Original title: "Is the Ethereum Roadmap Off Track?"
Guest: Max Resnick, Ethereum researcher working in the Special Mechanisms Group
Host: Ryan Sean Adams, David Hoffman, co-founders of Bankless
Original translation: BlockBeats
Editor's note:In the development of Ethereum, L2 solutions have gradually become the focus. However, with the improvement of scalability, these solutions have raised concerns about decentralization and trustlessness. Developers face a choice: to maintain the core concept of Ethereum or to pursue a faster and more efficient network? In this episode of Bankless, hosts Ryan and David and Ethereum core developer Max discuss whether the Ethereum roadmap has deviated from the direction, especially whether the second-layer solution has affected the development of Ethereum.
This episode of the Bankless podcast invited Max Resnick, a researcher at SMG (@specialmech). He has some different views on Ethereum's roadmap than the mainstream view. Max believes that the Ethereum community is overly focused on Rollups and has neglected the improvement of L1, making it difficult for L1 to support the application needs of key areas such as DeFi. He called for re-prioritizing the decentralization and performance optimization of Ethereum L1 instead of relying solely on L2, and believed that the Ethereum roadmap should be redesigned, and the most important thing to consider is to shorten the block time. During the discussion, Max also expressed his optimism about ZK technology and questioned the possibilities that the current L2 ecosystem will develop.
In this podcast, Bankless and Max raised and discussed several core questions:
1. Is Ethereum's roadmap off track?
2. Are L2 solutions still an extension of Ethereum? Are they hindering the flow of users, transactions, and value to the Ethereum mainnet?
3. Are the current Optimistic Rollups controlled by centralized sorters and Ethereum's goals misaligned? What will be the consequences if left unchecked?
4. If these issues are not just FUD, how should we respond?
Ethereum roadmap off track: Max Resnick believes that Ethereum’s roadmap has deviated from its original decentralization and censorship resistance goals, focusing too much on L2 solutions and ignoring the development of Layer 1.
Challenges caused by L2:While second-layer solutions can reduce transaction costs, Max is worried that the control of centralized sorters is undermining Ethereum’s decentralized vision.
Prioritization of Layer 1: Max advocates that priority should be given to improving the capacity and speed of Layer 1 to ensure that it can support core applications such as DeFi without relying entirely on L2.
ZK Rollups vs. Optimistic Rollups:The current roadmap favors Optimistic Rollups, while ZK Rollups, a technically superior solution, has not received enough attention.
Importance of decentralized sorters:Max believes that it is crucial to promote decentralized sorters for L2. Many projects claim to be decentralized but have not really taken action.
Debate on the relationship with L2:Max questions whether L2 is truly an extension of Ethereum, and believes that the current L2 is more like an independent chain and fails to fully utilize the security of Ethereum L1.
Worst-case concerns:Max is concerned that if centralized L2 continues to be allowed to develop, Ethereum may become too dependent on these centralized sorters, resulting in the loss of its decentralized nature.
Dispute over the technical development path:There is controversy within the Ethereum community over the timetable for the gradual decentralization of L2, and some people believe that these technologies should be given more time to be gradually improved.
The following is the original text of the conversation (the original content has been deleted and reorganized for easier reading and understanding):
Summary of golden sentences in this paragraph:
1. "I think a weakness in Ethereum research is the lack of communication between different research areas of our (Ethereum Foundation, EF)."
2. "We (Ethereum) need to increase capacity and speed, and we must compete with other L2s."
3. "Since The Merge, we (EF) have been looking for direction, but we found the wrong thing."
4. "Now the fees for Optimistic Rollups are actually cheaper than ZK Rollups, which is strange because ZK is supposedly a superior technology. I think we (EF) have been influenced by someone..."
David: Let's take a moment to learn about your field in the Ethereum ecosystem. What are you busy with now and what are you mainly focused on?
Max:I mainly study MEV and spend a lot of time in the "Dark Forest" studying areas such as order flow and supply chain. But recently I’ve started branching out to other parts of the ecosystem because I feel like if we’re going to solve some of the problems related to MEV, these problems require cooperation from other areas of the roadmap.
I think one of the weaknesses in Ethereum research is the lack of communication between our different research areas. For example, the consensus team is working on Orbit and single-slot finality, while the MEV team is saying, “If you don’t want to lose billions of dollars every year from users, you have to do something.” But there’s a fundamental conflict between the two. So, I started trying to collaborate across disciplines and communicate with other research and engineering teams. We’re solving very important problems on the MEV roadmap, but we really need help from other areas to solve these problems together.
Ryan: Max, let’s start with your “anti-mainstream” views on Ethereum. Can you briefly summarize your “anti-mainstream” views?
Max: I feel like we’ve lost our way on the roadmap. Ethereum L1 is good enough, but it's obviously not good enough. In the past few hard forks, we have been optimizing around Optimistic Rollups. But I think we should re-prioritize L1, make sure the block time is shorter, and make sure L1 is usable for the most popular application - DeFi. Although some people now even think that we shouldn't do DeFi at all, I think this needs to be discussed.
David: You're not against the Ethereum roadmap being centered around Rollups, right? You just think the priorities and order of execution should be adjusted, is that accurate?
Max: Yes, I think there's a case for using Rollups to compress transactions that are not high value, like buying coffee or making small transfers, which should all be done on Rollups, but the core Ethereum L1 should be where people transact. We shouldn't force applications like Uniswap to create their own application chains because L1 is too difficult to use. Similarly, we shouldn't force people to transact on L2 without any security guarantees, but L1 should be good enough to at least support the needs of these applications.
