Vitalik 香港演讲：谈以太坊未来五年，扩容、抗量子与 zk 验证

OK, hello, everyone.

OK, hello, everyone.

Good morning.

Good morning.

So, where the uh ethereum protocol is uh going? So, I think uh we uhHave seen um.

So, where the uh ethereum protocol is uh going? So, I think uh we uhHave seen um.

A lot of uh big changes in the yeah ethereum ecosystem over over the last couple of years.

A lot of uh big changes in the yeah ethereum ecosystem over over the last couple of years.

I think we've we've also seen a lot of changes um happening outside of the yeah ethereum ecosystem.

I think we've we've also seen a lot of changes um happening outside of the yeah ethereum ecosystem.

Including um all, including what's become possible with AI, including quantum uh computing, in a potentially coming up very soon.

Including um all, including what's become possible with AI, including quantum uh computing, in a potentially coming up very soon.

Including advances in formal verification, including cryptography zero knowledge proofs, and uh I think um one of the things that we have been uh really yeah doing a lot of is uh rethinking like what What really makes sense uh to use ethereum for?

Including advances in formal verification, including cryptography zero knowledge proofs, and uh I think um one of the things that we have been uh really yeah doing a lot of is uh rethinking like what What really makes sense uh to use ethereum for?

Why does ethereum, like what properties does ethereum need to have?

Why does ethereum, like what properties does ethereum need to have?

Why ethereum needs to have these uh properties, and uh like how this fits into the yeah.

Why ethereum needs to have these uh properties, and uh like how this fits into the yeah.

Roadmap for the Ethereum protocol over the next 5 years.

Roadmap for the Ethereum protocol over the next 5 years.

So, what is Ethereum for, right?

So, what is Ethereum for, right?

I think of it as having a two big functionalities.

I think of it as having a two big functionalities.

so one is that ethereum is a public billboard.

so one is that ethereum is a public billboard.

so it is a a place where an appli- applications can post messages and.

so it is a a place where an appli- applications can post messages and.

Everyone sees what messages were posted in what order.

Everyone sees what messages were posted in what order.

These messages can be anything.

These messages can be anything.

They could be transactions, they could be hashes, they could be encrypted data, they can be many other things.

They could be transactions, they could be hashes, they could be encrypted data, they can be many other things.

And actually there's a lot of opportunity for applications to use both Ethereum as a place to post to post a data.

And actually there's a lot of opportunity for applications to use both Ethereum as a place to post to post a data.

And at the same time, other kinds of protocols to do uh to interpret that data, yeah, to decrypts the data, to do compu- uh to do computation on that data.

And at the same time, other kinds of protocols to do uh to interpret that data, yeah, to decrypts the data, to do compu- uh to do computation on that data.

And then also the second is the computation, right?

And then also the second is the computation, right?

So, basically what Ethereum lets you do is it lets you have shared digital objects that are controlled by code.

So, basically what Ethereum lets you do is it lets you have shared digital objects that are controlled by code.

These digital objects can be many things.

These digital objects can be many things.

They can be Eath the asset, it could be ERC 20s, um, it can be NFTs, it could be NFTs that have meaning outside of a theorem.

They can be Eath the asset, it could be ERC 20s, um, it can be NFTs, it could be NFTs that have meaning outside of a theorem.

um, So ENS is one example.

um, So ENS is one example.

um, It can also even mean uh control over organizations, so things like DAOs are one example, but uh you can do many things.

um, It can also even mean uh control over organizations, so things like DAOs are one example, but uh you can do many things.

So both of these are highly valuable for decentralized applications that guarantee self-sovereign security.

So both of these are highly valuable for decentralized applications that guarantee self-sovereign security.

Verifiability and fair participation to all of their users.

Verifiability and fair participation to all of their users.

So, what do we mean by this?

So, what do we mean by this?

So, self sovereign basically means you as a user can participate, can verify, can be sure that you are secure.

So, self sovereign basically means you as a user can participate, can verify, can be sure that you are secure.

All with your own infrastructure.

All with your own infrastructure.

You do not need to trust any third party.

You do not need to trust any third party.

You do not need to trust a third party to run Ethereum, and you also do not need should not need to trust a third party outside of Ethereum, if you do not want to.

