There's an easy mechanic to understand about recommendation machines (which let's face it are the main reason for so-called echo-chambers to exist [1]) it's dec

There's an easy mechanic to understand about recommendation machines (which let's face it are the main reason for so-called echo-chambers to exist [1]) it's decontextualization. Any good scientific work has a introductory section explaining the stakes of the field, a section about related works, an auto-critic section about shortcomings and possible future improvements, some declaration of interest. It might seem artificial but it is key for turning bare information into useful and digestable information. It explains how, why and by whom this blob of content has been made. If we actually took down facebook and co and promoted content using a liberal network of peer-reviewing (in bazar-style of wikipedia, not in cathedral-style of big academia pubs) then i'm pretty positive a big chunk of information and institutional mistrust would go away. But then again really putting the true power of contextualized information in the hand of everyone is not something control-structures like states or multinationals have any interest in doing. [1] "fake-news", deceitful information, pseudo-science and fantasy-conspiracies have always existed, what's new about them is that they're getting so big. Yet they're not getting big out of nowhere, they spread on plateforms that (1) promote content based on past performance (which is not a curation method) and (2) ease low-context sharing like reposts, likes etc. They are also aggravated by goverments and entitled newsoutlets being pedantic and dogmatic, by fact-checkers that act like judges instead of investigators.