David: Do you think this is consistent with the idea of "non-monolithic" or "integrated"? Regardless of which word you use, you're still emphasizing solving the problem by significantly improving the capabilities of L1, right? Max: Yes, I think we need to increase capacity, increase speed, and compete with other L2s. Even if we can't be better than them, we will have more security, better mechanism design, and ultimately the combination of these advantages will help us win. For example, especially compared to Solana, I think we have a lot of advantages in other aspects, but if we refuse to compete on speed and throughput, we will lose anyway because these are the two most core functions of blockchain, and users don't want to wait 12 seconds to get on the chain. Ryan: Max, your point is that the Ethereum roadmap has gone off track. I'm curious, you've been in this space for a long time, when do you think it started to go off track? Why? How did it happen? Max: I feel like since The Merge, we've been looking for a direction, but we've found the wrong thing. The merge was a huge relief for the core developers, and we accomplished an amazing and significant thing. But then there was a sense of fatigue, the adrenaline went down, we took a step back a little bit, and then we came back to the table and ultimately decided that we were prioritizing the wrong things. It's not that these changes were bad, or that we shouldn't be making them. The problem is that we as an ecosystem have limited resources and can only do so much. I feel like we weren't prioritizing the most important things.
Ryan: Since the merger, from my perspective, Ethereum's priority has been to make L2 transactions very cheap. The launch of blobs has indeed achieved that goal. Are you saying that these achievements are good, but not as important as other things?
Related reading: "Detailed explanation of EIP-4844 proposal: Introducing "transactions with blobs" to reduce Rollup fees"
Due to the launch of blobs, the number of active users and transaction volume of Ethereum L2 have doubled in the past 6 months, but Ethereum L1's profitability has shrunk by 99%
Max: Yes, I think we have achieved this goal. But at the same time, because we delayed progress on other things, the Builder market was completely centralized, and the relay market looked extremely centralized. We had a lot of areas that were shelved because they didn't get enough attention, so we only achieved one goal, which was to make the data space very cheap.
Related reading: "Another major MEV relay service shut down, more than 90% of Ethereum's future block settlement will be in the hands of 4 companies"
By the way, I think this helps Optimistic Rollups more than ZK Rollups. So I would call this part of the roadmap that is currently completed the "Optimistic L2 Roadmap" instead of the "ZK L2 Roadmap." Right now, the fees for Optimistic Rollups are actually cheaper than ZK Rollups, which is strange because ZK is supposedly a superior technology.
I think we were influenced by some people. For example, a core development team member worked on the development of EIP-4844, but at the same time he was also working on another Optimistic L2 project, while the people from the ZK L2 project were not involved in promoting this part of the roadmap at all. Currently, it is almost all the people from Optimistic L2 who are promoting the roadmap, and the direction is beneficial to them.
Ryan: What do you think the (EF) decision process was like? My understanding of the decisions was that Optimistic Rollups were the first technology to emerge, ZK Rollups followed closely behind, and then probably in 2021, there was some confusion about the roadmap. Max: I think that's partially true. At the time, we saw the Optimistic path and it felt like the ZK path was far away, but it was actually much closer than we thought. So I want to give credit to teams like Succinct, who have made breakthroughs in compiling Rust code into ZK circuits, which has accelerated the development of ZK technology. ZK technology is a superior technology because it compresses the state and takes up less bandwidth. Bandwidth is basically the main limitation in the future world, and ZK is the only technology that can really make breakthroughs in bandwidth limitations, which Optimistic Rollups can't do.We thought ZK Rollups were going to be a decade away, but it’s only going to be about two years. We made some roadmap decisions based on that, but now we should step back and reevaluate whether we made the right decisions, especially if we misjudged the progress of ZK technology.
What should we do next? I think we should start optimizing the features needed for ZK Rollups. If we are going to continue to go the Rollups route, we need to immediately shift to making it cheaper to verify ZK proofs on-chain, rather than continuing to drive down the cost of data, because data on-chain is now almost free.
Also, I think fees on L1 have reached a level of equilibrium, but if you want to launch an application, what really matters is what will the fees become when your application explodes? If I am an application developer and I have developed a killer application that may grow 100 times and bring a lot of users, I will not choose to launch it on Ethereum. Because in the best case scenario, if I succeed, it will clog the chain and make fees so high that it is unusable. So what really matters is not the current equilibrium state, but how much scalability we have to attract top application developers who want to bring 100x user growth into the ecosystem.
Summary of golden sentences in this paragraph:
1. “They (referring to the Optimism superchain) loudly claim to be part of Ethereum, but they don’t enable fraud proofs... Do you want to be part of Ethereum, or do you just want to use Ethereum to create your own token and use it to pay gas fees?”
2. “I think the problem is that some Rollups may not want to be decentralized at all”
3. “If most transactions occur on a centralized server on Base, the outlook for the world will be very bleak.”
Ryan: If you are a developer with a killer application, which ecosystem would you choose to develop in now? Ethereum or Solana?
Max:I hope to build it on Ethereum.
