uspol, data science (?)
While things are kind of bonkers right now, this is an interesting observation about how state actors influence US elections (past and present): https://www.emptywheel.net/2019/02/25/on-august-2-2016-paul-manafort-gave-konstantin-kilimnik-75-pages-of-recent-detailed-polling-data/
One of the questions I’ve had is how polling data made it into the orbit of microtargeting/psyops companies and their state-sponsored counterparts, since that would be needed to go from how people behave (on social media) to how people vote (with high confidence).
He... gave it to them in paper.
uspol, data science (?)
I guess I’m wondering how nightly news orgs are going to explain how machine learning and inference models work, because so many folks still consider this to be magic.
re: uspol, data science (?)
@zetasyanthis Definitely. But given internal polling data and one of the most robust advertising databases on the planet, you can do some pretty scary things even with a relatively simple model.
re: uspol, data science (?)
@Goldkin Oh, definitely. I'm just sick of everyone also treating it like it's magic pixie dust you can sprinkle on every single problem.
re: uspol, data science (?)
@zetasyanthis I (literally) cannot wait for the Numb3rs explainer on this, because it is going to be so epically cringe-worthy.
re: uspol, data science (?)
@Goldkin I mostly consider it to be bullshit. :P (And yes I know how they work. :P)