Interview with Eli Ben-Sasson: How StarkWare is Planning its Layer 2 Road
This time, we’re joined by Eli Ben-Sasson, co-founder and President of StarkWare, which is a blockchain privacy solution provider based in Netanya, Israel. Eli Ben-Sasson and Alessandro Chiesa, co-founders of StarkWare, are also Zcash founders. The main goal of StarkWare is to further popularize the groundbreaking ZK-STARKs blockchain privacy solution so as to “bring scalability and privacy to a blockchain near you”.
Recently we carried out an online AMA (Ask Me Anything) with Eli Ben-Sasson, who shared with us more about StarkWare and its development in Layer 2.
Unitimes: Hi, Eli Ben-Sasson. Welcome to Unitimes’ online AMA Events. First of all, please introduce yourself. Why do you build up Starkware? Would you mind sharing the development history of StarkWare to us?
Eli Ben-Sasson: Hi! My name is Eli Ben-Sasson and I’m Co-founder and President of StarkWare. Prior to co-founding StarkWare, I was a full professor in the Computer Science department at Technion, Israel Institute of Technology.
I’ve been researching cryptographic proof systems since 2001, when I did my post-doctoral research at MIT and Harvard University, and found ways to make such proof systems more efficient. I funded and led research of such systems since 2008 and in 2013 realized they can help blockchains with both privacy and scalability. In 2015 I co-founded Zcash (the company behind the cryptocurrency) and after further improving the technology, in 2018 it was clear we can solve blockchain scalability with it, which led to the foundation of StarkWare.
StarkWare has 4 co-founders, the three others being Mr. Uri Kolodny, our CEO and a long time entrepreneur (and close friend of mine for the past 30 years), Dr. Michael Riabzev, Chief Architect, who was my PhD student and is a co-inventor of the ZK-STARK protocol, and Professor Alessandro Chiesa of University of Berkeley, Chief Scientist (in West) who co-founded Zcash with me and is a long-time collaborator of ours.
Today, StarkWare has 45 employees, all sitting in Israel. We have raised over $100 Million of funding from Paradigm, 3Arrows, Alameda Research and individual investors like Vitalik Buterin and Naval Ravikant. Most of our team are engineers, and we’re building scalability and privacy solutions using ZK-STARK technology.
Unitimes: Last year, StarkWare has launched a L2 scalability engine named StarkEx for Ethereum applications, and has been adopted by several well-known dApps, including decentralized exchange DeversiFi, margin trading platform dYdX, and NFT platform Immutable X etc. Can you briefly introduce how StarkEx works and the advantages it brings?
Eli Ben-Sasson: The way trading works on Ethereum today (say, on Uniswap) is that the nodes of the network need to process each transaction sent to them, and because of the limited gas of the Ethereum blockchain, users must pay very high prices to have the transactions included.
With StarkEx, the Ethereum nodes need only check a single ZK-STARK proof that shows that thousands of transactions have executed correctly. The size of the proof and the time needed to check it are exponentially smaller than the time needed to check all transactions in the proof. This way, instead of paying 200,000 gas for a transaction on popular DeFi applications, on dYdX a single transaction costs less than 1,500 gas!
Unitimes: As we know, StarkEx can be deployed in two ways: zk-Rollup (publishing data onchain) and Validium mode (publishing data off-chain). For example, DeversiFi and Immutable X both use Validium, while dYdX uses ZK-Rollup. Can you tell us more specifically about the differences as well as advantages and disadvantages of these two l2 solutions? When developers’gonna adopt StarkEx, how should they choose between zk-rollup or Validium?
Eli Ben-Sasson: Yes. As you point out, with Validium the data is off-chain, kept with a network of data availability providers, and with Rollup it sits on-chain. This means that Validium costs less gas, per transaction (because less data goes on the Ethereum network) and it also has higher privacy (because no one but for the Exchange operator can see the individual orders that led to the state change).
In terms of security, users of Validium need to trust the data availability committee, i.e., that at least one member of it has not been compromised. With Rollups, no such trust assumptions are needed, but on the down side, transactions cost more gas (because data appears on-chain) and there’s less financial privacy, for the very same reason.
Unitimes: It seems that StarkEX is somewhat similar to the L2 solution launched by Matter Labs, where zkSync focus on data availability onchain and its upcoming zkPorter focus on data availability offchain. How do you think about this?
Eli Ben-Sasson: Indeed, zkSync is using many of our innovations — like zkPorter which is really simply Validium, and other things. The main advantages we have over Matter Labs are:
(1) we invented and build our own ZK-STARK systems, whereas their tech stack is built by others (PLONK, built by Aztec). This also means our mastering of the technology and ability to improve it are greater than theirs.
(2) We have already multiple systems running in production, using a Turing complete programming language called Cairo that is readily available. Matter Labs have only a simple payment system in production, and no Turing complete language available.
(3) Finally, our core technology — ZK-STARKs, is faster, safer (cryptographically speaking), transparent (no trusted setup) and post-quantum secure, while the core technology they use (which was built by a different team) is slower, requires a trusted setup and is breakable by quantum computers.
