Lesson from machine learning: "tells it like it is" with no sense of where/why/how "it" came from/is going is buggy software...
Conversation
territory>map [data]; territory>narrative. Good narrative>map. Bad narrative<map. Truer narrative=spills out of map more
1
1
1
fwiw not all ML is opaque (like SVM). Decision tree algos can explain their reasoning => bones of a narrative.
1
this is good on the topic, though already obsolete going by convos with ML friends nautil.us/issue/40/learn

