Interesting thought, and I'd like to hear some of your ideas. The one idea I can think of would be to have academic papers upvoted/downvoted by the profession much like steemit posts are. You'd have big name academics as the "whales" and everyone else with voting power according to their academic reputation, which hopefully would correlate with their steem voting power. Papers of high quality and of value to society would be recognized by the amount of steem and upvotes rewarded.
This process would be more democratic and transparent than the current academic journal caste system, and therefore more robust to highlighting the most important research.