


Original article by Luyao
Gnosis 的创始人 Martin Koppelmann 3 月 4 日在 EthDenver 做了个分享,讲了 L2 的局限性,以及另一种扩容的方法。我觉得非常有趣,我做了一些摘录和笔记:
The original purpose of L2 was to complete a bunch of transactions in bulk and then synchronize the results back to L1. It's a very temporary space, rather than a new space for permanent assets.
In the long run, L2 will still have some fundamental problems even if it is perfectly implemented. For example:
One problem with this process is that it is only suitable for applications that do not have state inflation, such as trading platforms (which only need transaction results, not transaction history), but it is not scalable for applications with state inflation.
If 10 percent of the world's 8 billion population (800 million people) were to sign up for ENS, the entire transaction processing capacity of Ethereum would be used to process these requests, which would take two years to complete. During those two years, Ethereum will not be able to process any other transactions.
If all the stocks in the world (45,000) use Ethereum as a clearing layer, there will be less than 30 trades per stock per day (even if you use L2).
Transaction cost L2 peak gas sometimes reaches more than $1. Even if EIP4844 is implemented, gas can be reduced by 90%, but there are still two problems:
1. It still doesn't work for scenarios that require gas of less than 1 cent.
2. gas will continue to rise even as demand increases.
L2 Asset exit Problem:
1. Small assets may not be enough to cover exit gas fees,
2. The exit bandwidth is limited, if everyone wants to go out, it will be blocked......
Some apps cannot be rolled up.
CirclesUBI, POAP, for example, create a lot of states that can't be compressed, so L2 is useless for them.
How about staying in L2 forever, never leaving, and using it as a permanent space?
The sequencer of L2 is very central.
They can't take your money, but they have enormous power to decide whether to accept your deal, how much gas to charge for your deal, who is in front and who is behind...
I also hacked coinbase, saying that if you're going to be a trading platform on BASE, it can easily rank trades from it above yours...
Then, the centralized sequencer will most likely be censored or even forced to KYC (accept only transactions from KYC addresses).
The authors stress that this is very possible from the current regulatory perspective.
Here the author raises a very interesting question:
What's the point of having L2 if we're going to release an asset that's native to L2 and doesn't exist in L1?
Because L2 is safe because L1, but if you don't want L1, why do you need L2?
The problem of rigidity.
Ethereum L1 itself is still evolving and will undergo many changes in the next 5-10 years, which will present L2 with many challenges!
Example: The snapshots we use to vote are voted in L2 and then synchronized back to L1. They make Merkle Proof in L2.
However, Ethereum L1 is scheduled to switch from Merkle Trees to Verkle Trees in the next year or two, which will make the current version of Snapshot unusable.
Therefore, L2 may require some kind of "upgrade" mechanism, but this would also contradict its trustless goals.
The authors propose an interesting scheme similar to the Cosmos IBC model.
That is: Let's make another chain, run the same thing as Ethereum, and connect it over untrusted ZK-bridge to form the Ethereumverse.
In fact, the way the zk bridge was implemented was very sci-fi when I first heard about it, and now it's actually being built.
Is to use zk to run a light node client on one chain on another chain and then verify in it... It's trust-free and far safer than a traditional bridge.
My Personal opinion: This is a very interesting, underrated solution, and if you think about it, it really solves a lot of these problems.
Original link