Dropping ancient history is definitely a positive step, though it comes with downsides like a range of applications ceasing to work in a decentralized way because they depend on ancient historic events (I mostly blame the apps for this design, not the clients for dropping receipt history).
I am of the opinion that we are already way over reasonable state requirements for end-users, and small gains like dropping pre-merge history just help mitigate some of the damage we have already done. Part of the problem here is that there is no consensus on what the target end-user machine looks like, and there isn’t even consensus (sadly) that end-users should be able to run an Ethereum client at all.
The first step to making informed decisions on this topic is to come to an agreement (among who?) about what the target demographic is for running trustless RPC clients. Once we have that we can have more interesting discussions around whether a given change will cause us to exceed that target or not.