My thoughts are:

Was the 1 MB cap introduced as an anti spam measure when everybody used the same satoshi node, and did that version simply stuff all mempool transactions into the block in one go?

Big mining farms are probably not using reference nodes, since they probably wouldn't be able to pick transactions where they have been priotised using a transaction accelerator.

Increasing the block size cap in the simplest manner would avoid BU technical debt, as the emergent consensus mechanism probably wouldn't work very well if people do not configure their nodes (it would hit a 16MB cap in a more complicated manner.)

Miners have to way up the benefits of the higher processing costs required to build a bigger block versus the orphan risk associated with the delay caused by it. In other words, a more natural fee market develops.

So it won't be massive blocks by midnight.

Any comments? (probably a silly question  )