Please try to elevate the debate. See [1]. You're at DH0 right now.
> They're not truth, they're are nothing more than models. The truth can fit a plethora of models.
Models receive past data and emit predictions. We can then see how well those predictions match future data. We call one model "better" than another model when that first models' predictions more closely match future data than the second model's. Not all models are equivalent. The ML fairness people want to make model predictions less accurate because they don't like what the predictions say. Prioritizing truth over pleasantness isn't juvenile: it's the opposite. The mark of maturity is the willingness to accept an unpleasant reality instead of denying it.
Please try to elevate the debate. See [1]. You're at DH0 right now.
> They're not truth, they're are nothing more than models. The truth can fit a plethora of models.
Models receive past data and emit predictions. We can then see how well those predictions match future data. We call one model "better" than another model when that first models' predictions more closely match future data than the second model's. Not all models are equivalent. The ML fairness people want to make model predictions less accurate because they don't like what the predictions say. Prioritizing truth over pleasantness isn't juvenile: it's the opposite. The mark of maturity is the willingness to accept an unpleasant reality instead of denying it.
[1] http://www.paulgraham.com/disagree.html