Dialogue with Vitalik Buterin: A Brief Overview of the Differences and Trade-offs of Rollup and Other Ethereum Layer 2 Solutions

This article was published on imToken, original title: "AMA Dialogue with Ethereum Founder Vitalik", Interviewed by: Pan Zhixiong, Research Director of ChainNews, Organized by: ECN Esther

The discussion on Ethereum scalability is in full swing within the community, with multiple solutions being actively developed and expected to launch on the mainnet this year. On the eve of the explosion of Ethereum Layer 2 solutions, imToken, in collaboration with several outstanding Ethereum ecosystem communities and companies such as ETHPlanet, EthFans, ECN, Shanghai Frontier Technology Seminar, and HiBlock, has jointly planned a series of events themed around Ethereum scalability.

The first event was held on April 23: Rollup - Ethereum L2 Scalability New Paradigm Hangzhou Offline Meetup. Below is the text version of the AMA dialogue with Ethereum founder Vitalik from this Meetup.

Pan Zhixiong: Currently, EVM compatibility may become a very important competitive advantage for public chains or Layer 2. Many public chains of exchanges, such as BSC and HECO, leverage their own users or overall resources to provide a public chain that is relatively less decentralized compared to Ethereum, but with better performance, TPS, and lower user access costs. In the long run, do you think that the ecosystem strategies based on EVM compatibility developed by side chains or public chains like xDai, BSC, or HECO could potentially become a form of public chain ecosystem in the future? And could BSC or xDai possibly give back to the Ethereum ecosystem in the future, ultimately benefiting the entire Ethereum ecosystem?

Vitalik: Okay, understood. The current issue is that there is no blockchain that is scalable, has particularly high decentralization, and is EVM compatible. Right now, we have the power of Ethereum, as well as other side chain projects like Matic and xDai. There are also many projects that have nothing to do with Ethereum, like BSC. Currently, Ethereum's transaction fees are still particularly high. Some applications are not affected by the high transaction fees of Ethereum. However, many applications do indeed require lower transaction fees. There are currently no particularly good decentralized methods. So many people are starting to look at side chains and other chain projects.

But this is just the current situation. Next year and the year after, Rollup projects will probably all be launched. Additionally, we will have Ethereum sharding, so next year we will have some chains that support EVM, but we also have Layer 2 chains based on Ethereum. The two biggest advantages of Layer 2 chains based on Ethereum are security and connection to the Ethereum ecosystem. If you are creating your own chain, you might have 20 or 100 nodes, but it's hard to have 5,000 or 10,000 nodes. If you do a Rollup or create a platform that has a direct connection to Ethereum, you can benefit from the security standards of the Ethereum network. The second advantage is that if you have a close connection to the Ethereum ecosystem, you can gain some benefits from the network effects of Ethereum. However, when you build an application on these chains, you will have your own ecosystem. There will be issues with the connection between these ecosystems and Ethereum's ecosystem.

Pan Zhixiong: The second question is related to developers. We see that more and more Layer 2 projects may be launching in the coming months. However, when we talk to many developers, including those in China, about DeFi projects, they are still confused about which Layer 2 to choose. With so many options available, including State Channels, Plasma, Rollup, and side chains, what advice do you have for DeFi developers or ordinary developers looking to enter the Ethereum ecosystem in this multi-Rollup, multi-Layer 2 scenario? From what angles should they consider which ones to choose as their new direction?

Vitalik: The first question is whether you choose a project that supports EVM or one that does not. There are currently some Layer 2 projects that do not support EVM because they are simpler, so many of them have launched, including Loopring, zkSync, and OMG's Plasma. If you are creating a very simple application, such as something related to tokens or NFTs, you actually do not need EVM support. Right now, you can choose a Layer 2 that does not support EVM but is connected to the mainnet.

However, if you are working on something more complex or already have an application based on Ethereum, you will likely need a Rollup that supports EVM. You can choose another programming application or environment, but right now EVM has a large ecosystem with many contracts and code, so using an EVM-compatible Rollup will make your work easier. If you choose a Rollup that supports EVM, you have many options, including Optimism, Arbitrum, and zkSync, which support EVM, as well as some other projects.