Ryan: Some people might say, “You can build on L2, that’s also Ethereum.” What do you think? Max: Which L2? There are many options. I think the question is "Which L2?" There are so many options and you don't know which one will win. Maybe it looks like Base has attracted all the users, but at the same time we also know that Base is very unlikely to decentralize their sorter because they make a lot of money from sorter fees. You might say I want to choose the L2 that looks most likely to decentralize the sorter first, but that's not Arbitrum either, because Arbitrum is also trying to make money from the sorter, and they are not willing to give up this part of the income. There are probably three or four that may win now, and Linea is working hard to decentralize the sorter as soon as possible to provide this option. If you look at Solana, there aren't four different SVM L2s for you to choose from, you can just build on Solana and be confident that eventually L1 will solve the scalability problem, at least for your DeFi application.
David: There's been a lot of debate lately about whether L2 is an extension of Ethereum, Ethereum skeptics would say L2 is a new chain with independent execution fees, and these fees are captured by L2 and not shared with L1. Ethereum supporters would say that all L2s use Ethereum as security and are a natural extension of Ethereum, what do you think of this debate?
Max:Personally I don't think it's a black and white question, it's a continuum. But my position is that these are more like independent chains at the moment. If they want to continue the roadmap of decentralization, they need to make sure they push data to L1 frequently, make sure they have fraud proofs, make sure they have forced transactions, so that they will gradually move closer to becoming part of Ethereum, and if they start to have more composability features that allow for easier composability, then they will move further in that direction.
I think the problem I see is that the L2s that look like they're going to win have no incentive to push these changes, they loudly claim to be part of Ethereum, but they don't have fraud proofs enabled, and they don't show any signs of moving closer to a decentralized sorter. They're trying to build interoperability within their own ecosystem through super chains. So, do you want to be part of Ethereum, or do you just want to leverage Ethereum's users and influence, build your own token, and use it to pay gas fees? Ryan: Maybe you can look ahead to the next decade, Coinbase continues to invest in Base, Optimism's super chain becomes highly centralized and integrated, what are the long-term risks you worry about if we don't change direction? Max: Well, I'll give you a worst-case "doomsday scenario". I'm not saying those people are evil, many of them are doing valuable work, but the incentive mechanism exists. Coinbase has a corporate governance structure, which is different from independent founders like Jesse. I work at a large company ConsenSys, and we also have a corporate governance structure. The head of Linea and the head of our governance structure are different people. But we have some other interests in the ecosystem that give us an incentive to prioritize the decentralized sorter. When I look at Coinbase, I think their interests are different. We have MetaMask, we want users to use Ethereum, we want transactions to flow in the ecosystem, and we want decentralization. And our CEO is one of the co-founders of Ethereum, so we have a deep interest in Ethereum. Coinbase, which gets a lot of revenue from users trading Ethereum on its centralized exchange, doesn't have the same connection to Ethereum.
So my concern is that we push a lot of users to L2s like Base, and their financial incentives make them reluctant to decentralize sorters. They continue to grow their user base, increase market power, increase revenue, and eventually all activity happens on a centralized sorter. To me, this is no longer cryptocurrency, and it's no longer Ethereum. This is not decentralized at all, it's just a centralized server processing transactions on the block. Maybe they will have fraud proofs, maybe not, but they don't have them at the moment. So my biggest concern is that market forces and network effects happen, and these projects become too big to be shaken. Now we still have time to change direction, but five years from now, maybe all the activity is concentrated there (referring to centralized L2s like Base), and we can't bring users back to a truly decentralized financial system.
David: What is your ideal path? What changes do we need to make to get there?
Max: The ideal path is that everyone is working towards the same goal - DeFi, decentralized ecosystem, censorship resistance, trusted neutrality, Ethereum as a currency. If these are the directions we are working towards together, we may be able to quickly accomplish what we need to do, such as enabling fraud proofs and forced transactions. At the same time, we can also make some optimizations for ZK Rollups, which I think are a bit ignored at the moment because we don't prioritize them.
Compared to OP Rollup, the proof cost of ZK Rollup has remained high, data source: Dune
Maybe we can get to an ideal state where decentralized sorters are implemented on these Optimistic and ZK Rollups. I think this is the best case scenario for the L2 roadmap: these projects do what they have been saying they are going to do for five years, and finally we have a decentralized ecosystem. But the question is, how many times have they said they were going to decentralize the sorter, but never did? They keep delaying it again and again, when will we realize that we are being "ripped off"?
Ryan: Are you trying to start a civil war within Ethereum?
Max:Not really, I just want to say that if you claim to be supporting Ethereum and say you have been working on a decentralized sorter, just do it. Don't always say "next month" and then delay it again and again and never deliver.
My view on this is that, for example, Arbitrum, what they are doing with "Time Boost" may not be compatible with a decentralized sorter. So when they're moving toward a system that looks incompatible with a decentralized sorter, and then on the other hand they're saying they're going to implement a decentralized sorter, that doesn't make sense to me. So if you're paying close attention to what they're doing now, that's not quite right. I'm not trying to start a civil war, I'm just trying to put pressure on those who claim to be supporting Ethereum to actually deliver on the promises they've been saying for the last five years.
Ryan:Max, I think it's good to apply pressure, and Ethereum and its community do have some mechanisms in place for social pressure. I want to discuss ways that might be softer than the "hard landing" you describe. The basic scenario you mention is that there's a lack of coordination between the various L2s, there are uncooperative incentives, which leads them to ignore Ethereum's role as a federal government when making decisions that are in their own interests, and I think we need to be aware of that possibility.
