Thanks to Kevaundray Wedderburn, Alex Stokes, Tim Beiko, Mary Maller, Alexander Hicks, George Kadianakis, Dankrad Feist and Justin Drake for comments and revision.
Ethereum goes everything on ZK. Finally, we expect to migrate to the use of ZK tests at all levels of the battery, from the aggregation of signing of the Consensus layer to the confidentiality of Onchain with the side of the customer, and Improve the protocol to be easier and more user -friendly. But the first step will be an L1 ZKEVM.
How we can send an L1 Zkevm in less than a year
The quickest and most sure way to send an L1 ZKEVM is to start by giving validators the possibility of performing customers who, rather than reversing the useful execution charges, check the multiple events (let’s say three) generated by different ZKVMS each proving different EVM implementations. Because the verification of the evidence is so fast and the size of the evidence so succinct, the download and verification of several evidence is very reasonable and allows us to apply the same defense in depth as the diversity of customers existing at ZKVM.
In order for this plan to initially check the evidence of execution outside the chain, everything we need the protocol is a form of pipeline in Glamsterdam to allow more prouvance time.
Initially, we expect few validators to perform ZK customers. Over time, their safety will be demonstrated in production. The EF also puts resources in formal verification, writing specifications, audits and bug bonuses; We plan that adoption will increase slowly.
When a stake supermajority is comfortable managing ZK customers, we can increase the gas limit to a level that would require validators running reasonable equipment to check the evidence instead of reorganizing blocks. Once all validators check the evidence of execution, the same proofs can also be used by a Prepare For native ZK rollers.
Define Prouvance in real time for L1
Our greatest advantage in the execution of this plan is the capacity to exploit the whole ZKVM industry to make the largest ZK application in the world. Many ZKVMS already prove Ethereum blocks and performance breakthroughs are announced each week.
In order to maintain the properties of safety, liveliness and censorship of the resistance of the L1 Foundation, the Ethereum Foundation offers a standardized definition of proof in real time so that the ZKVM teams can work.
On the side of the proof system, the ZKVM targeting proof in real time should target 128 bits of security, which we consider as the right long -term objective for Ethereum L1. However, we are ready to accept a minimum of 100 bits of security during the first months of deployment, to meet short -term engineering challenges to reach 128 bits. The size of the evidence must remain less than 300 KIB and must not count on recursive packaging which use configurations of confidence. We expect Proof Systems to go to 128 -bit security at a time when ZK customers are in production and to further tighten safety requirements (for example with regard to conjectures) as time proving a decrease.
With the time of the current location of 12 seconds and the maximum time to propagate data on the network of ~ 1.5 second, means 10 seconds or less in real time. We expect the ZKVM can prove at least 99% of the Blocks of Mainnet in this window, with the rear end (as well as the synthetic back vectors) attenuated in the future hard forks.
In order to maintain the highest levels of liveliness and resistance to censorship, our definition of real -time prouvance aims to allow “prouvance at home” with the idea that some of the solo stakers who are currently directing the validators of the house will prove. Even if we expect to tighten the resistance to censorship through Forced transaction inclusion Before checking the proofs of ZK, compulsory, the prouvance at home is an important final safeguard.
Given that Prouvance in the Cloud is already cheap with multi-GPU punctuality bodies, the objective for ZKVM teams targeting real-time prouvance will be largely optimized for the execution of promoters on pre-prémal where the specifications are much more limited. Prouvance in real time on site should require a maximum capital expenditure of USD 100,000 (at the time of writing, this requires about $ 80,000 of participation to execute a validator). We expect it to drop over time even if the gas limit increases.
More than the cost of the equipment, the most important constraint for the house proving using GPU is energy consumption. Most residential houses have at least 10 kW which enter the street and some will have circuits for electrical devices or the load of electric vehicles with a capacity of 10 kW. Consequently, proof in real time must be possible on the equipment operating at 10 kW or less.
This brings us to our definition of work of proof in real time:
- Latence: <= 10s for P99 Blocs of Mainnet
- CAPEX on premal: <= 100K USD
- Power on site: <= 10kw
- Code: completely open source
- Safety:> = 128 bits
- Size of evidence: <= 300kib without configuration of confidence
The race towards real time
By DevconConnect Argentina, we hope to see the ZKVM teams continue to innovate towards the proof at home in real time, and so that the main ZKVMs become future basic infrastructure for Ethereum.
