Summary of the latest AMA from the Ethereum Foundation: New trends and challenges in Ethereum

Compilation & Organization: Karen, Foresight News

On January 10 at 21:00, the Ethereum Foundation research team held its 11th AMA on Reddit. Foresight News reviewed 313 comments, compiling and summarizing the key insights from Vitalik Buterin and members of the Ethereum Foundation. The discussions mainly covered the limitations of EIP-4844 in scaling Ethereum, liquidity fragmentation and composability challenges on Ethereum L2, reflections on Lido's dominance, Ethereum's security budget, native Rollups, intersections between AI and blockchain, privacy, Rollups, and more.

Vitalik's Key Insights

Question: How much can we safely increase the Gas Limit now? What about after Verkle?

Vitalik: Honestly, I think it's reasonable to moderately increase the Gas limit even today. The Gas limit hasn't increased for nearly three years. (Vitalik did some simple calculations, stating that this means the network throughput would increase to around 40 million.)

Question: What are the medium to long-term plans for bringing privacy to Ethereum users? In the future, is it possible that privacy will take a backseat to other more pressing issues (like scaling)?

Vitalik: There are definitely other privacy solutions, like Railway, which I have been using. I expect such solutions to eventually be available on top of existing L2s. We are not close to achieving a single ideal technical path for this yet, but there are a dozen methods, including (i) which ZK-SNARK scheme to use, (ii) how to implement privacy pools, (iii) what kind of UTXO system to build, and (iv) trade-offs between complexity and functionality.

Question: Can you name specific things where blockchain is better than existing non-blockchain technologies?

Vitalik: Many people, especially in politically unstable countries, are using cryptocurrencies to perform basic business functions and save. In some cases, over 10% of the population is affected by this.
ENS is the largest and most well-maintained domain/username system in the world, not relying on a single company or organization, and it has censorship resistance.
In many cases, multi-signature is a more convenient and easier way for groups to manage funds compared to any method offered by traditional finance.
NFTs are being used as a way for artists to raise funds, which was previously impossible. In my view, the average quality in the NFT space is improving.

Question: When will crList be released? Is it sufficient (along with PBS) to provide censorship resistance in the foreseeable future?

Vitalik: Probably around the same time as single-slot finalization, or shortly thereafter.

Question: Is it possible to run other clients as a backup to address the issue of Geth execution client preferences, and would doing so increase the technical workload?

Vitalik: I know that the infrastructure for running multiple clients as a staker is improving. I also expect upgrades like statelessness to further improve the situation.

Question: Do you think there are issues of security fragmentation and social fragmentation in the current Rollup Centric roadmap?

Vitalik: I agree this is a problem. Ultimately, I think it should be the responsibility of wallets, not individual users.

Question: What advantages does the ZK Light Client have compared to Helios, Kevlar, or Ninbus?

Vitalik: ZK light clients can be lighter and potentially cover the entire state transition functionality, not just the sync committee or consensus.

Question: Are you still exploring the possibility of lowering the Ethereum threshold required to run a validator?

Vitalik: There are two points: reducing client resource load and lowering the 32 ETH threshold.

Question: What do you think of Vitalik's 8192 signature post-ssf proposal? If validators are reputation-constrained and the entry threshold is high, wouldn't Ethereum become like Cosmos with DPOS? The prohibition of solo staking seems to pose a significant risk to Ethereum's decentralization.

Vitalik: Option 3 in this proposal is undoubtedly the most popular, largely for this reason.

Question: How can we responsibly abstract the complexity of modular blockchains from end users? Who owns it? Is this a public good in the space? Is it someone from the EF, or various teams building consumer applications?

Vitalik: I feel there are many aspects that can be improved at the wallet level. For cross-L2 transfers, I am optimistic about open permissionless cross-chain trading protocols like UniswapX. The rest is about presenting it to users; existing wallets certainly do a poor job and have room for improvement. I have started to see good progress, for example, Rabby is doing well in aggregating cross-chain views.