A typical Ethereum roadmap or L2 roadmap supporter might say, "Max, look at L2, these L2s all start out at "Phase Zero", which is their starting stage, and then gradually progress, add mechanisms like fraud proofs, and then they decentralize the sorter and gradually achieve full decentralization.
These technologies need to mature gradually, and they can't be fully decentralized until we ensure that they are safe for users. The worst-case scenario you describe will not happen. What do you think of this view?
Max: I think the problem is that some ecosystems, some Rollups may not want to be decentralized at all. Do these organizations, especially those that are profitable and in some cases publicly traded companies, really want to decentralize their sorters? From their own self-interest, it is actually more beneficial for them to retain control of the sorter. This is my point: look at the money, look at the incentive mechanism, these incentives are not inclined to decentralize the sorter. So as our ecosystem becomes more and more dependent on these players, our ability to exert pressure becomes less and less.
I guess I'm not the only one who's unhappy with the slow progress of decentralization. I remember Vitalik also mentioned a similar point on Twitter that the progress of decentralization and roadmap is: we reach stage one, and then stop there and do nothing. As a MEV researcher, the most important thing to me is whether we can achieve a decentralized sorter. If most transactions occur on a centralized server in Base, the prospects for the world will be very bleak.
Ryan: So what kind of behavior does L2 show to prove that they are not acting independently, but are willing to truly move towards decentralization?
Max: First of all, I don't think that L1's mandatory inclusion of transactions is enough to provide sufficient censorship resistance for L2. I think the sorter itself also needs to have a certain degree of censorship resistance and trusted neutrality.
Related reading: "Sequencer and Proposer fail, how does L2 implement self-sorting and emergency exit mechanisms? "
"The backup strategy allows for forced inclusion in the event of a sorter failure by forcing the transaction to be included after a specified time interval to ensure forced inclusion."
Because if these applications exist, if global markets are going to be on-chain, and they're running on AWS servers and can be censored by the US government at any time, that doesn't count as crypto to me, we might as well just have global markets running on the CME or the NYSE. If you're sanctioned by the US government or Coinbase, forced inclusion of transactions is not going to help you trade because Coinbase doesn't like you. So my view is that they need to implement decentralized collators that are censorship-resistant and credibly neutral.
If you want to be part of Ethereum, you can't claim to be part of Ethereum and have a centralized collator on AWS process all transactions and charge users a lot of fees that don't reflect the costs of L1.
Ryan: Do you think L2 can achieve the decentralized collator you mentioned in six months?
Max:Yeah, you just need to reach consensus on the blocks. We know how to do that, that's how blockchains work.
You can have permissionless entities building blocks today, and that's how Ethereum works today - we have a group of validators who have the power to propose blocks. This is a feature that all blockchain technologies already have, they just don't want to do it.
Summary of golden sentences in this paragraph:
1. "The final outcome of Ethereum that I hope for is not a cheap, undifferentiated DA chain, and then a group of powerful companies build Rollups on it."
2. "I think L2s are actually direct competitors to Ethereum now. How can you call something that directly competes with Ethereum part of Ethereum? This makes no sense to me. They are competing for the same users on the same set of applications."
3. "Eventually, all users will leave L1. If we can get L1 and L2 at the same level on the decentralized sorter, and everyone faces the same limitations, I think it will be better."
4. "The fees charged by Base are significantly higher than the amount it should charge based on L1 costs."
David: Are you looking for some kind of "fractal structure" where L2 splits off from L1? My original vision for L2 in 2019 was that you could stake your OP tokens, stake your ARB tokens, and then you also had a permissionless validator set, and you could participate in the validation of your preferred network. But that's not what's happening now.
Max: Yeah, the endgame I'm hoping for for Ethereum is not a cheap, undifferentiated DA chain, and then a bunch of powerful companies building Rollups on top of it and extracting as much revenue and power as possible from their monopoly position. They're not sure whether to decentralize the sorter, one reason is wanting to provide guaranteed execution quality to users, and another reason is revenue.
David: Assuming we do get to that structure, where we have a distributed validator set and sorter set on L2. Does that change your view on L2 not being an extension of Ethereum?
Max: Yeah, I would say they'll be closer to being part of Ethereum. I'm not sure if they'll be completely part of Ethereum, that depends on a lot of factors.
David: What are your considerations?
Max:For me, I'm an economist and I think about incentives a lot, and the key is how aligned the incentives are. Is Base's incentive now to try to make Ethereum better? Or is Ethereum actually a competitor to Base? If Ethereum gets better, does that mean Base will lose users? That's what I call "incentive misalignment."
If we're in a situation where the overall improvement of the ecosystem is good for all Rollups, then L2s will succeed together, not compete with L1s. But I think L2s are actually direct competitors to Ethereum now. How can you call something that's in direct competition with Ethereum part of Ethereum? It makes no sense to me, they're competing for the same users on the same set of applications.
We need to find a solution. There's no reason to have to maintain this competitive relationship. We can build an L1 that's very suitable for DeFi, and then move other non-sensitive types of transactions, such as basic transfers, to L2. This means that L2 can relieve a lot of burdens from L1, allowing L1 to better complete its tasks. This would be a symbiotic relationship, but I think the current situation is more like a parasitic relationship, because L2 is trying to take DeFi away from L1 and put it on its own platform, and L2 is still a centralized sorter.
