The study you linked doesn’t have a control group, which is a fairly fundamental oversight. In other words, it can’t answer the question of whether all negative political content is down ranked or just Chinese political content.