Vitalik: What Happens After ETH1 and ETH2 Merge?
This is the script of Vitalik Buterin’s speech on World Blockchain Conference 2021: (for video, click here)
Hello, everyone. I’m excited to be invited by 8BTC to speak on World Blockchain Conference held in Hangzhou Future Sci-tech City.
I’m sorry because of COVID-19, I can’t be there to meet you in person. But I’d like to share with you something about Ethereum’s latest progress of research and tech.
Today I’m going to talk about what happens after the merge.
We’ve talked a lot about many of these things that are happening to, if here I am, over the next time half year or so. We had the Berlin Hard Fork which happened in April. Then up upcoming very soon, we have the Altair Hard Fork on the Proof of Stake chain. The London Hard Fork under the existing Proof of Work chain. At some point, maybe about half a year from now may be more, we will get the merge, which is when the Proof of Work chain finishes and everything on the Proof of Work chain gets moved over onto the Proof of Stake chain.
But what we can talk about more is what happens after. So what happens after the merges finished? What are some upgrades to Ethereum protocol that will happen after that point?
What will happen after the merge?
The first major thing that will happen after the merge is what I call the post-merge cleanup fork. The idea here is basically that the merge is being done in the way that allows it to happen as quickly as possible.
The merge is fairly minimalistic. So there’s a lot of things that the Hard Fork that implements the merge that does not do. For example, it does not enable withdrawals is a one big example of this. Right now, if you have a yet deposit in the Proof of Stake system, you have no way of taking it out. After the merge, you still have no way of taking it out. The ability to take your deposit out and withdraw the deposit and your rewards, that’s something that will be implemented in this first hard fork that comes after the merge.
Aside from that, some different technical changes are removing this mechanism called ETH1 data voting, which is the mechanism that the Proof of Stake chain uses right now to talk to the Proof of Work chain.
Changing the execution layers of serialization, potentially removing RLP and more places adding SSZ, potentially agreeing that clients should stop trying to download the Proof of Work chain before the merge. So simplify clients, delete a lot of code that’s not necessary anymore, add some opcodes to access some more information about the Proof of Stake change. Just a lot of small changes that could have come at the same time as the merge, but to make sure that the merge itself happens more quickly, we’re delaying until this stuff work after the merge.
So after this, sharding is the one next major improvement. This is a sharding for data. The shards that we are adding has 64 shards that will each have 512 kilobyte blocks that come once every 12 seconds on each shard.
And these blocks just contain data, right? There’s no transaction execution happening inside of shards. The purpose of these shards is to give more data space to allow Rollups to use that space to have more scalability.
So today, Rollups can scale up to about 4,000 to 5,000 transactions per second. If hypothetically, the entire Ethereum ecosystem works us which to Rollup, but with a data shards, we have maybe 20 to 50 times more space. It would be possible for Rollups to go up to 100,000 transactions per second and in the future, even more than that.
Sharding will be added with a fewly basic level of security as first and then more security will be added later. And more improvements to sharding will come over time.
Data availability sampling. This technology that we came up with and developed to improve the security of sharding, basically to allow nodes to verify for themselves that the data in the shards has actually been published without nodes needing to download all of that data themselves. So if you just search for the phrase data availability sampling, there’s a lot more information about this in various places on the Internet. In my opinion, that’s a very fascinating technology and is very important for increasing the security of sharding and making sure that the increased scaling and the increased amount of data being a past rule on a sharded network does not lead to a risk that blocks get accepted where nobody can access the data, because that would break relatives and that would break applications. So just increasing the security of sharding more generally.
And there’s also other security improvements. So this is just a various different cryptographic technology is that have fancy names like a Single Secret Leader Election, Verifiable Delay Functions, Proof of Custody.
I don’t have time to go into what all of these things are. But just very quickly. Single Secret Leader Election, for example, is just makes it harder to see who the proposers of future blocks are going to be. That makes it more difficult to attack the network, because you don’t know like which nodes you need to attack if you want to prevent that blocks from being published.
Verifiable Delay Functions make randomness in Ethereum. More secure Proof of Custody forces nodes to actually keep and to validate block data. So Proof of Custody can increase decentralization and just reduces the risk that everyone gonna start using centralized services to run all of their nodes. So also important.
Some improvements to the execution layer.This is the part of the system that handles things like transactions, the EVM, smart contracts, accounts.
Address extension is one thing that we are working at increasing addresses from 20 bytes to 32 bytes.
Verkle trees is to remix stateless clients possible. A stateless client is a client that verifies the chain without actually having to store the state of the chain locally, without having to store locally all of the accounts and all of the contracts. Instead, a stateless client would receive and verify a block that contains the information that is like the part of the state. So the only the accounts and the contracts that were accessed in that particular block together with a proof that those accounts that are being included in the block actually are valid and they are the correct state at that particular time. So also a very fascinating technology. This would just allow block verification to happen without nodes requiring to have a really any hard drive space.
State expiry is another solution to Ethereum growing a state size problem, which basically just says clients stop needing to store accounts and other storage and other objects that have not been accessed recently. Instead, anyone trying to access those objects just need witnesses. They need to provide proofs to be able to recover them.