David: Even if it develops according to the ideal state you described, L2 still has its independent characteristics. They are still independent blockchains and do not have much value capture back to Ethereum L1. So I am a little confused, why can these changes make L2 an extension of Ethereum?
Max:This is a question about power dynamics. If L1 is a decentralized system, and we have to move slowly because there are many stakeholders, and the services we can provide to users are limited by decentralization, then it is fundamentally unfair to have DeFi competition on L1 and L2. Because users tend to make short-term choices and don't consider long-tail risks, just like people don't often buy enough insurance. Users don't think about decentralization, even if it would be better for them to think about these issues.
Eventually, all users will leave L1. If we can get L1 and L2 to the same level on decentralized sorters, and everyone faces the same limitations, I think that will be better. The more Ethereum properties L2 inherits, the closer it is to Ethereum. They already inherit security and finality, and if they can also inherit trusted neutrality, that would be even better. But I think there is still a fundamental tension between whether user activity will happen on L1 or L2, because they are directly competing for Ethereum users, and users are the lifeblood of Ethereum.
David:So your concern is that many L2s are somehow "cheating" and "interfering" with Ethereum in terms of value capture, which is why you think they are not enough of an extension of Ethereum.
Max:Yeah, I think there will still be a substantial difference even so. Right now L2s are competing with each other, and they are also competing with L1 for users. We need to get to a point where the types of activities happening on L1 and L2 are very different, otherwise this tension will always exist.
Now that the bull market has come, things are relatively calm, but eventually, economic forces cannot be ignored and incentives will emerge. I think this incentive has already emerged, such as Arbitrum's "time acceleration" proposal, or Base charging fees that are significantly higher than it should based on L1 costs. Optimism's superchain is quite unique. Optimism is more committed to achieving a non-exploitative and decentralized sorter in the short term, so I don't want to put them in the same category. But for Base and Arbitrum, the incentive mechanism is very strong, and they have no motivation to be decentralized sorters. To some extent, they are in direct competition with Ethereum.
Base's DA operating costs are far from the actual network fee income, data source: Dune
Summary of quotes in this paragraph:
1. "I hope to achieve a transition to Based Rollup, although I think it's unlikely to happen. I believe in both points at the same time."
2. "It's ridiculous that the blockchain still maintains a 12-second block time in 2024. We are too far behind other projects."
3. "A negative research attitude is very common in the Ethereum ecosystem, as if to say: Why are you researching this problem? It's too difficult."
David: I want to talk about Based Rollup, one part of which is that it is basically tied to Ethereum, outsourcing a lot of power to Ethereum L1. They have a formal alignment with Ethereum L1 in terms of block space and economics. Do you think Based Rollup is a viable solution?
Related reading: "Detailed Explanation of Based Rollup: A Solution that Fully Inherits Ethereum's Activity"
Based Rollup, also known as L1-sequenced Rollup, is a Rollup whose sequencing is completely driven by the underlying L1.
Max:I have two points of view. First, from an incentive perspective, they are indeed more aligned. Whether a user moves from Based Rollup to L1 or from L1 to Based Rollup, there will be no problem for both parties. This is a fundamental incentive alignment provided by Based Rollup, which is a good thing.
However, one of the disadvantages of Based Rollup is that the block time of L1 is too long to match the functions of other Rollups. For example, some Rollups have a block time of less than one second, but if you are a Based Rollup, you can only generate a block every 12 seconds, because this is the current block time of Ethereum.
Comparison of block times of various L1 public chains, the difference between Ethereum L1 and various L2 and EVM public chains is more than 10 seconds, data source: Coin98
Another question is incentives. I know I keep repeating this word. What motivation do Optimism, Base, Arbitrum have to become Based Rollup? They get a lot of income from the sorter. Giving up control of the sorter is like giving up "spice" (a rare energy source in the science fiction novel "Dune"). They won't be willing to do so. If we choose Based Rollup from the beginning, this approach will work, but now we need very strong social pressure, otherwise it will not move in this direction at present.
Ryan: If all L2s become Based Rollup overnight, will it ease your concerns?
Max:Of course. And I think we have not fully explored the potential of Based Rollup. But if we have some specific property advantages, such as real-time anti-censorship capabilities on L1, we can take on the role of maximizing an arbitrary objective function to optimize block sorting. Now, if you try to do a Based Rollup on L1, it basically has to maximize the profits of the validators, and maximizing the profits of the validators is usually in direct conflict with maximizing the interests of users, which is the basic contradiction of MEV in a single-leader chain. But if you have a censorship-resistant L1 that can adjudicate in any block auction, it can build "super Based Rollups" that no longer maximize MEV, but maximize user welfare.
I think this is my ideal end of the Rollups roadmap: every Rollup is a Based Rollup, but it is a "super Based Rollup". I think this is what Ethereum should pursue in the long run: maximizing user welfare. When everyone agrees on this point, I will be more willing to call them part of Ethereum.
I hope that the transition to Based Rollup can be achieved, although I think it is unlikely to happen. I believe in both points at the same time.
David: You tweeted a while ago that "The goal is not for Ethereum to generate the most profit for ETH holders, but for Ethereum to be a force for positive impact in the world, provided it is sustainable. I don't think the current L1 roadmap is the best way to achieve this." What do you want to see on Ethereum L1? How would you adjust its direction to achieve the goals you set out?
