Yes, that's a potential issue, but I think it would be an edge case in practice.
We know there ain't no such thing as a free lunch, which is what RC has been all about. So given that, where do we go from here?
In the method I am exploring above, I think the vast majority of transactions in the near term would occur at zero cost on an excess capacity basis. These transaction costs would only emerge as a necessary element when there are enough users to fill the blocks beyond what is subjectively reasonable.
Just for arbitrary numbers, let's say that's 1 million active users. At that point, the witnesses would have a choice to make between expanding block size to keep a reasonable number of spots free for the small fry or to limit new user growth. Or maybe some combination of the two.
We already pay a voluntary tax to subsidize low-SP users. That's what the witness earnings are. So that might also be an adjustment for when blocks become full.
It's not the fullness of a block that they're worried about, as I understand it, but the total size of the historic blockchain. (Which is where most of the literal resources go, why witnesses need substantial and expensive servers, nodes are even worse, and replays take forever.)
Hmm. That would imply that keeping the chain in RAM is the constriction. On a text-based chain like STEEM with a growing user base that creates a square-law geometric progression.
If that's the case then it's an absurd design.
I think I'd need to talk to aggroed to see if that is the constraint.