I think one thing that might improve factchecking is moving away from the data product model where every check has to have some kind of subjective ranking. When you ask a fact-obsessed reporter to go through a political speech with a fine-tooth comb it's no surprise when they have notes for every line! If you have to justify every note with a label you'd probably have to go with something like "misleading". But that does a poor job capturing the overall truthfulness of a speech or event.