I think Optimistic Rollup will be safer because zero-knowledge proofs are still a relatively new technology. It is complex, and not many people particularly understand how to read zkSNARK circuit code. However, if you look at the code for Optimism and Arbitrum, both are relatively simple. So, I think in the short term, the likelihood of issues with Optimistic Rollup is lower, but in the long term, ZK EVM Rollup is better. However, long-term could mean 3 years, 5 years, or 8 years, and current Optimistic Rollup projects may also evolve into ZK Rollup. If you want to use a Rollup that supports EVM right now, I think Optimistic Rollup is safer, and I also think it is more likely that they will actually launch without delays into next year or beyond.

Between Optimism and Arbitrum, I personally have great respect for both teams, and I know there are now more Optimistic Rollup teams. I think everyone needs to continue to watch their progress and see how these projects, their communities, and developments are doing.

Pan Zhixiong: Okay, thank you. The next topic I want to discuss is about zkPorter, which we mentioned earlier. They will be launching on the mainnet in the next two to three months. The zkPorter solution essentially moves data availability off-chain. This is somewhat similar to Ethereum 2.0's sharding, which also moves data availability off-chain. I want to ask if there is any comparability between the two, and how do their solutions compare to Ethereum 2.0's sharding plan? What are the points of comparison?

The second question is if zkSync can deliver on time and launch in August, what advantages would developers have in choosing Optimism? Because in the long run, zkSync's security will definitely be higher, right?

Vitalik: Actually, I am a bit concerned about zkPorter's method of data availability because whether in Ethereum's sharding or data availability supported by the entire network, if you cannot successfully attack the Ethereum chain, you cannot attack the data of a shard. However, zkPorter's method of data availability is not supported by the Ethereum chain, but rather by some of their nodes, so attacking their data availability would be easier. In fact, I think if you compare their zkRollup and zkPorter, zkPorter might be cheaper, but transactions on a Rollup are already quite cheap. Currently, sending a transaction on the mainnet costs about 20 RMB, while sending a transaction on a Rollup this year will be 0.2, and next year or during the 2.0 sharding phase it will be 0.002 or 0.0002, so transaction fees on Rollups will be very low. Therefore, I think a platform that has off-chain data availability is unnecessary.

Secondly, if they successfully launch in August, what advantages and disadvantages will they have? A significant advantage of zkRollup is that there is no waiting period for withdrawals, unlike Optimistic Rollup. In Optimistic Rollup, if you withdraw now, you have to wait 7 days. In fact, Optimism, Arbitrum, and some other teams are using a Layer 2 method to solve this issue, which is that if you withdraw now, a Liquidity Provider will provide you with tokens, and this Liquidity Provider will wait for 7 days.

But zkRollup does not have to worry about these issues, so the user experience may be better. However, if Optimism and Arbitrum successfully implement a Liquidity Provider mechanism, Optimistic Rollup can avoid the waiting period issue. If they launch in August, I think their biggest disadvantage is that the risk of security vulnerabilities will be higher. This issue is not the fault of their teams, but rather that zkSNARK technology is very new and complex, and there are fewer developers who understand zkSNARK code, so there is a greater likelihood of issues that they have not discovered. But this is a current disadvantage; in 3 years or 5 years, they will have more time to confirm that their zkSNARK EVM is problem-free, and the ecosystem will be more mature. So this is a short-term disadvantage, while the long-term advantages of zkSNARK are significant.

Pan Zhixiong: Thank you, Vitalik. The last question is about the broader applications of Layer 2. We see that Ethereum is currently more focused on NFT, DeFi, or payment-related application scenarios. However, Ethereum's original intention was not just to engage in purely financial-related businesses. In fact, it has now become a financial settlement layer, right? Recently, we have seen data that exceeds PayPal in terms of settlement amounts. So, is it possible for us to see more general application scenarios unrelated to DeFi emerging or exploding on Layer 2? Have you seen any interesting projects that could potentially play a role in the low-cost and high TPS scenarios of Layer 2?

