Vitalik evaluates L2: L1 and L2 should continue to develop synchronously in the short term, and the future market will determine the scaling method.

BlockBeats News, on February 19, Vitalik, co-founder of ETH Fang, replied to the question of "how to evaluate Layer 2, what is good or different from what was originally expected, and whether rollups are generally good or blood-sucking for ETH Fang" in the text AMA held on the Tako platform, saying: Until now, the expansion method of ETH Fang can be roughly understood as hybrid L1 + L2, But I don't think anyone has clearly defined enough which transactions should be at L1 and which should be at L2. The answer "everything is on L2" is more difficult to accept because: · In this way, it is easy to lose the "medium of exchange" and "store of value" of ETH. If you're worried that L2 steals L1's users and doesn't give L1 anything back, this problem is even worse in a "L1 does almost nothing" situation. L1 is still required for operations across L2. If there is a problem with one L2, the user still needs to have a way to move to another L2 by themselves. So there are some use cases where L1 is hard to avoid. The answer "everything is on L1" is also more difficult to accept because: · If L1 supports a lot of transactions, it is easy to become centralized, i.e., using technologies such as ZK-EVM. The world's demand for on-chain transactions is unlimited, no matter how high the TPS of L1 is, it can always be found that an application needs 10 times more TPS (e.g., artificial intelligence, micropayment, microprediction market, etc.) · L2 doesn't just scale, L2 also provides faster confirmation through "preconfirmations" and avoids MEV issues through "sequencers". So we need hybrid L1 + L2. I think the role of L2 will continue to change, for example, now it seems that EVM compatibility with L2 is enough, it's possible that we'll see more privacy-focused L2s (aztec, intmax, etc.), and maybe more application-specific L2s (if an app wants to control its MEV situation, there's a benefit here, etc.). So in the short term, I think we should continue to improve the capabilities of L1 at the same time, add blobs to give L2 more space, drive interoperability across L2, and then the market itself will decide which scaling is right for which application.