In honor of #ScholarStrike, I’m going to share some resources I know about math and policing. People use math to build policing that disproportionately harms people of color. And people use math to try to dismantle it.
This reporting by @propublica about recidivism algorithms is excellent. High schoolers could read it. It highlights the math in a rigorous way while also being very human. propublica.org/article/machin…
Here are papers that show the mathematical impossibility of a fair recidivism algorithm. They focus on the human cost of baked-in unfairness, especially for people of color. Almost fair is not nearly good enough.
arxiv.org/abs/1609.05807
papers.nips.cc/paper/7151-on-…
arxiv.org/abs/1609.05807
papers.nips.cc/paper/7151-on-…
@YehCathery shared this article yesterday about mathematicians calling for colleagues to stop contributing to predictive policing. Predictive policing has been shown to lead to over-policing of neighborhoods in which people of color predominantly live.
insidehighered.com/news/2020/06/2…
insidehighered.com/news/2020/06/2…
And here is a paper showing one reason why predictive policing has this effect, because of feedback loops.
proceedings.mlr.press/v81/ensign18a.…
proceedings.mlr.press/v81/ensign18a.…
If it surprises you that math is used to build and bolster systemic racism in policing, read this stuff. Dig into the mathematical research around it. No matter your formal math training, you know enough math to ask the right questions.
Here is a talk about how data gleaned through biased policing makes its way into predictive policing programs, perpetuating the problem.
cyber.harvard.edu/events/dirty-d…
cyber.harvard.edu/events/dirty-d…