Question: Once the EVM is SNARKed, will Rollups appear on Ethereum's roadmap? How will this affect scalability, and what does it mean for users running light clients?

Vitalik: If the sync committee is removed, it’s because the underlying consensus itself has become light enough to allow light clients to verify it directly (or light enough for specialized nodes to SNARK it for light clients to verify). This is the main reason for considering the method of 8192 signatures per slot: it sets a fairly low bar for the complexity of verifying basic consensus.

Question: How likely are we to see Bitcoin implement ZKP verification opcodes or Bitcoin Cash attempt to become a better DA layer?

Vitalik: For technical reasons, interoperability between Bitcoin and Ethereum at that level is unlikely: Ethereum is moving towards fast finality, while Bitcoin sticks with PoW, which does not provide this type of finality (in fact, there may be a week to a month of factual finality). Reorganizations beyond this time may be rejected by the community, but a week or a month is already a long time.

Question: Is there really a plan to increase the block time to 1 minute? Why?

Vitalik: No. There was a brief consideration to increase the block time to 32 seconds or even 64 seconds to better support single-slot finality, but I think the current single-slot 8192 signature strategy is more mainstream.

Limitations of EIP-4844 in Scaling Ethereum

Question: EIP-4844 introduces Blob transactions, but the capacity is limited to about 0.375 MB per block, which only corresponds to 440 TPS of simple transfers on L2. To drive blockchain adoption, the Ethereum community aims to reach VISA-level scale of 2000 TPS. What other options must we take to get closer to this goal besides waiting for complete danksharding (which seems far off)?

Ethereum Foundation researcher Dankrad Feist: I agree that EIP-4844 is just the first step in providing data publication (previously known as data availability). In fact, it only economically separates data publication without scaling, but it does introduce the necessary cryptographic techniques to allow for full scaling. Currently, we are indeed working on some exciting research aimed at achieving a certain degree of scaling as soon as possible. It seems that PeerDAS is relatively easy to implement, and we could launch a basic version in a relatively short time, providing a certain level of scaling, hopefully within a year after 4844 is released. While we won't immediately see scaling to 256 blobs or 1.3 MB/s, we might initially scale to 32 blobs per block, which is 10 times the current amount, and will be positioned between 4844 and full danksharding. Scaling Ethereum's data capacity is absolutely the most important thing in 2024 and even 2025.

Note: The design of PeerDAS aims to reuse well-known, battle-tested p2p components already in production in Ethereum to bring additional DA scale above 4844 while keeping the workload for honest nodes in the same realm as 4844 (downloading <1MB per slot).

Ethereum Foundation researcher Justin Drake: There are various strategies to reduce the data consumed per transaction:

Contract redesign: Gas golfing for L1 contracts will soon be replaced by Data golfing for Rollup contracts. Perhaps Uniswap v5 can be customized for Rollup, and data efficiency will be significantly improved. Transaction amounts can also be appropriately adjusted and compressed.
State differences: By using SNARKs, all witness data of transactions can be removed: signatures, Nonce, Gas limits can be removed, while information like Gas price and tip amounts can be compressed into minimal differences. Only the minimum information necessary to reconstruct the state is required.
Batch compression: Transaction data in batches can be compressed. For example, if 10 transactions in a batch are minting the same NFT, significant advantages can be gained. This can be likened to gzip compression: having more uncompressed data can yield better final compression.
Ethereum Foundation researcher Justin Drake: Technically, EIP-4844 itself is not an expansion solution at all. EIP-4844 lays the groundwork for future scaling by separating Rollup data from the main block. In the future, we may even transition to smarter data availability checks based on sampling methods (data availability sampling, or DAS) without additional hard forks. Work in this area is already underway, and you can explore it by searching for peerDAS.

Liquidity Fragmentation and Composability Challenges on Ethereum L2

Question: How will Ethereum address liquidity fragmentation and composability issues on L2?

Vitalik: I am personally excited about things like https://uniswap.org/whitepaper-uniswapx.pdf. We need to accomplish cross-L2 transfers through open permissionless protocols rather than proprietary cross-chain bridges that use their own tokens and on-chain governance.

