Bankless Interview with Vitalik: Centralization Risks After Transitioning to PoS Are "Overhyped"

Source: Bankless Podcast

Compiled by: Mary Liu, Derrick Chen, BitpushNews

After the Ethereum Merge was completed, Vitalik Buterin (hereinafter referred to as V God) accepted an interview with Bankless, sharing his views on the Merge, Ethereum roadmap, staking, and more. The Bitpush editorial team has summarized the key points from the interview.

Host: After comprehensive consideration from various parties, we feel that the final PoS design is a simplified version and does not meet everyone's ambitions. Do you think if we wait a bit longer, we will find a new design, or is the current design already the final and logically optimal design?

V God: I feel you raised two questions. The first question is whether there is a better form of PoS chain in the long term. The second question is whether our research team could have found a better transition method if they had spent more time researching.

For the first question, we are still far from the optimal state of a PoS chain, and there is still much improvement to be made. For the second question, I think we can say that this is the best form of the Merge. If I had the chance to go back to 2014, I might have used a simpler design and transitioned to PoS in 2018. We have now achieved some of the goals we set at the beginning, but there are also some security issues that have arisen, and we are working to address them.

So if I were to make any changes, I might suggest a simpler PoS and an earlier Merge. Of course, I also hope that other PoW chains will migrate to PoS mechanisms in the next four years.

Host: For everyone, the Merge came too late. There must have been some attacks and community disagreements in between, and of course, there were some benefits; the PoW mechanism has benefited miners and many other entities. So, is there anything particularly important that you would want to redo and change?

V God: If we could have avoided those attacks and disagreements, that would have been best. I think the Layer 2 roadmap could have been optimized more; we could have spent more resources on Optimism and Arbitrum.

I feel that many of the things we did were like "discovering a hundred wrong ways to make a light bulb." We spent a lot of resources exploring different paths and trial and error. Regarding scalability, it took us a long time from realizing the need for scalability to developing solutions like Plasma and Rollup, and we also found that many people's goals were not that complicated; for them, scalability might not be that necessary.

Bankless Host: Where are we now in the Ethereum roadmap?

V God: If we look at the Merge roadmap, the Merge has been completed, and what remains unfinished in the Merge part includes the post-Merge hard fork with withdrawals, which is clearly one of the next important priorities after the Merge. This is a very simple hard fork, and I think the main debate at this point is whether to do it simultaneously with EIP-4844 or to do the fork first and then do 4844. Distributed validators are still under development; although they are not 100% complete, I guess they have made considerable progress (such as the validator node project DVT aka SSV).

From a longer-term perspective, we have made some great progress on Single Secret Leader Election (SSLE), and the relevant cryptographic principles have taken shape.

Next is Single-slot finality, which is a larger project. I think it is one of the projects that started earlier in the roadmap because people recognized its greater value and importance, and I believe this is one of the things we will have to have a large public discussion with the community about at some point, because single-slot finality has huge benefits, but there are also some costs, such as what final outcomes can be expected with a deposit size of 32 ETH; we hope to reduce the finality time to a single slot even with 32 ETH or similar conditions.

Then there is The Surge; things have clearly been reshuffled, and the new shard design EIP-4844 is more advantageous and is basically waiting for deployment. Next is The Verge - Verkle Tree (further decentralizing the network, allowing individuals to run nodes). I believe we have made significant progress in implementing Verkle trees; the current main bottleneck is that the transition from our existing tree structure to Verkle trees will be a huge development challenge, and there is still debate about how to do this. I think overall, relative to achieving scalability, its priority has been lowered because scalability is very important for Ethereum.

Regarding The Purge, History Expiry EIP-444 has made great progress, and Ban-self-destruct can be deployed at any time. State expiry has been lowered in priority because of proposer-builder separation; basically, if you have PBS and you have Stainlessness, then the number of participants who must hold the entire state is very small, and ordinary validators do not have to hold the state.

Because ordinary validators only need to validate other blocks instead of creating their own blocks, the lowering of the priority of State expiry has indeed given us a lot of freedom to first figure out other things, like making the Ethereum protocol more concise, getting rid of RLP, and cleaning up the block structure, but it may take another year or so to pave the way for this.

Host: Is there a plan to change staking economics at some point, and if so, what would that look like?

V God: I can see one possibility where there are some ways to make deposits and withdrawals at least faster under normal circumstances, for example, if the chain finalizes, allowing a large number of deposits and withdrawals to occur, which would make the validator experience easier and reduce the incentive to participate in staking pools, making it easier to have smaller and more decentralized staking pools, so that’s the first.

The second is that changes to the Miner Extractable Value (MEV) architecture will obviously affect staking economics. We will add an MEV smoothing mechanism, which basically forces MEV income to be redistributed to all ATF validators rather than concentrated in one, which will reduce the variance of staking income and also lower the incentive to be part of a staking pool.

So I think due to some changes that may occur in staking, staking economics will undergo many changes, just like I believe the long-term dream for Ethereum is basically that all a staker has to do is download and validate a bunch of data and then sign. If that’s the case, then staking on a mobile phone would be very feasible, but I think it will take five to ten years to achieve.

Host: Is there a concern about staking centralization after Ethereum transitioned to PoS?

V God: I do think this issue is being overhyped. Let’s look at Bitcoin; three mining pools control over half of the Bitcoin network, and five mining pools control 80%, which is not lower than what is currently done in Ethereum's proof of stake. Many people in the Ethereum research team have criticized certain aspects of LIDO. I know it’s important to defend Ethereum because you love it, but Lido is definitely not a single centralized participant; it’s not like the owners/admins/developers you imagine have the ability to pull the plug and become participants in some kind of attack. It’s a protocol, like some quite a few sub-validators, and each of our sub-validators only has a few percent of the shares.

Clearly, combining Lido, Coinbase, Kraken, and some other participants does add up to quite a bit, which is a concerning issue. I think the good news in the short term is that these participants are all people who love Ethereum and genuinely want Ethereum to thrive, so I think the risk of them doing terrible things in the short term is low. So my point is clear; we should welcome every centralized asset staking provider in the ecosystem. Obviously, good intentions from people are not something we can rely on in the long term, because we want to become a decentralized ecosystem. I think there are some good solutions that are right in the long run.