Type 2 as an aggregation of Type 1 processes — LessWrong