A model is an opinion embedded in mathematics

O'Neil's foundational claim is that every mathematical model, however sophisticated, is built from human choices: what to measure, what to optimize for, what proxy to use for something unmeasurable, and what data to feed it. None of that is neutral — each choice bakes in the modeler's assumptions and blind spots, whether or not the modeler intends it.

She compares building a model to describing a family dinner using only the food eaten — technically accurate, but omitting the conversation, mood, and relationships that actually define the event. Every model necessarily simplifies reality, discarding what its builder decided not to measure.

The problem isn't that models simplify — all models must. The problem is when a simplified, opinion-laden model is presented and treated as objective truth, insulated from scrutiny by the appearance of mathematical rigor. A formula is not automatically fair just because it's a formula.