Archive

Kaleidoscope: Positionality-aware Machine Learning

Positionality is the spe­cif­ic posi­tion or per­spec­tive that an indi­vid­ual takes giv­en their past expe­ri­ences, their knowl­edge; their world­view is shaped by posi­tion­al­i­ty. It’s a unique but par­tial view of the world. And when we’re design­ing machines we’re embed­ding posi­tion­al­i­ty into those machines with all of the choic­es we’re mak­ing about what counts and what does­n’t count.

The Essence of Sandwichness,” Including the Official New York Definition of a Sandwich

I love read­ing gov­ern­ment mem­os because the world is an unclas­si­fi­able place, but the peo­ple that write these mem­os have to try to clas­si­fy it any­way, and the results get weird. You get these clas­si­fi­ca­tion odd­i­ties, these reg­u­la­to­ry platy­pus­es.