Max: There are three things, some symbiotic, some in tension. First, we must start reducing block times immediately. It is ridiculous for a blockchain to maintain a 12 second block time in 2024, and we are too far behind other projects. If we continue to maintain a 12 second block time, it is obvious that we will be beaten by our competitors.
Second, we need to provide better real-time censorship resistance. This does not just mean "users wait 30 minutes to meet an independent validator who does not run MEV and is willing to package your transaction into a block." It means that users can pay a high enough fee to enter the next block, and this fee should not be so high that basic DeFi activities become impossible.
Finally, we need to increase throughput to support developers who create "killer applications." We need to be able to support 100 times or even 1,000 times the current usage, so that when people have great application ideas, they will choose to develop on Ethereum instead of going elsewhere.
One of the attitudes I hate most about Ethereum research is: "You want to scale L1, but we don't have the tools today to make it handle as many transactions as Visa." This attitude is too negative, as if to say: "You want to do this, but the technology is not there yet."
I read an article about flying that was published six weeks before the Wright brothers' first test flight. The article said that the technology for human flight does not exist yet, and maybe humans can achieve flight in 10 million years, but everyone working on it now is wasting their time. I think this negative research attitude is very common in the Ethereum ecosystem, as if to say: "Why are you working on this problem? It's too hard."I believe that if we have a group of smart people who are willing to work hard, we may be able to achieve this goal instead of giving up at the beginning.
Ryan: If we shorten the block time, suddenly Based Rollup becomes more feasible. Maybe it can prompt new Based Rollup teams and frameworks to emerge and compete more competitively with existing opponents, right?
Max:Absolutely right. And as a user, which chain do you want to be on? On a chain that maximizes your personal value or on a chain that maximizes your personal welfare? Of course I would choose the latter, but if the block time of this chain is still 12 seconds and the fees are very high, it will affect my choice. But if we can keep fees low and block times fast enough while maximizing user welfare, I think that will be a winning solution.
Summary of golden sentences in this paragraph:
1. "The bandwidth usage of Optimistic Rollups is almost the same as that on L1. The only thing saved is the execution part, and execution is not the bottleneck."
2. "We do need to borrow some tools from new blockchains, one of which is the concept of parallel execution and state storage."
3. "I think Vitalik's point is reasonable, but I also think that giving up on solving problems that have not yet been solved is defeatist. We should at least try, otherwise Ethereum will gradually decline."
Ryan: When I discussed this topic with others in the Ethereum community, I learned that the problem is not the hardware requirements for running Ethereum nodes, but the bottleneck that we cannot achieve a 2-second block time. This may prevent many users from continuing to participate in the network, which is why this proposal is often met with resistance, because it will actually reduce the decentralization of Ethereum L1, which is often considered "blasphemy". We don't want high bandwidth to be a reason that prevents people from participating in Ethereum L1 verification. What do you think about this?
Max:I think it is very reasonable, and bandwidth is indeed a bottleneck. This is also why I am more optimistic about ZK technology than Optimistic Rollup technology. ZK technology is essentially compressing, while Optimistic Rollups have to put transactions on-chain, which may do some compression, but not nearly as much compression as ZK Rollups can achieve. My view is that we need to use ZK technology more, not just to do a ZK Rollup, but to integrate ZK compression technology into L1, which will reduce bandwidth requirements.
Optimistic Rollups don't actually reduce bandwidth requirements significantly, at least without improving the data availability layer. I think we can also achieve some improvements with L1 blocks. We often go down a technology path and become too focused on that path and ignore the many new technologies we have invented that can be used in other technology stacks. Now the blob you download contains all the transactions, and from a bandwidth perspective, Optimistic Rollups have almost the same bandwidth usage as on L1, the only savings are in the execution part, and execution is not the bottleneck.
If our hardware requirements are still at the same level as they were eight years ago, execution may be an issue. But even if we use the past two or three generations of consumer-grade hardware as the minimum requirement for nodes, I think we can easily handle all current transactions. State growth may be a bottleneck, bandwidth may also be a bottleneck, but the execution optimization brought by Optimistic Rollups is not the core issue we really need to focus on.
Ryan: You're saying that while there are increased hardware and bandwidth requirements, this won't take away people's ability to stake from home. But can we really achieve a 100x increase in throughput without major changes? Another objection is that even if you achieve a 100x increase in Ethereum mainnet, is that really enough to support all the computation that needs to be done on Ethereum in the world?
Max: Yeah, I think we do need to borrow some tools from new blockchains, one of which is the concept of parallel execution and state storage.
One of the advantages of Rollup is that the state is highly isolated, so you don't need a big hard drive to store all the state. Once the amount of data reaches a certain level, you need to move from RAM to storage, and from storage to slower storage, and that's the problem with state growth. But if the state is isolated, you know which part of the state you need to access. You can store the part you're working on in memory, and when you're done you move it out of memory and load the other part you're working on, so I think parallel execution and state isolation are very important.
Another question is, is a 100x improvement enough? I discussed this with Vitalik the other day. He mentioned a point: "If we actually need 100%, but we're only at 1%, and even if we do all the work we can only get to 5%, isn't that a waste of time?"I think there's some truth to that point, but I also want to say that unless we try, we'll never know how far we can go.
I think we should optimize Ethereum to be a platform for financial activities, and financial activities are not just transactions like Visa, but include some controversial state changes, etc. So I think Vitalik's point is reasonable, but I also think it's defeatist to give up on solving problems that haven't been solved yet. We should at least try, otherwise Ethereum will gradually wither away.
