I have long thought that implementing something inspired by a 2nd price auction for rewards in conjunction with a superlinear rewards curve (to discourage stake-splitting) would go a long way towards solving the valuation problem.

Unlike today's system where a voter can maximize curation rewards by voting at maximum strength, an algorithm inspired by a 2nd price auction could drive high-powered voters to self-limit their own voting strength to accurately match the perceived value that they place on a post. Otherwise, they'd wind up wasting their voting power.

Among other places, I wrote about that in Simulating a Steem curation rewards distribution that is modeled after a 2nd price auction