Reading the accounts from yourself and Mike Wacker, I could ask many questions about Google’s internal culture, HR policies etc. But I’ll stick to the development of Google’s algorithms for now. You state that it’s not possible for political bias to creep into their development, but then link to a set of guidelines where judgement is required by the rater. So any potential political bias is simply postponed to a later stage in the process i.e. the rating rather than development. In the section you specifically link to (section 7.3) the definition of what constitutes hate or violence is so broad that it could include pretty much anything. Including content that is “hateful-sounding” is far too abstract. Leaving these key decisions up to the rater is “precisely” where political bias comes into play. What you and I and anybody else considers hateful is entirely subjective. If the person who was made to feel unsafe by Jordan Peterson’s criticisms of gender pronouns (see Mike Wacker’s article) happended to be a rater, then clearly there is a reasoned question to be raised as to whether Google does indeed have a problem with serving politically neutral content.