It's not nodes that need to be incentivized, that's really easy to do. It's also not latency we're looking for either. What we really want is a way to decentrally, and trustlessly, prove that nodes are geographically distant (as a proxy for node anarchy), and incentivize it. It's pretty easy to prove that any two nodes are closer than or equal to a certain distance, but I don't see how to do any better than that.
We could have 3 trusted centers, which would triangulate a node and mining rewards could then be adjusted based on how geographically separated they are, but that's neither trustless nor decentralized.

Like I've said before that seems like a paradox in Bitcoin. We want Work that is nontrivial, to prevent attacks, while also wanting nodes to be trivial to start and operate, to prevent other sorts of attacks. Seems to me as long as mining is profitable it will always struggle toward centralization, and Work minimization, which is always economically optimal. The only way for Bitcoin to remain secure is by being economically unoptimal, which would require charitable node operation, which has become more and more unfeasible as operation has gotten more expensive and tedious.

I expect we'll see even more centralization when we hit the next reward halving, which isn't exactly prescient, that's been the norm. Unless someone can find a clever way to incentivize node distance (node anarchy), which seems like a pretty hard problem, or we start pouring wine on the ground for the Bitcoin gods, we're going to see the slow death of Bitcoin as a decentralized trustless currency, though it may survive as some other sort of abomination.