Excellent post, Scott. I remember talking with you about some of these differences a while back, and you've laid it all out quite nicely here as well.
I've been thinking for quite some time that I should get a full API node up and running and this has me thinking all the more how important that is for top witnesses. I look forward to hearing more about your build when you release details. Hopefully we'll continue to see improvements to how Steem manages data so we won't need expensive, experimental servers or days of replay time to run full nodes.
'Bout time I get some thoughts to paper, eh? :)
Indeed, the "part 3" of this series will really just be me geeking out about building the ultimate STEEM server -- hopefully my experience can help inform other witnesses and we can push for a more decentralized data environment!
Re- how steem manages data, I totally agree. It's something I want to dig in to soon, since it seems like it's currently a second class citizen as it currently "works". We can do much, much better.
It may relate to some of the things you mention in your post, but what do you think of this EOS: An Architectural, Performance, and Economic Analysis? I wonder how much if it relates to Steem as well. I've been asked to take a look and give a perspective and figured you might be interested as well.
Saw that this morning, first impression is that it has an extraordinary amount of inaccuracies.
I might do a peer review to explain how misleading their statements are.