>>6396
I would guess singletons could simulate societies to get their benefits, making them strictly superior.
But possibly real societies are more likely to have beneficial value drift than a singleton carefully controlling a simulated society (depending on the original value system). Or societies are more fit, but exploring to find out if they are is too costly to have a positive prior expected value for a singleton. Or the overhead of the singleton makes simulations underperform compared to the real thing.