You do not need to trust a third party to run Ethereum, and you also do not need should not need to trust a third party outside of Ethereum, if you do not want to.

um, So, verifiability, the ability to verify that the chain is a pers- is progressing correctly, and uh to ver- uh to verify that uh.

um, So, verifiability, the ability to verify that the chain is a pers- is progressing correctly, and uh to ver- uh to verify that uh.

The yeah, uh, everything that's happening is what is uh supposed to be happening, and also the ability to guarantee anyone's right to actually publish messages to this billboard.

The yeah, uh, everything that's happening is what is uh supposed to be happening, and also the ability to guarantee anyone's right to actually publish messages to this billboard.

So, this is the core.

So, this is the core.

We should think of ethereum as being a technical building block, and think about all of the yeah, what kinds of applications this technical building block enables.

We should think of ethereum as being a technical building block, and think about all of the yeah, what kinds of applications this technical building block enables.

Most interesting applications will be a combination of on chain and off chain.

Most interesting applications will be a combination of on chain and off chain.

This includes ENS, uh, this includes prediction markets.

This includes ENS, uh, this includes prediction markets.

Prediction markets have an on chain component, which is the yeah the assets that they create for every event.

Prediction markets have an on chain component, which is the yeah the assets that they create for every event.

The ability to trade those assets.

The ability to trade those assets.

um, Also, they have uh off chain components.

um, Also, they have uh off chain components.

One of them is the oracle, another one is that sometimes uh there's a prediction market designs where trading where or and order matching happen off chain, and then also anything with privacy.

One of them is the oracle, another one is that sometimes uh there's a prediction market designs where trading where or and order matching happen off chain, and then also anything with privacy.

So for example There have been, there has been decades of work on uh cryptographic protocols that uh make it easy yeah to, or or that try to enable um secure electronic voting.

So for example There have been, there has been decades of work on uh cryptographic protocols that uh make it easy yeah to, or or that try to enable um secure electronic voting.

And a lot of these protocols, they often rely on a yeah a public billboard, a place where people can post information, in this case, so they are encrypted votes.

And a lot of these protocols, they often rely on a yeah a public billboard, a place where people can post information, in this case, so they are encrypted votes.

In a way that that guarantees everyone's ability to participate.

In a way that that guarantees everyone's ability to participate.

um, Anything with privacy has to have an on-chain part publishing data on-chain, and an off-chain part interpreting that data.

um, Anything with privacy has to have an on-chain part publishing data on-chain, and an off-chain part interpreting that data.

If he be, interpreting the data has to happen off chain in a yeah private protocol.

If he be, interpreting the data has to happen off chain in a yeah private protocol.

So, we we talk a lot about L twos, and the way that I think about L twos, and the way that I think about the types of L twos that make sense, is that I think The type of L two that does not make sense is just you take a theorem.

So, we we talk a lot about L twos, and the way that I think about L twos, and the way that I think about the types of L twos that make sense, is that I think The type of L two that does not make sense is just you take a theorem.

you make a clone of a theorem, you make it 100 times bigger, you make it more centralized, and you are done.

you make a clone of a theorem, you make it 100 times bigger, you make it more centralized, and you are done.

The type of L two that I think make sense is, you at, you start to look through these applications, you ask what are the off chain components that they need, what are the parts that they need other than the Ethereum L one.

The type of L two that I think make sense is, you at, you start to look through these applications, you ask what are the off chain components that they need, what are the parts that they need other than the Ethereum L one.

and then you build that.

and then you build that.

So what this means for a theorem we need scaling for data We need the ability to publish much more data on chain and Pura达斯 which came in the yeah most recent hard fork last year already includes that but we still need to push that further Scaling。

So what this means for a theorem we need scaling for data We need the ability to publish much more data on chain and Pura达斯 which came in the yeah most recent hard fork last year already includes that but we still need to push that further Scaling。

of compute is also important because scaling compute that is that is part of the ethereum chain allows different applications to compose with each other to talk to each other without requiring intermediaries

of compute is also important because scaling compute that is that is part of the ethereum chain allows different applications to compose with each other to talk to each other without requiring intermediaries

You, now, this is uh the straw map.

