Vitalik in his latest post argues that Ethereum should stop treating L2s as canonical “scaling shards” and instead view them as diverse, partially attached chains, because L2s are slow to reach full trustless interop while L1 itself is rapidly scaling and can natively verify rollups.