Justin Drake: I acknowledge that the fragmentation of liquidity and composability across Rollups (more generally across L2s, including verification) is a problem. However, this is just a temporary transition; we are in the adolescent stage of a Rollup-centric roadmap, and we can regain general synchronous composability across Rollups. It is expected that Ethereum will be able to organically merge and "fix" itself through natural market forces. The future of Ethereum is seamless, and we will not need asynchronous cross-chain.

Regaining general synchronous composability has two elements:

Shared ordering: In any given slot, multiple Rollups choose to share the same orderer, but there are three non-negotiable requirements: trusted neutrality, security, and pre-confirmation, with L1 compatibility as an optional requirement.
Real-time proof: SNARKs can be used in real-time to prove that transaction execution is valid.
Espresso struggles to meet economic security with their tokens, and it is impossible to meet the L1 compatibility optional requirement. The only orderer that can meet the L1 compatibility optional requirement is Ethereum L1 itself. Few realize that Ethereum can also serve as a shared orderer.

My personal view is that in a few years, native Ethereum orderers will become Rollup shared orderers. Rollups using native orderers will merge with L1. But the biggest obstacle is pre-confirmation; Ethereum's 12-second block time is too long. The good news is that pre-confirmation provided by L1 is possible, and the most effective method is to use execution tickets, which requires a hard fork and patience.

Regarding the overall roadmap for Rollup sequencing, my best guess is that we will gradually achieve ordering decentralization: from centralized ordering (the current state) to federated ordering by a trusted committee elected by governance (e.g., Arbitrum's plan), then to decentralized ordering (e.g., Espresso), and finally back to pre-confirmation-based ordering. This will take time, but it is well worth it.

Is Lido's Dominance a Cause for Concern?

Question: If Lido gains 40% of the validator share, do you support actions to reduce the validator share socially (e.g., discouraging attacks, scrutinizing block construction for stETH trades)?

Barnabé Monnot: No, I do not support that. If we want Lido or any sufficiently large participant to become smaller, we should focus on lowering the barriers to entry into this market, including by specifying certain features in the protocol.

Justin Drake: I personally do not support it, but it will be up to the community to decide. I do not believe in Hasu's "LST maximalism," nor do I believe that nearly all staked ETH will ultimately converge to a single LST. I am not too worried about Lido's dominance; it is even possible that Lido's dominance will decline in the coming months and years, one reason being that ETF issuers may choose not to put their ETH into Lido.

Ethereum's Security Budget

Question: How much security budget does Ethereum need? How did you arrive at this number?

Justin Drake: First, let’s distinguish between the concepts of security budget and economic budget.

Security budget: The income earned by consensus participants (stakers in PoS, miners in PoW) over a period (usually a year). For most chains, the security budget is issuance plus MEV. Transaction fees are a subset of MEV. For example, in Ethereum, the security budget is approximately: $2 billion/year (issuance) + $500 million/year (MEV) = $2.5 billion/year. Bitcoin's security budget is about $10 billion per year, with the vast majority going to issuance.

Economic security: This is the value of assets deployed by consensus participants to protect a chain, which is incentivized by the security budget. In PoS, it is the amount staked multiplied by the value of each staked unit. For Ethereum, this is: 28.7 million ETH * $2,370/ETH = $68 billion. In PoW, economic security is the hash power multiplied by the value of deploying each unit of hash power (ASIC costs plus all fixed data center costs: land, power infrastructure, cooling infrastructure). For example, Bitcoin is $550M TH/s * $18/(TH/s) = $10 billion.

The economic efficiency of PoS is about 30 times that of PoW. I personally believe the security budget should be large enough for the amount of staked ETH to represent a quarter of the total supply. Why a quarter? Because it is a power of 2, while half might be too high and one-eighth might be too low. The concept of Ethereum's "minimum viable issuance" suggests issuing the least amount necessary for the staked ETH to reach a quarter of the total supply. Coincidentally, the current issuance plan has the staked ETH at 24%, although it should be noted that the amount of staked ETH is still growing, and MEV distorts the incentives, so this ratio may need to be treated with caution.

