Polygenic embryo screening arrived faster than the public conversation. By the late 2010s, IVF clinics in the United States quietly began offering reports that ranked embryos by predicted risk of schizophrenia, diabetes, heart disease, and—more controversially—by predicted height, educational attainment, and IQ. The underlying math is unremarkable: take a genome-wide association study, sum the weighted contributions of hundreds of thousands of common variants, and produce a single number. The cultural shift is not unremarkable. For the first time, prospective parents at the embryo stage can choose between their own potential children on the basis of probabilistic traits that go well beyond Mendelian disease.
This is a collective question dressed as a private one. Each parent's decision is intimate, but the aggregate of millions of such decisions reshapes the gene pool of the next generation, redistributes who gets born, and recalibrates which traits a society treats as defects versus differences. Disability rights advocates have warned for decades that prenatal screening, applied at scale, expresses a population-level judgment about whose lives count. Polygenic selection sharpens that warning because it does not even claim to identify disease—it claims to optimize.
The technology's predictive power is also routinely overstated. A polygenic score for educational attainment explains roughly 12-15 percent of variance in the population it was trained on, and substantially less in populations underrepresented in the training data, which means nearly every non-European family receives less accurate scoring. Selecting between sibling embryos—who already share half their variance by descent—shaves that predictive power further. The expected gain from picking the "highest" embryo out of five is on the order of two to three IQ points or an inch of height. Parents are being sold near-certainty and delivered statistical drift.
The sixth law—Revise—matters here because the field is moving while the framework is being built. The American Society for Reproductive Medicine has issued cautious statements; the UK's Human Fertilisation and Embryology Authority has not approved polygenic selection for non-medical traits; clinics in the U.S. operate in regulatory silence. A parent presented with an embryo report is asked to make a decision that the discipline itself cannot yet underwrite. They are conscripted into the experiment.
The deeper distortion is to parenthood itself. Selection presupposes that the relationship is forward-engineered—that a child's traits are inputs the parent specifies rather than discoveries the parent receives. This is not new in kind (parents have always chosen schools, neighborhoods, mates) but it is new in degree. The choice now reaches before the existence of the chosen. What follows is a subtle redistribution of responsibility: when a screened-against trait appears anyway, the parent is positioned as having failed at a task they were never told they could fail at.
Kathryn Paige Harden has argued, against both the eugenic right and the blank-slate left, that genetic differences are real and that ignoring them entrenches inequality. Her position is uncomfortable because it requires both honesty about heritability and commitment to redistribution. Polygenic embryo selection threatens to satisfy the first half while abandoning the second—privatizing genetic advantage as a consumer good for those who can afford IVF cycles in the first place. The collective consequence is a stratification that compounds the existing one: families with the resources for selection produce children with marginally optimized scores, and the social meaning of "natural conception" shifts toward something a poorer person resorted to.
A society that wants to revise this trajectory has limited tools. Outright bans tend to push the practice across borders. Disclosure mandates produce informed consumers but not informed citizens. The more durable lever is what gets reimbursed and what doesn't: insurance and public health systems can cover screening for serious Mendelian conditions while declining to cover polygenic ranking for cognitive traits. That line is contestable, but contestability is the point. Drawing it forces the public argument that the technology, deployed quietly clinic by clinic, otherwise avoids.
The parent's choice, in the end, is not really a choice about the child. It is a choice about what kind of parent the chooser intends to be: one who pre-specifies, or one who receives. Both positions can be loving. Only one of them is honest about how little of a person is predicted by the variants we currently know how to read.