Thread · 3 tweets · 19 Sept 2019

Happy to agree to disagree! Your description leads me to a question though (if you don't mind). You seem to describe a system of strong internal accountability, through transparency, docs, etc.

So when research like twitter.com/Miles_Brundage… is published (not a curve ball, literally just something I have in a tab) I presume you can look at some internal docs that eg. disprove the research or outline a plan to fix the issue?

My question is: if I indeed correctly understood the accountability that grounds your trust, how do you explain that the public-facing response on this topic is so poor (basically "the algorithm did it")?