What might make sense is to write various vignettes exploring phenomena that happen around each sort of model. The ways in which big platforms exploit user ignorance, the dragon for an open platform dealing with a tricky bot network, the internal arguments around antiscrapers using AI to detect AI.