If anything, we should get the blocksize to the largest rationally supportable size
This is exactly where reasonable people differ.
That is a key word right there "reasonable". I certainly haven't had any great epiphany as to how to choose the correct size or how to make it a reliably updated thing. Predictability and anything that removes human consensus making from an ongoing process is what I would favor.
The only way to remove human consensus from the ongoing process is to leave it exactly the way it is. No changes to the consensus rules ever (MP argument).
That is btw pretty much Wladimir's view. He's not going to back any change that doesn't have human consensus. So you either have human consensus, or removing it, no changes.
I suspect Wladimir was thinking in context of a hardfork vs. context of ongoing blocksize increases chosen by an algorithm. I might not have seen his specific commentary though so I could be wrong. It is hard to keep up with all the goings on nowadays.
I can certainly see how a complete removal of the hardcoded block size limit would be an attractive option. There are several large systems that have unrestricted scale but are controlled by each individual actor according to their capability. Bitcoin is much more likely to end up like the big I Internet in this respect than something that everyone can run on their raspberry pi.
With respect to "no changes to consensus rules ever"; that way lies madness
. I don't see a solution where Bitcoin retains value as a monetary unit or a ledger of record without a blocksize increase. Lightning and Sidechains are great but they need a strong ledger of record to support them. Strength requires broad adoption. Broad adoption requires high TPS. With Sidechains I suppose you could argue that everyone will just transition out of Bitcoin into something else but that probably won't fly with those bought into the existing ecosystem.