You, now, this is uh the straw map.

If you go to a straw map dot org, uh, you can find this.

If you go to a straw map dot org, uh, you can find this.

This is a a proposed roadmap for the next 5 years of Ethereum.

This is a a proposed roadmap for the next 5 years of Ethereum.

So, the core goals for the protocol in the short term, one is short term scaling, continuing to aggressively increase the gas limit.

So, the core goals for the protocol in the short term, one is short term scaling, continuing to aggressively increase the gas limit.

One of them is to begin the yeah, zkEVM rollout.

One of them is to begin the yeah, zkEVM rollout.

I'll talk about this.

I'll talk about this.

zkEVMs allow Ethereum to scale much more, have much more computation.

zkEVMs allow Ethereum to scale much more, have much more computation.

While still making it very easy to verify the chain.

While still making it very easy to verify the chain.

um, Early post quantum readiness.

um, Early post quantum readiness.

uh, We yeah have been thinking about quantum computing for years.

uh, We yeah have been thinking about quantum computing for years.

We have recognized for years that it is a problem, and.

We have recognized for years that it is a problem, and.

We have some things in very soon in the short term that will improve Ethereum's quantum safety, and the entire roadmap, and already like this roadmap by the end of it, Ethereum will be fully quantum safe for all parts of it, and very efficiently.

We have some things in very soon in the short term that will improve Ethereum's quantum safety, and the entire roadmap, and already like this roadmap by the end of it, Ethereum will be fully quantum safe for all parts of it, and very efficiently.

Improving the block building pipeline, and improving support for privacy.

Improving the block building pipeline, and improving support for privacy.

So, short term scaling, there are a number of EIPs, uh proposals that will go into the next Ethereum hard fork, block off access list which enable paralization.

So, short term scaling, there are a number of EIPs, uh proposals that will go into the next Ethereum hard fork, block off access list which enable paralization.

Gas re-pricing, which enables uh more uh more efficiency, it's uh also makes it safer safer for gas limits to be higher.

Gas re-pricing, which enables uh more uh more efficiency, it's uh also makes it safer safer for gas limits to be higher.

um, E PBS makes it safer for it to take a longer to verify a theorem blocks.

um, E PBS makes it safer for it to take a longer to verify a theorem blocks.

And improvements in ethereum's ability, uh, and ethereum nodes ability to download the state.

And improvements in ethereum's ability, uh, and ethereum nodes ability to download the state.

um, EIP 8 141.

um, EIP 8 141.

This is the EIP, account abstraction proposal.

This is the EIP, account abstraction proposal.

It's both very simple and very powerful.

It's both very simple and very powerful.

Basically, a transaction is a list of calls, and so one of the calls could be verification, one of the calls could be execution, and this makes it very easy for a theorem to have native support for smart contract wallets.

Basically, a transaction is a list of calls, and so one of the calls could be verification, one of the calls could be execution, and this makes it very easy for a theorem to have native support for smart contract wallets.

For paying for other people's transactions, for quantum resistant signature algorithms, and for privacy protocols.

For paying for other people's transactions, for quantum resistant signature algorithms, and for privacy protocols.

So, this makes Ethereum more general purpose, and so it supports many thing, many yeah functionalities at the same time.

So, this makes Ethereum more general purpose, and so it supports many thing, many yeah functionalities at the same time.

um, Quantum resistance.

um, Quantum resistance.

So, quantum resistant signature algorithms exist.

So, quantum resistant signature algorithms exist.

They have existed for 20 years.

They have existed for 20 years.

We know what they are.

We know what they are.

We know how to build them.

We know how to build them.

The problem is that they, that that they are less efficient.

The problem is that they, that that they are less efficient.

So, a quantum resistant signature takes two or three kilobytes compared to sixty four bytes right now, and it takes 200 200,000 gas on chain.

So, a quantum resistant signature takes two or three kilobytes compared to sixty four bytes right now, and it takes 200 200,000 gas on chain.

Compared to 3,000 gasoline chain right now.

Compared to 3,000 gasoline chain right now.

And so, there are two types of signatures that we can use.

And so, there are two types of signatures that we can use.

One of them is hash based, the other is lattice based signatures.

One of them is hash based, the other is lattice based signatures.