Unitimes: The ultimate goal of Starkware is to launch your own general L2 network StarkNet. According to Tom Brand, product manager of StareWare, StarkNet is expected to go live on testnet in the next few weeks and the mainnet later this year. Seems like that the L2 scalability engine StarkEx you have launched, along with the Turing complete language Cairo, the STARK Provers for generating proofs, etc., are all used as a prelude to the launch of StarkNet. Can you briefly introduce Cairo and STARK Provers? And how do they promote the launch of the StarkNet network?
Eli Ben-Sasson: Yes! StarkNet is a very exciting development. Think of Ethereum, it allows you to write and deploy any smart contract, and then others may send any transactions to it. That’s amazing! But due to the amazing success of Ethereum, the network is now clogged and it costs a lot to execute a transaction on it.
The STARK proving system exponentially shrinks the amount of time it takes to verify a proof for the correct execution of programs, so can be used *theoretically*, to exponentially scale the rate of transactions on Ethereum (and other blockchains), or, stated differently, to exponentially reduce the gas cost of each transaction.
The main problem thus far was accessibility to this amazing STARK technology. It was very hard to write programs in a way that would make STARK proofs for it efficient. It was also very hard to build and operate STARK provers. All this is about to change over the next week or two: StarkNet is permissionless, so anyone can write any smart contract and submit it to the StarkNet chain, just like on Ethereum!
Also, anyone can submit transactions to StarkNet, and they’ll be executed. But StarkNet transactions cost exponentially lower gas on Ethereum, due to STARK scaling.
Now, StarkNet smart contracts are written in a new programming language, called Cairo (just like Ethereum smart contracts are written in a special language — Solidity). The reason for this is that we need a language that optimizes the usage of STARK proofs and scale, and this is exactly what Cairo does. By the way, you can download Cairo at this website — cairo-lang.org, and you can follow StarkNet at this website: https://starkware.co/product/starknet/
Unitimes: It is expected that more L2 solutions will be launched this year and next year, including L2 networks based on Optimistic Rollup such as Optimism Ethereum and Arbitrum, as well as L2 networks based on zk-Rollup such as StarkNet and zkSync 2.0, and of course, sidechain solutions like Polygon and xDai which has already launched its mainnet. The current mainstream dApps seem to have completed the L2 team. What do you think of the competition facing StarkNet? And how will StarkNet attract more dApps and users’ adoption? What is the future plan of StarkNet?
Eli Ben-Sasson: I think there’s a lot of room for many different L2s because there’s a lot of need for scale, and one could imagine several different L2s using similar technologies and offering bridges between them.
So I’m looking forward to all these different teams deploying! I think StarkNet has the most scalable, safest technology among all L2s, and its technology also has instant blockchain finality time which means its capital efficiency is optimal (and certainly far better than that of optimistic rollups). Thus, in any configuration of L2s, StarkNet will hold a prominent place.
Unitimes: Speaking of different L2s, what do you see the future of L2 look like from the long run? For example, will there be only one monoply in the end? What’s your opinion?
Eli Ben-Sasson: I think we’ll see many different L2s and I don’t believe there will be one monopoly. Also, an L2 that’s decentralized isn’t really a monopoly, just like Bitcoin and Ethereum are not monopolies because they’re decentralized. We’ll also see bridges between different L2s, and in fact we’re constantly talking to various other L2 teams about building such bridges!
Unitimes: It has been more than 10 years since the birth of Bitcoin, we have witnessed the development of a large number of new technologies and applications in crypto. As for you, where will the next battlefield of blockchain be after L2? Any prediction?
Eli Ben-Sasson: I think we’ll start seeing adoption from NFTs and gaming applications, onto virtual reality and through it to social networks and interactions. Then, finally, conventional enterprises and financial services will arrive at the scene, but they’ll be last, not first, to adopt.
Additionally, we’ll see demands from the public to apply the transparency and integrity standards of blockchains and proof systems like ZK-STARKs to other areas, and require that conventional businesses apply similar standards even in permissioned environments. We’re already starting to see this in the area of CBDCs (central bank digital currencies).
Unitimes: In the past year, DeFi has attracted many traditional capitals. Some people think that DeFi is the “killer app” of Ethereum while some think that the threshold for using DeFi is too high — for example, its user experience is still not friendly, the logic behind it is still complicated, and the risks are high. What is your opinion on this? What do you think is the key for the large-scale adoption of blockchain in the future?
Eli Ben-Sasson: I think the killer app of Ethereum will actually come from the virtual worlds of gaming, entertainment and virtual reality. The reason is that in those worlds people anyways already assign a lot of value to things that are digital and scarce, so putting those things on a blockchain is a natural evolution.
So, while DeFi is really cool and nice, I think gaming, NFTs, social networks and virtual reality will be the killer apps. The reason these are not taking off yet is due to the limited scalability of Ethereum. But StarkNet will resolve this, so we’ll see this really taking off on StarkNet.
