Proof of Work scaling DOS for new candidate blocks

[Craig-Macomber]

I'm not sure if this is the right place to bring this up, but in README.md:

The efficiency of the algorithm declines with the number of blocks the network has to choose from in each round. We may use Proof of Work, with no reward and difficulty retarget as a throttle rate on the introduction of new candidate blocks.

I might have misunderstood the explanation, but I think the proposed solution here seems to open a hole for DOS of the entire network by someone willing to do a lot of work. Since the only reward for this work is control of the network, it seems like fair players won't be willing to do much compared to attackers. One adversary could spam candidate blocks from one or many nodes and drive the work higher than anyone wants to bother with, and then control what gets accepted into blocks by minting all the new candidate blocks.

I realize this attack isn't too harmful (can only delay transactions) and is trivial to detect, but I think designs that prevent it are practical.

A proof of stake based mechanic (limit candidate blocks per skycoin per unit time, or rank blocks based on a related value) might work. I suppose just using the proof of work to knock out blocks that are redundant with ones with more work might make sense though.