When the Optimum is also Blind: a New Perspective on Universal Optimization