The content isn’t the problem. Or, well, most of the content, anyway. The problem is who controls its recommendation algorithm and collects the data.