This article critiques Cass Sunstein's argument that algorithms can eliminate cognitive biases and noise to improve decision-making in public life. I argue Sunstein's theoretical position on the uses of algorithms largely holds, but important empirical and technical perspectives must be considered to avoid misuses. This article has two parts, with the first part focusing on the confluence of algorithms and bias and the second part focusing on algorithms and noise. I review several studies that show algorithms in the public sector produce more complex outcomes than Sunstein's theoretical position suggests. I also challenge the notion that algorithms eliminate noise by considering measurement error and non-deterministic systems. While it is important to consider the potential uses of algorithms in society, this question must be met with another: how are algorithms used?