This seems discontinuous with the Center for Election Science's earlier push for approval voting. Approval voting maximizes the consent of the governed; this, while it may arguably do a better job at evaluating candidates, minimizes the consent of the governed, as most constituents would not have the opportunity to consent to any candidate. It's an interesting idea, but I would be very concerned about the outcome being perceived as illegitimate by the general population.

Hell yes. Also just citizens assemblies in general, whether for appointing officials or directly deliberating on issues. Turns out randomly selected citizens actually perform far better than elected officials.