I do support modest superlinear (~n^1.3ish, but capped after a certain point, lets say after 1,000,000 sp worth of votes, then linear afterwards) in conjunction with higher curation (~50%) and increased downvote incentives

I think with n^1.3 you probably get most of the benefits of n^2 (wisdom of the crowds, forcing all 'profitable' behavior into the light, making it difficult to put a price on vote buying, higher curation incentives, gets rid of all 'profitable' spamming etc.) at a fraction of the cost (whales less overpowered, minnows less underpowered)

Basically at n^2, someone with 10x your voting power has a vote value that is 100x that of yours. In other words, they get 10x more voting power PER sp.

Under n^1.3 someone with 10x your voting power only has a vote value that's 20x that of yours. So their voting power per sp is only 2x that of someone with 10x fewer sp. It gives minnows a fairer chance and should be significant enough to enjoy most of the benefits of n^2 outlined above, especially paired with higher curation and more downvote incentives.

Basically it all comes down to trying to get profit maximization behavior to shift from what it is now to actually voting for good content (at least subjectively), but at the least cost in terms of trade offs.

So sure, you can do this by making curation 100%, or with a curve that's n^10 (extreme examples to illustrate this point), but the trade offs are too high. The idea is to shift things around just enough to move the economic equilibrium away from brainless vote selling/self voting but leave as much on the table for content creators and minnows' voting power as possible so they have a worthwhile time being here.