Make vote power loss dependent on the relationship of SP of the voter to the author. If they have the same SP it costs normal. If you're voting on a minnow with much less SP it should cost a whale much less of their voting pool to upvote their content. That changes the incentive to upvote minnows.

This really activated my almonds. I think this would be an effective adjustment. Good write up.

Also, concur on not mixing more than one major variable at a time. Easier to assess changes, UNLESS a solid argument could be made that the efficacy of the two changes are interrelated/dependent on each other for an optimal outcome hypothesis.