Credibility scores are critical. For platform usage and accountability; for publisher accountability; as training data for machine learning; etc.
Without credibility scores, how can we even measure progress on misinformation goals? The crucial property for these scores is that they are meaningful and defensible — that they can be trusted.
In an ideal world, credibility scores would be citizen lead — but we live in a world of trolls and manipulators with vast resources. So I vote strongly for credibility scores (it is the focus of my work), but without the citizen lead component at least initially. In the long run, I think we can build complex checks into rating systems to avoid manipulation and have true citizen-led scoring—but we can’t wait for “the long run” to get meaningful credibility scores.
(I’ve written more on this here: https://medium.com/@aviv/how-to-fix-a-news-ecosystem-4bf623f91d81)