Social site Facebook has announced that it is going to start cracking down on posts from individuals and pages that are meant to grow influence by baiting users into engaging with them, and it's using a machine learning model to do so. The posts in question are the ones that ask users to put in a certain reaction as a form of voting, share something, tag friends, or otherwise engage with the post when they otherwise would not have. The company will be actively scanning for such posts using its purpose-built machine learning model and will be demoting them, negating the reach advantage that they gain through engagement baiting which often comes down to begging for likes and shares.
To be specific, the model here was trained on five different categories of engagement baiting across a large swath of posts in a teaching sample. It learned to identify posts that are trying to get users to vote, react, share, tag friends, and comment. All of these forms of engagement can help a post get promoted, netting a wider reach and appearing higher on users' news feeds. The model has also been trained to recognize posts that are legitimately trying to circulate something and need additional reach, such as posts about missing children, fundraisers, or posts soliciting advice. Such posts will not be affected by this wave of enforcement, Facebook said, without elaborating on the manner. Pages that continually use engagement baiting will be demoted on a page-wide level over time, meaning that the more a page uses engagement baiting, the lower its overall post ranking will be, even for posts that don't use such tactics.
This is another front in Facebook's battle to legitimize the platform, making it a place for users to interact and share the things they truly care about. This comes on the heels of a fake news epidemic that saw the American government and prominent figures calling on Facebook and others to work harder to identify and take down misleading, fabricated, and radical news stories. With the fake news problem being diligently worked on within Facebook at this point, moving on to entities that pursue growth inorganically by seeking to undermine and skirt around how Facebook's algorithms work seems to be a logical next step.