Theoretical comment spam like you described could be fought with automation pretty effectively imo. Once a farm is detected with the help of a relational database and some sophisticated analysis algorithm, the farm is done, and the cost to set up a large number of accounts puts a huge risk on starting an effort like that. Even with the relatively low account creation cost right now.
Imo shooting all small users because of that possible scenario is the wrong approach.
Also, the scenario you described doesn't need a threshold of an equivalent of currently ~3$
I made a small test. My 13k SP vote adds 6c to a 0$ post, but 24c to one with 30$. Raising the rewards for the big one in return lowers the rewards of the small ones. I will not like this ever. Going to linear was good.
RE: The EIP and how curation will matter again - a.k.a false hopes