Academic publishing is a complete farce, both for textbooks and the papers. The experts who write these books don't really do it for the money, they do for the prestige of furthering their field.
And I actually do think free content is the way things will go. If you look at machine learning, almost all the content is available on
https://arxiv.org/, and I am meaning real cutting edge stuff. And most of the good papers in this field have their code published on github, such that other people can use it. Which actually makes it more scientific as anyone can easily repeat the experiments to test the claims made in the paper.