3 ms·
The (Im)Possibility of Fairness
- cbnotfromthere 6y agoYou know you're dealing with SJW zealots masquerading as researchers when they base their whole paper on such wrong absolutes: "Current sentencing risk assessments are racially biased and job advertisements discriminate on gender" Racial bias is absolute? No degree whatsoever? And no chance that "bias" may reflect general cultural differences (such as higher rate of criminality)? "systems might use more directly measurable—and more problematic—features such as college GPA upon graduation" Why and how could possibly college GPA be a problematic feature at all? Looks like they have a problem with reality.
- derbOac 6y agoInteresting read... most of this is restating work in the psychological measurement literature on measurement invariance and predictive invariance, especially with regard to notions of fairness. There's a large literature on this in the areas of IO and educational psychology, and there have been official position papers about these same ideas in the context of litigation and case law.