The point is to collate publically available information into an accessible single location that will allow easy discovery of hidden agendas and conflicts of interest. You don’t have to trust the site for anything, it’s simply a jumping off point.

You write: "because of how easy it is for any shady organization to set up a defamatory site masquerading as trustworthy and impartial." I can see emotions, confirmation bias and backfire effect playing a role by, as you note, "a defamatory site". I'm curious if there are options or solutions to counter this?

MIT has a cool project looking at how individual terms are covered though: https://www.media.mit.edu/projects/media-cloud/overview/

The relationships when modeling from the author/publication get really big really quick. I suspect as NLP gets better it'll get easier to model though.