And then the idea is that we add vectorization into the EVM.

And then the idea is that we add vectorization into the EVM.

Basically, the same kind of logic that allows computers to run AI very quickly.

Basically, the same kind of logic that allows computers to run AI very quickly.

We used like basically process many numbers in parallel at the same time.

We used like basically process many numbers in parallel at the same time.

The same logic uh can be uh can be used for lattice based signatures.

The same logic uh can be uh can be used for lattice based signatures.

Basically working aggressively to make quantum resistant signatures much more efficient.

Basically working aggressively to make quantum resistant signatures much more efficient.

Mid-term goal, the state. So the state stores ethereum accounts, balances, smart contracts.

Mid-term goal, the state. So the state stores ethereum accounts, balances, smart contracts.

Scaling execution is easier, scaling the state is harder.

Scaling execution is easier, scaling the state is harder.

There is a lot of work that we yeah have to do on this.

There is a lot of work that we yeah have to do on this.

So that is all the short term.

So that is all the short term.

The longer term And this is where we really want Ethereum to go, right?

The longer term And this is where we really want Ethereum to go, right?

So Ethereum uh is what is not meant to be Is is not meant to compete with high frequency trading platforms.

So Ethereum uh is what is not meant to be Is is not meant to compete with high frequency trading platforms.

Ethereum is not meant to be the fastest chain.

Ethereum is not meant to be the fastest chain.

Ethereum is meant to be the secure chain, the decentralized chain, the chain that continues to be online, the chain that you can rely on.

Ethereum is meant to be the secure chain, the decentralized chain, the chain that continues to be online, the chain that you can rely on.

So, one goal is maximally safe consensus, meaning If the net, if the network is uh if the network is secretness, meaning if the network does not have problems, it can survive 49 percent of the of the and of the notes having of having faults.

So, one goal is maximally safe consensus, meaning If the net, if the network is uh if the network is secretness, meaning if the network does not have problems, it can survive 49 percent of the of the and of the notes having of having faults.

And it can actually survive almost all of the nodes going offline.

And it can actually survive almost all of the nodes going offline.

So, same property as Bitcoin.

So, same property as Bitcoin.

And then, if the yeah network is uh has problems, then you still have finality with 33 percent security.

And then, if the yeah network is uh has problems, then you still have finality with 33 percent security.

That's one, part two is formally verify everything, and we've been aggressively starting to use AI to generate mathematical proofs that mo-, that the yeah software that will be running the long term version of a theorem.

That's one, part two is formally verify everything, and we've been aggressively starting to use AI to generate mathematical proofs that mo-, that the yeah software that will be running the long term version of a theorem.

Actually has the properties that it is supposed to have.

Actually has the properties that it is supposed to have.

We've made already made progress that would have been impossible two years ago.

We've made already made progress that would have been impossible two years ago.

AI is progressing quickly, and so we are using that.

AI is progressing quickly, and so we are using that.

um, Full quantum resistance, also maximum simplicity, keep long term ethereum as simple as possible, and maximally future proof, right?

um, Full quantum resistance, also maximum simplicity, keep long term ethereum as simple as possible, and maximally future proof, right?

So ethereum needs to pass the walk away test. Ethereum needs to be a platform that.

So ethereum needs to pass the walk away test. Ethereum needs to be a platform that.

You can rely on, even if there are no core devs left, right?

You can rely on, even if there are no core devs left, right?

Because uh like, and and this is the same thing that Bitcoin strives for, this is the same thing that you have to strive for, if you wants to be a base for long term uh for long term secure assets, for uh for long term digital objects. We need, you.

Because uh like, and and this is the same thing that Bitcoin strives for, this is the same thing that you have to strive for, if you wants to be a base for long term uh for long term secure assets, for uh for long term digital objects. We need, you.

need to be something.

need to be something.

Who is ongoing safety does not depend on one team continuing to exist and one team always continuing to work.

Who is ongoing safety does not depend on one team continuing to exist and one team always continuing to work.

Lean consensus, so lean consensus takes The best of both worlds between the yeah bitcoin style approach, but that that's called the available chain, and the yeah BFT approach, that's the the uh um uh finality.

