Ethereum

if ethereum is planning to be rollup-centric, there is no need to increase the gas limit.

we should, in fact, be targeting to lower it once L2s get permissionless block proposals

I'm good at making mistakes. Sometimes I improve, but I never make the same mistake twice.
https://makemake.site

Hard disagree. Increasing the gas limit won't solve our scaling problems, as we may have only another 10X (before SNARKification), but it sure does help a lot of things. Some operations must happen on L1, like validator management. It's also a (far from reached) bottleneck for rollups, that must submit proofs (ZK). The cost of reasonably increasing it, today, is near zero.  It's going up, not down. Moreover, at some point the gas limit may be gone completely. Do as much operations as you want, as long as: (1) (reasonably small) proof of validity (2) state-diff not too big. (this specific design has a bonus in the form of unexpected privacy)