Vitalik: Hmm, understood. I think there are currently no major challenges for Rollup. The reason Ethereum has become a finance-focused project is that transaction fees are currently very high. Financial projects can afford these fees. However, if you are working on a non-financial project, you cannot afford such transaction fees right now. Rollup's transaction fees are lower, so I think there is a great possibility that we will see these applications on Rollup. In fact, we can already see these situations. If you look at some successful non-financial projects in the Ethereum ecosystem, like Dark Forest, some are still on the Ethereum mainnet, some have moved to our testnets (like Rinkeby, Ropsten, etc.), and some have moved to side chains (like xDai or Polygon). So when we have Rollup, there will be more non-financial applications that will move to Rollup.

A particularly important example is ENS. I think ENS is currently the most successful non-financial application in the Ethereum ecosystem. Many people now have ENS domain names, but creating or updating an ENS domain name has become particularly expensive, making it difficult to use. However, if ENS or part of ENS could move to Rollup, it would solve this problem.

Audience Question 1: Now we see so many Rollups, so what is currently the most critical issue that needs to be solved in the Layer 2 technical solutions? Are they facing any significant problems or challenges?

Vitalik: Hmm, understood. I think these Rollups do not have major challenges right now, but there are quite a few smaller challenges. There may be 100 small issues, and if they cannot solve these 100 small issues, then the performance and user experience of the Rollup will be particularly poor. However, if they can solve, say, 50 or 80 out of the 100 issues, then the user experience will be particularly good. For example, one issue I previously raised is the withdrawal problem of Optimistic Rollup. Withdrawing from Optimistic Rollup requires a week of waiting. So they want to create a Liquidity Provider mechanism to help users withdraw faster.

A second example is that if these Rollups start to be particularly successful, the TPS inside may be very high. For example, a Rollup might have 100 TPS or 300 TPS, and the nodes within these Rollups will have synchronization issues. Because there are still gas nodes that need to process transactions within these Rollups. So right now, people are not particularly concerned about this issue because if a Rollup has 100 TPS, if they have this problem, it means their Rollup is already very successful. However, when they are that successful, their nodes will need to be more efficient. This is the second issue.

The third issue is that if we envision a future of scalability with not just one Rollup but multiple Rollups, such as Optimism, Arbitrum, Loopring, etc., then cross-rollup transaction issues will become particularly important. For example, if I have some assets in a certain Rollup, how can I move my assets to another Rollup? In fact, I have thought about this issue; one day I may need to pay a friend some ETH, maybe 0.05 or 0.03, but the current transaction fees are quite high. My friend has a Loopring account, and I also have a Loopring account, but I don't have enough for 0.03, while my zkSync account has 0.1. But at this moment, I have no way to move my 0.1 from zkSync to Loopring. Of course, I can withdraw from zkSync and then deposit on Loopring. But this takes a long time and incurs a lot of gas fees. However, if we had a cross-rollup exchange, it could solve this problem. So, I think we currently have many of these small issues, and if we can solve most of these small problems, I believe the Rollup ecosystem will be very good. Some of these small problems are not research issues; we know what needs to be done, but it requires developers to write code and conduct tests.

Audience Question 2: Next is a technical question. Does Ethereum currently have plans to support more zero-knowledge proof algorithms in pre-compiles? For example, recursive zero-knowledge proof algorithms like Pickles?

Vitalik: Are you asking whether Ethereum has plans to add more pre-compiles to help developers and researchers with zero-knowledge proofs? Or to create zero-knowledge proof compilers? Because there are some zero-knowledge proof systems where if you write this algorithm, it is already researchable on the current Ethereum chain. If it is an algorithm based on elliptic curves pairing, it can be done on Ethereum.

Audience Question 3: The next question is technical. What do you think about polynomial commitments in Rollup? Can we replace Merkle trees with polynomial commitments?

Vitalik: This question is a bit complex because when considering our stateless clients and state expiry, we also looked into this issue. We found that using polynomial commitments has a significant problem: if there is a particularly large state, meaning a state with many accounts, for example, if there are 5,000 accounts in a block, if some accounts' balances or storage change, we need to compute the witness for all accounts. We found this problem particularly difficult to solve, so in Ethereum's roadmap, we did not choose to move to polynomial commitments. We chose to move to Verkle Trees, which is a method between Merkle Trees and polynomial commitments. It has some characteristics of polynomial trees but eliminates their disadvantages.

Pan Zhixiong: Okay, thank you. This concludes the event. Thank you, Vitalik.

Vitalik: Thank you all, goodbye.