I don't think it is possible to do actual research on it, it is always a guess (as to node loss).
So besides research on node utilization costs/attacks and block propagation costs/attacks,
the major problem is: no one really can do research on what a post-hardforked blocksize
node count will be. It could be 4500 or it could be 5. There is no way to study it legitimately,
so its outright just a blind risk. I don't like the idea of jumping off a cliff blind and not knowing
if the cliff is deadly or just a simple street curb.

I was thinking today, if the BU and other developers figured out a mechanism to "incentivize nodes"
in the same manner that Miners are incentivized to verify Bitcoin txs/blocks, then not only would they
provide one of the greatest services to the Bitcoin community, but also allow a more secured future for
forked upgrades, since the node system can grow and not diminish.

If I had the know how, that some of the members of this forum have, that is what I would be working on.
Miners don't work for nothing, but node maintainers do. If this puzzle could be legitimately resolved, it would
be a great success for Bitcoin, even more so than a 2 to 8mb blocksize increase.