Start by questioning the premise: _should_ algorithms make moral decisions at all?