x
Selective similarity metrics for imitation - History — LessWrong