4MB blocks going towards 2020 seems like it's leaning towards protecting decentralization vs growth at all costs. And it's a number I bet most BIP101 fans would find absurdly small. As polarized as this debate has been, I think the winning proposal should be the one that equalizes displeasure between the XT-Unlimited and 1MB4EVA camps. Doubling max block size in tandem with halvings mimics and compliments the simplicity and predictability of the reward schedule.
This dose nothing to alleviate the genuine concerns of Luke Jr, and other Bitcoin Engineers that have good evidence and reason to believe that the 750kb blocks that we have now are already hitting the block size scaling limits that the network can safely manage.
People who propose such large blocks, such as 2mb or 4mb, never seem to address the concerns that the Bitcoin Network is *already* struggling under present loads.