Beyond algorithmic equivalence: self-modelling

Link post