Thanks @kfx. This is a thoughtful analysis and you’re not wrong about the potential for v2-like pools on v4 enabled by hooks. And I also agree that it would have been useful to have v2 everywhere v3’s been deployed since we started branching off of mainnet. C’est la vie.
In my view the liquidity fragmentation is a problem but one that’s solved (at least at first) at the router level. Eventually of course we’d love to get all Uniswap liquidity for any given chain into the Singleton so that we can realize the gas optimizations it can provide. But a) I don’t think that’s necessary to plan for on day 0 of launch and b) i think there are a bunch of interesting ways we could incentivize that transition (e.g. some implementation of the credit facility @mjc716 describes here).
Infrastructure-wise, there will need to be a new universal router deployed on the chains with new v2 deployments. Integrators that talk to the current Universal Router will need to note this change and update the address they call. Any apps that interact with the protocol directly will need to update their liquidity provision UIs.
I think the benefits (increased user safety, potential market share gains) outweigh the valid cons you point out.