This proposal aims to discuss how social media platforms and the understanding of how algorithms work in different digital environments impact the work of science communicators who develop online scicomm projects. The hypothesis assumes that scientific knowledge alone is no longer sufficient for the effectiveness and efficiency of science and health communication on the internet, since the information overload and the effects of the infodemic during the covid-19 pandemic indicated the need not only to talk about such subjects with a scientific basis, but also to know how to stand out in a vast universe of online content – which includes misinformation, disinformation, and conspiracy theories. Based on examples collected from Twitter profiles of Brazilian science communicators, we will discuss how sharing the perception of the influence of algorithms and platform policies impacts the way these science communicators talk about science and health, and how they manage to share perceptions on how their content is edited and mediated by algorithms and platforms. In addition to Twitter, posts from Instagram and YouTube videos are also empiric examples that help us identify similarities and patterns of content that are consolidated as more interesting for the logic of the algorithms. We question whether the process of adapting content to the logics of algorithmic mediation, although contributing to a greater reach of science communication, can transform the communication of health and science into a clash between form and content, and to what extent algorithmic mediation homogenizes the content and has implications for the type of science communication that is built to pleasing machines in order to inform the public.