To close that gap entirely, the expectation of rewards at the time of voting should be based on an algorithm which is not computable or estimatable/predictable by a bot or in which all choices produce the same estimate.
If the said estimation and the maximization of it also motivate the human voter to interact online, then that other desirable goal of maximum real human usage is simultaneously met.
Afaik, the uncertainty in the curation reward marginally adheres to the former requirement but it is difficult for users to estimate and doesn’t (maximally) motivate the said goal. Steem’s rewards are much too complex (and much more complex than they need to be) which is one of the reasons why users leave.
RE: Proposing Steem Equality 0.19.0 as the Next Fork