It seems Bitcoin community can't reach consensus on this issue, a lot of people against the proposal.
There seem to be a few people who want some OTHER algorithm for increasing the block size, but I'm hearing very broad consensus that we do need a hard fork to increase the size.
That is progress.
I'll be running some experiments using actual blockchain data to see where the current code breaks; if it can already handle 20MB blocks then I'll work through the details to propose a hard fork (write BIPs, write code, write tests, ...).