I’m starting to worry that the most popular perspective from which people not familiar with ML can tell loss functions and sampling algorithms apart is through the prism of BLM culture. Although, I agree with your sentiment that there is always a thin end of the wedge.