Related reading: EF has no dreams
David: The current Ethereum roadmap has five main directions. This roadmap takes into account some of the things you mentioned, but it does give priority to issues such as L2 and data availability. How different is your proposal to modify the roadmap from the existing roadmap?
The five-phase roadmap for Ethereum after the merger, which are:
The Merge, refers to the merger of the Ethereum execution layer (current mainnet) and the beacon chain (new POS layer).
The Surge, refers to the introduction of sharding on Ethereum.
The Verge, the introduction of Verkle trees, will eventually improve the scalability of ETH.
The Purge, reducing the hard disk space required for validators. Eliminating historical data and bad debts.
The Splurge, a series of miscellaneous, "smaller" upgrades to ensure that the network runs smoothly after the first 4 phases.
Max: I think the biggest change is to shorten the block time, which is not currently on the roadmap. I think this is a very big shift in focus because there was a group of people who wanted to achieve single slot finality by extending the block time, which is completely different from what I proposed. So I think this is the biggest change.
I work in the MEV space and am most familiar with the Surge roadmap. I have a proposal that would achieve the goals of Scourge in a different way. If you're talking about the goals, the changes aren't huge, but if you're talking about the engineering approach, the differences are huge. I think it would achieve the same goals, but it would require rethinking these issues from fundamental principles, rather than designing a bunch of gadgets and piecing them together.
In my opinion, Scourge (the MEV part of the roadmap) is the one that's most focused on L1. I think we should shift that part toward reducing block times and addressing all MEV issues through multiple proposals.
Summary of golden sentences in this paragraph:
1. "I'm not a CIA agent who wants to dismantle this ecosystem. I just see the slow decline of our users and this ecosystem, and then want to ask: "Can we turn this around?""
Ryan: When I talked about the "civil war", I was partly joking, but also a little serious. These statements make some people in the Ethereum community feel uneasy because they are used to hearing similar statements from some L1 competitors who are trying to defeat Ethereum. These competitors often make some compromises on the current L2 roadmap, such as centralization and lack of anti-censorship. Many people in Ethereum do not trust these VCs and L1s who are trying to promote Ethereum alternatives. They know that these people are motivated by interests. So, Max, what is your motivation? Are you fighting for technology? Do you care about decentralization?
Max: I’m a researcher and when I see problems, I want to solve them. I see a problem where I think we’re optimizing for the wrong goals, and that’s why I’m out there and saying on Twitter, “Hey, I think you well-educated, well-intentioned people are wrong.” Because I see different paths and we’re not optimizing for the right things.
From a financial perspective, I’m not involved in any angel rounds. I only have one advisory position, again at Espresso System, which is not a big position, and it’s focused on the block auction part because I’m an auction theorist and working on decentralized sorter implementations, which is the important problem I’ve been discussing here.
Beyond that, my financial exposure is primarily through ConsenSys stock and my future career development. So I don’t own any Solana tokens, or any other L2 project tokens, and I’m not planning to launch an Alt L1. I work at ConsenSys and our business and user activity is directly tied to the success of Ethereum. So from a financial perspective, the most important thing is to make sure that Ethereum succeeds, that there are users, and that those users stay in the system. If users move to Solana, that's very bad for me because the wallet I work on has very low usage on Solana.
My goal is really just to make Ethereum as great as it can be and get as many users participating in L1 and L2 as they can, as long as they're using Ethereum and are happy with it. There's no conspiracy in my opposition, I'm not a CIA agent who wants to tear down the ecosystem, I just see the slow decline of our users and this ecosystem and want to ask, "Can we turn this around?" Because I really want the only decentralized smart contract chain to continue to be a positive force in the world.
Ryan: Max, in addition to Ethereum supporters, there may be other communities listening to this conversation, such as the Solana community. They hear you say L2 is parasitic, Layer 1 execution is good, and users should return to Ethereum, and they might think, "Max, why don't you just come to Solana? We're optimizing all the things you care about, and we're willing to re-architect the whole system to do it the 'right way'. Why don't you join us?"
Max:Ethereum is the first ecosystem I started working on, and I think we have a responsibility to get it on the right path. There are a lot of architectural designs in Solana that I don't completely agree with, and it may be too difficult to change now. For example, Solana may not necessarily reach consensus in some cases because its votes may be censored. These kinds of issues are beyond my ability, even if I join Solana and Anatoly says, "Hey, Max, can you come here and build this stuff?" I'm not a consensus mechanism researcher, and it's out of my control to re-architect the whole system in a sense.
By contrast, what Ethereum needs to do is more in line with my thinking, and I can also help with that. And as I said, I work for a company that builds products for Ethereum. I think Solana already has a lot of smart people with vision, how much leverage can I add to them if I join them? They already know the things I'm talking about. I'll just be another person over there saying these things. We need people in the Ethereum ecosystem to say these things because it's very important.
Also, some people may say that what I said is basically defending Solana. Actually, I'm defending the good parts of Solana. We are in an open source ecosystem, we take good ideas from everywhere and combine them to build the best things. Solana is not doing everything wrong, they have good ideas and bad ideas. We should take good ideas and incorporate them into our roadmap. The same is true for Monad, it can't be considered bad just because it is doing certain things, we should also take good ideas from Monad and implement them in our ecosystem.
