Well written @dantheman. It will be very interesting to see how this plays out.
Is the idea of 21 block producers such that they'd be large VM farms, similar to EC2 or Azure, over time? Naturally, starting out, they'd be smaller. I'm guessing a smaller number of block producers is easier on the governance layer and overall governance capacity.
Also, regarding the ability for block producers to accelerate a "patch" to the network, is there any planned governance to this process? It's possible that this could become a loophole by which block producers pull a "kill net-neutrality" event.
that would defeat the purpose of distributing consensus, increasing reliability, and minimizing the opportunity for collusion among witnesses / block producers. It will be similar to the STEEM model, with block producers global distributed in different colocation facilities, managed by different providers (ie. multiple ISPs / cloud services), around the world in multiple countries, each managed by independent (and unrelated) parties.