Gradient Update #9: Bias Bounties and Hierarchical Architectures for Computer Vision