The thread How is Farming Centralization Disincentivized seems like the best discussion of this topic. I will ensure a link is included in the ‘Mining Centralization’ section.
I still think large vaults (like large bitcoin miners) are inevitable. The economic logic I understand is that farming 20% above average is most rational, therefore a trend toward larger nodes is always more desirable (from a farming perspective).
The only way for small farmers to manipulate this is to start lots of near-zero-resource vaults to bring the average down, but because they sit low on the sigmoid curve these farmers will earn very little. So relying on this to happen is like assuming people will always be willing to run full bitcoin nodes (which is probably true, but is economically irrational given the rules of the network).
I’d like to be shown this logic is wrong, but currently I can only see economic motives for larger vault sizes over time, not smaller or ‘balanced’ vault sizes.
One important point I will make sure to include (@Seneca makes this point in the thread linked above): “What the algorithms will do is prevent centralisation in XOR space.” I will definitely add that to the ‘Mining Centralization’ section.