At some point there will be another block size war. Probably when the cost of a single onchain transaction approaches the cost of running a full node for a year. Who knows which way it will go.
Discussion
Agree. We'll have to revisit that debate eventually, unless somebody much smarter than me comes up with a more clever and elegant solution to the problem. Arbitrary block size, big or small, has problems one way or another. Dynamic block size, with certain constraints, might be a solution.
I still think L2s are the way though. The more we grow lightning and add channels the better probably.
All I know for certain is custodial solutions are not the way.