In the long term future, Rollup may be the best way for Ethereum to achieve global scaling. Except for specific alt-L1s, most alt-L1s will (probably) become a Rollup on Ethereum.
The most essential difference between Rollup and alt-L1 is that Rollup can directly use Ethereum as its own security layer and DA layer, and directly inherit all applications and users in the Ethereum ecosystem.
In turn, Rollup can bring benefits (more users) to the Ethereum ecosystem.
Among all L1s (except Bitcoin), Ethereum is undoubtedly the blockchain with the largest number of users, the longest-running time, and the most decentralized and secure blockchain.
Inheriting the security of Ethereum, Rollup can focus on the execution layer and does not need to design a complex token mechanism (on the contrary, Alt-L1 needs to design its own security layer).
The end is that Rollup (relative to Alt-L1) has more freedom and more room for innovation, and because it inherits the highest talent density in the Ethereum community, Rollup will become a test field for all innovations.
Many important innovations and breakthroughs will occur on Rollup, I don't think we can imagine the type of applications that will eventually be built on Rollup.
The end of Alt-L1 is likely to be that most of Alt-L1 will disappear or become the rollup of Ethereum. Except for L1s such as arweave (storage) or ceramic (data warehouse) that solve specific needs, faster and cheaper "Ethereum" (Solana, BSC) is not competitive in front of rollup-network.
Of course, this is just one of countless futures (the most likely one). I'm always willing to change my opinion as long as you can give me enough evidence.
It is not enough to achieve global scaling of Ethereum through three Rollups (starkware, zksync and scroll).
Starkware and zksync have the first-mover advantage, but Scroll's zkEVM has the strongest legitimacy, and Scroll is the most open and the easiest for community members to participate in.
Vitalik mentioned in an article the power of this legitimacy.
So I think there may be 100+ various Rollups in the future. There are general-purpose rollups like Scroll and zkysnc, and there will be application-specific rollups like dydx and loopring.
If I want to build an app chain today, I would definitely choose Rollup instead of Tendermint (polynya wrote a great thread on this topic, highly recommend reading it).
So I'm excited about "application-specific recursive zk rollups".
For application-specific rollups, Scroll and zksync can develop an L3 SDK to facilitate other developers to create and deploy a rollup (not just smart contracts) on L2 network.
For recursive zk-proof, we can generate zk-proof from L3 to L2, or outsource it to a third party. As for proof outsourcing, Scroll's hardware-accelerated algorithm has the potential to open a new market.
It's still very early, after L2's zk-rollup is mature enough, there will also be liquidity aggregation and composability solutions between rollups. Two very interesting proposals I see are "dAMM" and "Fractal scaling".
I believe it is definitely easier to do composability between Rollups than between Alt-L1 (since Rollups share the Ethereum DA layer).
The Rollup dilemma we propose today will eventually be resolved, and Rollup is a better choice than the Alt-L1 solution. Now is the best time to build Rollup and scaling Ethereum, Welcome to join the team :D.