Decision makers often desire the subjective predictions of experts. Any
mechanism for eliciting and aggregating such assessments ideally should
be: (1) incentive-compatible; (2) information-weighted (i.e., individual
assessments should be weighted according to the a priori unknown amounts of
private information held by agents); and (3) self-verifying (i.e., the mechanism should
not require any ground truth observations). Although subsets of these properties
are attainable by previous elicitation methods, our mechanism of collective 
revelation is the first to simultaneously achieve all three attributes. 
Moreover, we derive budget-balanced and online variants of the mechanism.