To ensure we do not pay too high a price for economic security (especially with the emergence of re-staking), there is a potential upgrade called stake capping, which would reduce issuance (even possibly turning negative) when total staking approaches the cap. Ethereum researchers have reached a rough consensus that stake capping is desirable, and EIPs and further discussions are expected to emerge soon.

Are Native Rollups the Future of Ethereum?

Question: Are enshrined rollups the future of Ethereum? When can we expect to see them on the mainnet?

Justin Drake: Many people, including myself, are trying to rename "enshrined rollups" to "native rollups," as the term enshrined has caused confusion. Native Rollups are Rollups that natively use the L1 EVM for transaction execution, rather than deploying custom Rollups with non-native fault proofs or validity proofs. As it stands, it is currently impossible to build native Rollups on Ethereum; every existing Rollup is custom, including EVM-equivalent Rollups. However, I believe native Rollups are the future of Ethereum. My prediction is that every successful EVM-equivalent Rollup will eventually upgrade to a native Rollup. It will take several years (3 years or more?) to see native Rollups on the mainnet. The reason is that implementing EVM precompiles on the mainnet requires type 0 zkEVM to mature in terms of security, performance, and diversity.

Intersections Between AI and Blockchain

Question: What are some non-obvious potential applications/incentives/intersections between AI and blockchain?

s0isp0ke: We recently wrote a paper focusing on: 1. Blockchain as infrastructure to ensure AI safety and cooperation through trusted commitments; 2. Researching and implementing collaborative AI in existing blockchain games and in scenarios with real-world incentive mechanisms (like MEV).

Justin Drake: An important intersection is "AI currency." We are building programmable digital currencies that AI can host and trade without permission. Advanced AI will pay each other with cryptocurrencies and keep their savings in cryptocurrencies, ultimately making AI the wealthiest entities in the world, partly thanks to cryptocurrencies. This is bullish in the medium term, but incredibly so in the long term. Another intersection with AI is security; AI is much better than humans at identifying vulnerabilities.

Privacy

Question: What are your views on privacy implementation on Ethereum? Vitalik previously mentioned Railway and Nocturne; given government crackdowns, what are your thoughts on the long-term viability of privacy on Ethereum?

Nerolation: There are currently more complex privacy tools that do not simply mix assets but also provide compliance tools simultaneously. Ameen's privacy pool project is a great example, striking a good balance between providing privacy for everyday Ethereum users and making it difficult to be attacked by black hats. Here is a paper on the subject. The only downside is the cost, as on-chain zero-knowledge proof verification is very expensive.

Another option is stealth addresses. They provide a weaker form of privacy—only unlinkability, not untraceability—but are much easier to implement and are significantly cheaper to use. Stealth addresses have great potential for donations, payroll checks, or everyday grocery shopping.

Others

Question: What are your thoughts on VanEck's commitment to provide 5% of BTC ETF profits to core developers? Is this a potential risk? If a large amount of cash starts flowing from TradFi to the ETH core team and becomes a significant part of their income, how would you mitigate centralization risks?

Justin Drake: I personally think it's a good idea. If the specific institution receiving the donations becomes corrupt, a ten-year commitment might be a bit long. I would recommend VanEck to make a shorter-term commitment, with regular reviews and revisions to maintain consistency in the incentive structure.

Question: What are your views on the re-staking risks of Eigenlayer?

Barnabé Monnot: The risk Ethereum faces is a large EigenLayer slashing event, but if the slashing is legitimate, then the Ethereum protocol will self-repair. Higher rewards for lower staked values will attract new stakers. For illegitimate slashing, such as smart contract risks, EigenLayer plans to set up safeguards.

Justin Drake: At this stage, I do not see a practical or meaningful way to "ensure" re-staking. The closest might be to limit staking through reduced issuance (even possibly negative) when staking approaches the cap. In practice, the staking cap can be viewed as a form of "re-staking destruction" mechanism.