Account abstraction is a technology to increase ease of use for smart contract wallets and for some other applications, basically make it as easy to use a smart contract wallet as it is to use regular accounts today. This has used cases like multisig-wallets, social recovery wallets and many other use cases. There are different paths to doing this and we’re exploring all of them, but trying to actually get this kind of finished in them included in some form.
EVM improvements. So improve Ethereum Virtual Machine to make it possible to add or implement more advanced forms of cryptography, more efficiently remove the need for precompile, so we can simplify the protocol. A lot of uh… Actually, not that many improvements that hopefully but a few but very important improvements to the EVM.
Long-term things: improving consensus algorithm and post quantum security
So that is what is coming in the near-term future. But we can also talk about what’s happening in the long-term future. So this is maybe 2 to 5 years from. Maybe if things take longer than even after that, there’s a few things there.
One is continuing to improve Ethereum’s consensus algorithm. So we’re looking at CBC Casper for this. We’re also looking at some other things, but basically just continuing to improve the simplicity efficiency and the security of Ethereum consensus protocol.
ZK-Snarks or Zero Knowledge Proofs is the very important and very powerful technology. The idea behind the ZK-Snarks is that they are proofs that prove that for example, everything in a block was computed correctly, that all of the transactions are valid, that the entire block is valid, where that proof can be verified very quickly. Even if a block is very large and very complex and has a large amount of computation, the proof that block is correct, it takes a long time to generate the proof. But once the proof is generated, the proof can be verified very quickly.
Snarks have a lot of use cases. We can snark the Beacon Chain. Snark the Proof of Stake chain to make the Proof of Stake chain easier to verify and participate in. We can Snark the Ethereum Virtual Machine or potentially some other virtual machine that extends the Ethereum Virtual Machine, make it easier to add full smart contract capabilities to shards, if we want to do that. A lot of things that we can, add Zero Knowledge Proofs around. And if we add Zero Knowledge Proofs to things, then this will make it much easier to run in an Ethereum node and for anyone to verify that Ethereum blocks actually are correct. So it will make it even harder to attack the network.
Post quantum security is also something which is very important. Quantum computers are coming and they will be here at some point, not very soon, but definitely at some point in the future. And being secure against quantum computers is definitely something which is very important.
For quantum security, the good news is that there are solutions, right? There are changes that can be made to the protocol to make the protocol more secure against quantum computers and so we can use cryptography that is known to already be quantum resistant. And the best quantum resistant cryptography actually is hashes, right? So even today, hashes, functions like SHA 256 and SHA 3 are already quantum resistant. What we can do is we can replace existing cryptography with hash-based cryptography. We can replace some Snarks with Stark. We can replace BLS signatures with aggregate signatures that use the same technology as Starks. We can replace verkle tree that we talked about to make proof sizes for stateless client smaller with a merkle tree that uses Starks. We can basically replace everything with Starks, potentially replace some things with lattice. The Ethereum Foundation has a cryptography team that is a looking more and more at lattice-based technology as well.
But these are things that only needs to happen like maybe 5 to 15 years in the future. So this is not in near term concern. But for the time when this becomes a concern, we already know the kinds of changes that will need to be made. So those are some of the long-term improvements. At that point, I think we’ll basically be done, right?
Future of Ethereum: layer 1 for decentralization, layer 2 for innovation
I think Ethereum’s general technical vision is that we’re trying to improve quickly in the short term and settle down in the long term, because blockchains in the long term, I think do need to settle down and then you need do to become more stable. They do need to just give more guarantees to the users that things are not going to change, so that users can feel safe and developing on them.
But in the short term, there’s just all of this technology that has been developed after bitcoin came into existence, and even after the original version of Ethereum came into existence.
And this technology, it’s just very powerful. And it can really improve the security of Ethereum, the scalability of Ethereum, the privacy of Ethereum, all of these properties. And it’s just really important to be able to add these things. So that the blockchain can be secure, so that we do not have this very high transaction fees, so that the blockchain can be easier to use.
Over time, the expectation is that once the blockchain itself has enough functionality, anything beyond that can be done at layer 2. So there’s still room for improvements at layer 1, but more things can be done at layer 2. And layer 1 can focus on decentralization and eventually layer 2 will be the main place where we see people doing the most innovation. But we’re not there yet, right? I think that is a place that we will be much closer to 2 years from now.
But right now, they’re still and after the merge, all of these other improvements to just improve the scalability of Ethereum protocol and to make Ethereum protocol easier to use and to make the Ethereum protocol more secure.
And those things are very important, right? But these things are all happening in parallel, right? There are teams working on Proof of Stake, and working on the merge. There’s a team working on sharding. There’s a team working on account abstraction. There’s a team working on verkle trees. So there’s also many projects working on layer 2 scaling and continuing to improve scalability of Ethereum. Optimism is one of the layer 2 projects and recently launched support for Uniswap. So scaling is happening and it is happening more and more quickly.
And I think over the next 2 or 3 years, we’re going to see it become much cheaper to use Ethereum. And we’re going to see it become more possible for many more kinds of applications to use Ethereum. And the Ethereum ecosystem is going to become a lot more interesting and fun. So thank you!