on some marketing spree and promise and say that I know that we made the right choice, it’s too early. But let me just point out the two trade-offs. Well, the trade-off is once at least I thought of them and then something that I realized only very recently on the positive side of the choice that we made. So, there’s no doubt that building a VM that is dedicated to the constraints of your technology means you will utilize that technology better. This is true about CUDA and GPUs and it is true about Cairo and STARKs, okay? At the same time, it also means that it’s going to be harder to get the tooling infrastructure the developers care for.
And by the way, it’s the same thing with solidity, with respect to the Ethereum virtual machine or a blockchain machine compared to using whatever pre-existing programming language was there, right? Vitalik and Gavin Woods and whoever was writing the yellow paper also probably struggle with similar questions. Should we take the Java virtual machine and Java? Should we take Python or should we build a virtual machine for this new thing, which is harder. I think in that case it’s clear that they made the right choice definitely, even though now Solana with Rust maybe, challenges that. So even there, the jury is out. Now, those are the two sides of the trade-off. You will get more proving juice out of a VM that is built with proving in mind than you will get out of using some other VM. But the developer toolkit will take more time to build.
So now, which will win? Too early to say. One advantage that at least I didn’t realize early on, but actually is related to this UX and the developer experience in the user experience. By owning on a very deep level the whole stack from the virtual machine and all the tiers above that, as opposed to saying we take something standard, it was built somewhere else. The ability to offer new features, to innovate and to say, okay, this doesn’t make sense, we’re going to change it, is much greater. So for instance, our peers or competitors across the other L2s that are now embracing ERC-4337 with account abstraction, it is never going to give the same level of experience to the end users who are coming, why? Because they’re coming from Ethereum with the Ethereum standard, they already have their EOAs and their ledgers and their MetaMask in a certain way, and now offering them an ERC account abstraction as an additional thing is going to be a very, very problematic thing, and now you have two standards.
With StarkNet, everything is based on smart wallets, account abstraction, everything is already biometric key, biometrics secured, social recovery, things like that as a baseline. And so this is the ability to innovate and offer the developers and then the users, an integrated experience that is modern and better is something we didn’t realize. It’s not related necessarily to the proof systems. So summarizing, we made a bet, we made a large number of bets. We started by asking about KZG and groth16 versus Stark. There again, the question could be asked. We’re also being challenged every day on that front as well. There are new innovations on folding and stuff like that, and still we’re very confident our technological path will scale best. Same thing applies to Cairo. I believe we made the right choice, but time will tell and honestly, we don’t know yet.

Eric Wall:
Thank you so much for that answer. And I see we’re at least in my local time, we’re way ahead over the hour. Also, the recording time is one minute away from the hour cutoff. Liron, I think that you had another meeting that you’re probably late to, unless someone has something that they really urgently want to say before we close out, I want to thank all of you guys for your time and I’m excited for this next year to come.

Eli Ben-Sasson:
Thank you Eric.

Eric Wall:
Thank you Eric.

Avihu Levy:
Me too, thank you very much.

Eric Wall:
Thank you so much.

Liron Hayman:
Bye.

Eric Wall:
All right.