Twitter doesn’t know how to address the troll problem without diminishing the power of the platform’s network effect. This is a (notoriously) hard problem; but an engineering-driven culture is unlikely to believe in, let alone adopt, the vision of someone whose view is based on an understanding of human behavior at a large scale (as opposed to an understanding of tech).
So a tech-driven platform is left looking at raw metrics with limited insight. There’s no good way to know whether, if they police users (or allow users to restrict interactions with other users) in a meaningful way (which would require programmatically limiting certain types of engagement with the platform), net engagement will go up or down.
And it’s not clear how they ever go from not knowing to investing in the kinds of fundamental changes that would be necessary to help address the problems they are experiencing now.