Question: Are one-shot signatures the ultimate game for blockchain?

Justin Drake: Yes, I believe one-shot signatures can fundamentally change the outcome of consensus. Of course, we are decades (possibly over 30 years) away from this goal.

Question: What impact would the majority or all L2s moving to DA have on L1?

mikeneuder: There are two scenarios. The first is that new L2s gain massive adoption; however, if L2 has its own native asset used as Gas, and users on L2 do not have to interact with Ethereum (they can directly join L2). Now, the only contribution L2 makes to Ethereum is paying for the data published in blob form.

The second scenario is that different L2s still use Ethereum as a settlement layer but publish blobs to Celestia. This is often referred to as "validium." Therefore, this L2 no longer pays L1 data blob fees. This is unlikely to change its settlement layer, as ETH as the Gas token on L2 is crucial for cross-chain bridges between Ethereum L1.

Justin Drake: There are strong network effects around shared security for DA. If Rollups stop consuming Ethereum DA, it would indicate that Ethereum has lost to some competitors in the settlement game, leading to a loss of fee revenue, a decrease in currency premium, and a contraction of economic security and economic bandwidth. I predict it will slowly but surely die.

Question: What is the structure of the Ethereum Foundation research team? Is there a team responsible for specific risks (like re-staking or liquidity staking centralization)? How many researchers are there?

Barnabé Monnot: The research team has several smaller groups within it; for example, our team is the Robust Incentives Group (RIG). The topics are very diverse, and many teams may be interested in the same subject, so you will see many teams discussing LST or re-staking. Therefore, more focus is on each team's default scope and approach. We have about 35 people in total, with 7 members in RIG.

Question: Should Ethereum strive to maintain the ability to join the validator set without permission?

Francesco: The validator set should definitely never be permissioned; in this sense, if the current validator set cannot handle the activation of validators that meet all requirements, it should be considered a secondary censorship attack. The tricky part is "meeting all joining requirements," as this could range from 32 ETH to other values, depending on the choices made. Personally, I believe it is very valuable for Ethereum to maintain the accessibility of solo staking, or even to improve its accessibility.

Question: What ideas from other blockchain networks are interesting and might be implemented in Ethereum in the future?

Barnabé Monnot: I don't know if they will be implemented, but I am interested in the following ideas:

Multiplicity as a censorship-resistant tool, introduced in the context of Cosmos chains.

Batch block space from Polkadot, which could provide long-term supply guarantees for projects like Rollups that pay for blobs, at a fixed cost.

Question: If Ethereum continues on its current path, do you think other blockchains will still be needed in 20 years?

Barnabé Monnot: Rollups are other blockchains. I believe in heterogeneity; Ethereum is one model, but not the only model, and there are many other ways to build blockchains that make sense for various use cases, and I hope their design choices are validated as well.

Question: Will the amount of ETH required to run staking nodes decrease? If so, when will it decrease? What about transaction fees?

Barnabé Monnot: For transaction fees, we do not have active control; fees are determined by the market. The best thing we can do is scale Ethereum as much as possible through L2 or improvements to L1, which should allow us to increase L1's Gas limit and provide more capacity.

Question: Is the current consensus that Ethereum's monetary policy no longer needs to be modified since the merge and EIP-1559? If not, how would issuance be adjusted? Increased or decreased?

Domothy: There will still be some changes regarding monetary policy, but they will not be as drastic as EIP-1559 and the merge. Most changes will be related to consensus, and we will need to fine-tune it to achieve roadmap goals (for example, single-slot determinism may require adjustments to the staking yield curve, increasing the maximum effective balance, etc.), rather than changes made specifically to adjust issuance.

Question: What are the most underrated application scenarios and use cases that you hope to see more investment and development in?

Justin Drake: I believe a zero-fee "Ethereum World Lottery" is an easily achievable goal. We can now use VDF for lottery-level randomness generation, and the smart contract logic is very simple in other respects.

Nerolation: Combining user-friendly wallets with enhanced privacy, such as account abstraction wallets.