It was probably inevitable. With the growing use of preimplantation genetic diagnosis (PGD) , in which embryos created by in vitro fertilization are screened for genetic defects, the day was going to come when fertility doctors used it not for the well-established purpose of identifying glitches thatinvariably lead to disease—mutations such as those causing hemophilia, fragile X syndrome, neuromuscular dystrophies, Rett syndrome, cystic fibrosis, Tay-Sachs, sickle-cell anemia and Huntington disease. Instead, PGD was also going to be used to look for mutations that onlymight lead to problems. Doctors announced that the first baby screened for the breast- and ovarian-cancer gene BRCA1 was born in Britain.

Granted, this was in a family who, doctors said, had had three generations of women stricken with breast or ovarian cancer in their 20s. The family’s desire to spare their baby girl a similar fate is understandable. As Paul Serhal, Medical Director of the Assisted Conception Unit where the diagnosis took place, said, “This little girl will not face the specter of developing this genetic form of breast cancer or ovarian cancer in her adult life. The parents will have been spared the risk of inflicting this disease on their daughter. The lasting legacy is the eradication of the transmission of this form of cancer that has blighted these families for generations.”

But here’s the rub. The risk of developing breast cancer if you have a mutation in BRCA1 is very likely not the oft-quoted 82%. In late 2004, scientists led by Sholom Wacholder of the National Cancer Institute went public with long-simmering concerns about the 2003 study that established (in the mind of the public and doctors, though not geneticists well-versed in the pitfalls of this kind of study) the risk of breast cancer conferred by carrying mutations in BRCA1: an 82% chance of developing breast cancer by age 80, compared to 12% for women overall.

As Wacholder and his colleagues pointed out, that estimate is much higher than what other studies of women with the mutations have found—46% in one, for instance, with other estimates in the range of 50% to 60%. The study that yielded the higher estimate, Wacholder told me in 2005, contained serious flaws.

In a nutshell, it recruited cancer patients and then examined their DNA for BRCA1 mutations. But that means that people with the mutationsand no cancer were excluded from the analysis. That would inflate the risk estimate. (How might you have the "bad" form of BRCA1 and not develop breast cancer? Presumably because other genes undo whatever damage BRCA1 mutations do.)

“This is not a small disagreement, such as whether the true risk is 90% or 95%,” Wacholder said. “The disagreement is whether the true risk is closer to 50% or to 80%. That’s the difference between a coin flip and a virtually-sure thing.”

Just to repeat, in the case of the British family their BRCA1 mutation plus genetic background seemed to make breast cancer as close to a sure thing as you can get. But what of the countless other families without this history? If we start screening for mutations that might or might not bring on disease, and discarding those embryos that don’t make the cut, let’s at least acknowledge what we’re doing: an unprecedented experiment in directing human evolution. Will families demand PGD for genes associated with risk-taking? neuroticism? marital infidelity? Do they care that most such gene-disease links have failed to hold up, and that the track record of genes and behavior is even worse? I'm betting not. The desire for the perfect, flawless baby will not be thwarted by science, let alone logic.