It's not obvious what fixes might work here without actually testing things. Sometimes ideas sound good, but contain some only later obvious flaw that dooms them. And online we see immense web business that initially must have seemed absurd, and were never expected to loom large or last longer than a fad usually would.
In which spirit, some unorthodox experimental notions:
How about an experimental gutters section, where first-timers can post, and vote as much as they please for their own posts, (but not for the posts of others); an added enhancement to prevent gaming would be to limit downvoting, maybe a new user could downvote their own post only once. The theory to test would be that low-vote posts might be better quality.
There could be awards and badges, like "Million time voter", or "Stuffer" for those who qualify.
The opposite method would be an experimental section hades where users are permitted infinite downvotes of others posts. (But only one upvote.) Logically it's much the same as the gutters, but the psychology might work out differently. Here the rate of retaliation might be the most useful metric. Badges and awards might be named "Vendetta", "Joffrey", et al.