On the 'vouch' option, this seems helpful. However, at some point one just accidentally creates a tier of "Super Upvotes" and "Super Downvotes". Trying to establish a reasonable heuristic/estimate/algorithm/whatever to identify who is or not a 'trustworthy user' worthy of the "vouch" permission might produce an interesting paper or blog post, but I doubt it'd help to increase transparency in real life, especially if there is a need for judgement-calls.