I will do my best to make Ethereum as good as it can be. If the ETH research community is truly focused on doing the best things for Ethereum, I will continue to be here. Like I said, I firmly believe that all people working on Ethereum have no bad intentions and our goal is to make Ethereum as good as it can be. My task is to convince others that the things they think will make Ethereum better are not as effective as the things I advocate.
We need to re-order priorities, and my point is that when you say that things that are planned to be implemented after five hard forks should be completed in the first hard fork, and things planned in the first hard fork should be done after five hard forks, it is actually a re-organization of the priority queue.
Ryan: I think what you are advocating is to let the pendulum swing back slightly to L1. This is not inconsistent with Ethereum's overall roadmap. I remember a conversation with Justin Drake a few years ago, where he talked about Ethereum's sci-fi roadmap, and his vision was to have L2 innovate and build, and maybe even bring some technologies back to L1, such as integrating a ZK EVM. I would say that this is actually in harmony with Ethereum's long-term plan: focus on L1 first, then move to L2, and then return to L1. If we can achieve these goals, then we will have the best solution. Do you think my understanding is too harmonious?
Max:I think if we have a lot of places where people can try new technologies and there are incentives to develop these new technologies, then new technologies will be developed, and then we can decide which technologies are suitable for L1. The existence of L2 gives us the opportunity to test these ideas in real financial activities, rather than in testnets, because testnets cannot simulate MEV. So this is an opportunity to verify these new technologies and then introduce them to L1.
However, I think the problem is that we are competing for a key resource - bandwidth. Whether it is ZK Rollup or Optimistic Rollup, everyone is using bandwidth. ZK Rollup uses less bandwidth, but it is still using it. If we want to expand L1, we need bandwidth. In the current roadmap, although the plan to increase the number of blobs has been withdrawn, as more blobs are used, it will be more difficult to implement my proposal in the future.
Summary of golden sentences in this paragraph:
1. "If Ethereum L1 becomes the main place for people to trade, then ETH will become valuable, and other claims about ETH as a currency and as a store of value are too illusory."
2. "If most of the net assets of L2 developers are in L2 tokens and these tokens perform poorly, they will do things that are not aligned with Ethereum but aligned with L2 tokens. This is the direct point of conflict between L2 and L1-they have their own native token value."
Ryan: Many Ethereum holders will now ask why ETH is performing poorly, and what impact will your proposed roadmap adjustment have on ETH and its value accumulation?
Max:I think if Ethereum L1 becomes the main place where people trade, then ETH will become valuable, and other arguments about ETH as a currency, as a store of value are too illusory. The most direct explanation is: "If you want to trade on Ethereum, you have to pay ETH", which is the reason that resonates most in my mind.
If you think ETH is performing poorly, you should look at the performance of Arbitrum and Optimism tokens, which are even worse. So there will be some attempts to increase the value of these tokens, such as allowing them to capture more MEV or fee income, or allowing L2 to support native tokens to pay gas fees. I think this is a big problem. If L2 starts paying gas fees with ARB or OP, they will become more and more like independent blockchains rather than part of Ethereum.
There is also a strange argument that we need L2 to pay data availability fees with ETH so that ETH can become a currency. So why can’t they just pay execution fees in ETH on L1? This ETH as money argument is a little confusing to me. As a newer Ethereum participant, I don’t find the ETH as money argument that appeals to me that much. Instead, the ETH as a useful asset, such as being used to trade, swap tokens on Uniswap, or take out loans, the ETH as “digital oil” argument is more appealing to me.
Ryan: You mentioned a question that I have to ask, which I call "Samani's paradox", which is that the people who think that L2 is parasitic and that they will suck all the value from Ethereum L1 are usually the same people who think that L2 tokens will go to zero and have no value. So my question is, where is the value going? Is it in the L1 tokens, or is it in the L2 tokens?
Max: Yeah, I think L2 has other issues to solve, such as their issuance and whether anyone is willing to hold these tokens for the long term. Part of the reason why the tokens are underperforming is that they issued too many tokens, which led to too much selling. If there are more sales than purchases, the token price will naturally go down. One solution is to start directing more value to these L2 tokens. If L2 developers have most of their net worth in L2 tokens, and those tokens don’t perform well, they’ll do things that are not aligned with Ethereum but aligned with L2 tokens, and that’s a direct point of conflict between L2 and L1 — they have their own native token value.
The overall market value of L2 tokens has been falling since April this year, and at the same time, a large amount of token quotas are unlocked every week; data source: Token Terminal, Token Unlock
David: What would you call on the audience or other researchers who are motivated? What do you want them to do?
Max:First, I would say, contact your “senators”, and here “senators” refers to the Ethereum core developers, and tell them that we need to shorten the block time. Especially if you are an application developer who is considering leaving Ethereum L1 due to long block times, then contact your core developers and tell them that you urgently need to reduce block times in the upcoming hard fork, otherwise you will leave L1.
Second, if you are a solo staker, please share whether your network connection can support more bandwidth than it currently does. I feel that many times the abstract arguments we hear about solo stakers do not actually match reality. Many solo stakers actually use very good hardware and network connections, so bandwidth is not an issue for them. Therefore, if you are a solo staker, please stand up and tell everyone that you actually have enough hardware and network connections to support more bandwidth needs.
欢迎加入律动 BlockBeats 官方社群:
Telegram 订阅群:https://t.me/theblockbeats
Telegram 交流群:https://t.me/BlockBeats_App
Twitter 官方账号:https://twitter.com/BlockBeatsAsia