Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

TLDR: There are two different intuitive definitions of predictive fairness. Given populations with different base crime rates, it is impossible to meet both fairness criteria.

If you can survive the WP payroll, the clearest article I've found on this is here:

https://www.washingtonpost.com/news/monkey-cage/wp/2016/10/1...

The full paper is here: https://arxiv.org/pdf/1610.07524v1.pdf




The clearest I've seen is this interactive visualisation: https://research.google.com/bigpicture/attacking-discriminat...


That is great to play with, although it seems to gloss over the fact that you can't have both kinds of fairness.


It defines 4 separate objectives (profit maximisation and 3 types of fairness), and shows that they each result in different thresholds. Just click any of the 4 red buttons.

It then goes on to advocate a particular kind of fairness.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: