CONCEPT
Opinions Embedded in Mathematics
Cathy O'Neil's foundational claim that mathematical models are not neutral discoveries but constructions shaped at every step by human choices—about what to measure, what to optimize, and what counts as success—and that those choices are opinions, not facts.
The belief that algorithms are objective is not merely wrong; it is dangerous in a specific way. It is dangerous because it takes the human judgment embedded in every model—the choices about what data to use, what outcomes to target, what counts as a good result—and hides that judgment behind the authority of mathematics, which most people are trained to regard as neutral and beyond dispute.
Cathy O'Neil's central claim is that this concealment is not incidental to how algorithmic systems operate but is often their most dangerous feature. A person rejected by a biased human cannot argue with the human—but there is at least a human to argue with. A person rejected by an opaque algorithm cannot argue with it at all, because the bias has been laundered into mathematics and presented as an impersonal verdict. The phrase
opinions embedded in mathematics is O'Neil's attempt to reverse the laundering: to make visible the human