I’m not sure if you’re talking about optimality as if it is something attainable. It plainly isn’t. It may or may not exist but, frankly, that’s irrelevant as long as we can never learn what it is, and we can’t, not even in much simpler domains. So, just like in the markets, where the independent actors together try to find the “correct” price but never quite reach it (maybe this is what you meant?), we’ll need to depend on simple but robust heuristics. What we definitely shouldn’t seek is something like the Starcraft bot that nobody, not even its makers, can reason about.
The difference between optimality and “works fine” is that of equations an inequalities, where the former are about finding the precise (“optimal”) answers while the latter are about finding the ranges within which we are okay. While the former are “neater” in theory, the latter are more useful in practice, and even more so when we’re dealing with uncertainty or complexity.
Based on your “platonic optimum” paragraph, you seem to be talking about this but call it optimality. I’m not quite sure, but please no Starcraft.
It’s fine to apply things at any levels of abstractions as long as we have guarantees for correctness (not optimality) at all of those levels.
However, if we can’t guarantee (not wish or hope for) that the system will stay within healthy operational parameters (we won’t run out of either coins or space), that’s a sign that we don’t have the right rules at the level of abstraction that’s responsible to deal with those parameters.
I’ve written about ideas concerning that here and there but I don’t have anything specific at the moment, sorry.
I would prefer a free market where vaults would compete for content while seeking to maximize their earnings while also protecting the network from filling up. However, that’s not how chunks are assigned to vaults and I have no good ideas about how to go around that. Unless…
Maybe if sections assigned chunks to the N lowest bidding vaults out of M that are considered? Greedy vaults would remain empty so they would be incentivized to ask a fair price, vaults that are getting full could stop receiving chunks until the others caught up, and users unwilling to pay up would not get their stuff stored.
EDIT: Clever vaults, if they couldn’t grow, would set their prices so as to always keep some free storage around for times when the rest of the vaults are also getting full, lest they miss out when the price got higher. It’s similar to airlines that ask more and more as they have fewer and fewer free seats left.
The strength of this model is that it needs only two components:
- Sections need to keep track of the bids of a pool of vaults.
- Vaults need to have a method to set the price.
At this point, we have a free market and the prices will be set accordingly. Please note that the chunks will get stored at the lowest possible price and the network will be protected from getting full as long as most vaults are working according to their own interest and the precise method they use to set their prices is not very important (people will try to come up with better and better algos for that).
EDIT2: I think it’s just PAC without all the complications. I think I’ll just copy/paste it there. I’m sorry but I’m quite busy recently and I missed out on much. Though I even commented on that thread a few days ago 
We’re on the same page, I’m talking about PUT price and safecoin. @oetyng suggested, or so it seems, incorporating external information into the price and I was referring to that.