They are talking being able to segregate yourselves via the internet - local segregation, while a real thing, still exposes you to a variety of view points. Sel

They are talking being able to segregate yourselves via the internet - local segregation, while a real thing, still exposes you to a variety of view points. Self-segregation via the internet exposes you to a pool 1000x (very conservative here) or more than what's available locally, without too much effort. It's a problem of both how specialized you can get and how easily it can be achieved. You could compare normal/traditional self-segregation to approximating a data set. Self-segregation is akin to overfitting a data set.