| So, Facebook has a vast cache of data on what people do and don't like and respond to in their feed. Facebook also has any number of 'features' in ML language they can slice and dice this data by, from likes and shares through to time spent on article before returning to the main feed. Browsers of the Facebook feed themselves have any number of ways of accessing information, which they did not have before. If I want sports news on teams from Chicago, I'm no longer beholden to Chicago Tribune or even Facebook for that, I can download something like Bleacher Report TeamStream and have a customized feed of what I find interesting and nothing else. It's probably not massively difficult to work out the kind of content that makes people like, share and click at scale on Facebook - humans are humans after all - but it's probably not consistent with the output of the Chicago Tribune (national stories that are covered in greater detail elsewhere and content which is probably only of any real interest to Chicagoans). This is probably even more the case at the moment with so much of the action politically happening at a national level - Trump is a once in a generation kind of news story. It seems they have two options: - Accept that Facebook is a pay-to-play platform and that they will have to pay to narrowcast news to Chicagoans - Accept that Facebook is going to account for an ever-declining share of their traffic. Either way, I'm not sure there is a way forward for online media that doesn't involve a huge rethink about what the product they serve up is. Blaming the algorithm seems incredibly fruitless. |