Yet, any company who open sources an algorithm can simply apply any private alternative algorithms using any alternative/undisclosed data sources to any undisclosed subset of users as they wish.
Hell, whose to say that some people aren't chosen for long term "experiments" with different behavioral/belief modification goals? That's what I would do.
Even if there were some sort of open source spec like the W3C specs, for instance, Twitter would still be a black box that could choose to conform to the spec or not choose to.
What we need is something more enforceable. So either you do something completely green field like https://blueskyweb.xyz/, or you spend $43B & roll up your sleeves to try and fix the problems with the current platform.
Hell, whose to say that some people aren't chosen for long term "experiments" with different behavioral/belief modification goals? That's what I would do.