These videos dangle between what YouTube’s terms of service find acceptable and in violation.
This article states that YouTube has seen a 70% drop in watch time on “borderline” videos by non-subscribers following the alteration of its recommendation algorithm to reduce such content. YouTube’s has also outlined the specific criteria external moderators use to determine if a flagged video is borderline.
This information is used to inform machine learning tools regulating the platform. Further, every evaluated video obtains about nine opinions, of which the critical areas need the validation of certified experts.
YouTube says it is focused on giving diverse voices a place on its platform while protecting viewers and creators from “harmful content”. It is also surfacing authoritative, reliable and accurate sources for topics around news, science and historical events.
[4 minute read]