Lean consensus, so lean consensus takes The best of both worlds between the yeah bitcoin style approach, but that that's called the available chain, and the yeah BFT approach, that's the the uh um uh finality.

It gets both with optional security properties, quantum safe, and fast finality.

It gets both with optional security properties, quantum safe, and fast finality.

So, finality within 1~3 slots basically expect the chain to finalize within like ten or maybe ten or twenty seconds, maybe less.

So, finality within 1~3 slots basically expect the chain to finalize within like ten or maybe ten or twenty seconds, maybe less.

ZK-EVMs,ZK-EVMs let you verify the chain without having a big computer that uh runs everything yourself.

ZK-EVMs,ZK-EVMs let you verify the chain without having a big computer that uh runs everything yourself.

Everyone should be verifying the chain before they trust it, even your phone should be verifying your chain, even IoT devices should be verifying the chain.

Everyone should be verifying the chain before they trust it, even your phone should be verifying your chain, even IoT devices should be verifying the chain.

And as zkEVMs are already sufficiently fast to prove real time EVM execution, the goal this year is to make them sufficiently safe.

And as zkEVMs are already sufficiently fast to prove real time EVM execution, the goal this year is to make them sufficiently safe.

And the goal this year is to start with small percentages of the network using zkEVMs, and then increasing that over time, and to move to pro- to to zkEVMs being the primary way to validate the chain by 2028.

And the goal this year is to start with small percentages of the network using zkEVMs, and then increasing that over time, and to move to pro- to to zkEVMs being the primary way to validate the chain by 2028.

And so this allows Ethereum to scale much more, process many more transactions without becoming centralized.

And so this allows Ethereum to scale much more, process many more transactions without becoming centralized.

So, what is division behind these things, right?

So, what is division behind these things, right?

So, Ethereum is the world computer.

So, Ethereum is the world computer.

Ethereum is both a global shared layer for making commitments for publishing.

Ethereum is both a global shared layer for making commitments for publishing.

Data that represents actions, that actions that the different users wants to take.

Data that represents actions, that actions that the different users wants to take.

It is a a place where you can publish data, where you can prove the data has been published.

It is a a place where you can publish data, where you can prove the data has been published.

And you can also prove that data has not been published, and that is very open for everyone to use.

And you can also prove that data has not been published, and that is very open for everyone to use.

And it's also a global shared layer for high value guaranteed Execution of rules.

And it's also a global shared layer for high value guaranteed Execution of rules.

Ethereum needs to be maximally robust and maximally easy to verify.

Ethereum needs to be maximally robust and maximally easy to verify.

I think in the future with AI, it will actually be easier than ima- than we've imagined.

I think in the future with AI, it will actually be easier than ima- than we've imagined.

To write actually secure software, if you wants to be secure.

To write actually secure software, if you wants to be secure.

But if people do not wants to be secure, then software will have ten times more bugs than before, it will be attacked ten times more than before.

But if people do not wants to be secure, then software will have ten times more bugs than before, it will be attacked ten times more than before.

And so, ethereum as a blockchain needs to be on the secure side first, on the decentralized side first.

And so, ethereum as a blockchain needs to be on the secure side first, on the decentralized side first.

And when those things are guaranteed, make that security available to as i- to users as much as possible.

And when those things are guaranteed, make that security available to as i- to users as much as possible.

Possible.

Possible.

So, Ethereum is there to build decentralized applications that guarantee self-sovereign security, verifiability, and fair participation for their users.

So, Ethereum is there to build decentralized applications that guarantee self-sovereign security, verifiability, and fair participation for their users.

This includes financial, this includes decentralized social, this includes identity, this includes applications that have one part financial and one part not financial.

This includes financial, this includes decentralized social, this includes identity, this includes applications that have one part financial and one part not financial.

Includes things like which includes things like ENS, includes things like prediction markets, it includes many things.

Includes things like which includes things like ENS, includes things like prediction markets, it includes many things.

And uh Ethereum it will make building applications that are like this um easy and default, and This is the core goal that the yeah roadmap for the next four years is uh designed around. Thank you.

And uh Ethereum it will make building applications that are like this um easy and default, and This is the core goal that the yeah roadmap for the next four years is uh designed around. Thank you.