An advantage astrologers have over genetic testing is that the prediction of astrologers are personally verifiable. An astrologer once emphatically stated that I had no chance of a career in international cricket or Bollywood. So far both claims have turned out to be remarkably accurate.

How does one personally verify a “12.5%” increased chance of lung cancer, the sort of number the vendor for genetic testing 23andMe produces? If one develops lung cancer how would one know that the chances were indeed 12.5%, not 6.25% or 25%?

We die only once. Whether one ends up with lung cancer or doesn’t, the veracity of the claim can be made only empirically. Meaning we need to see how many develop lung cancer out of 10, 000 people just like us.

Yet there is an element of scientific precision in the number, augmented by the decimal point. And it is precisely because genetic testing tends towards science not metaphysics that it falls within the dominion of the Food and Drug Administration (FDA). FDA does not regulate palm readers.

FDA has asked 23andMe to stop sales of its genomic testing.

As a libertarian seeped in the Austrian school of economics, I am generally disposed against regulations. I also share the sentiments of the monetarist Milton Friedman that the true costs of the FDA must also include the treatment opportunities foregone in their lengthy review process.

So it hurts me to be somewhat sympathetic of FDA’s stance on 23andMe, even as I think an outright ban was a tad harsh.

Genetic testing falls in that spectrum of the market where one reaps the benefits but spreads the costs. Economists call this a negative externality. Bankers are most familiar with this. Remember the adage “privatize your gains and socialize your losses.”

Who will counsel the person wondering if she should take anti-coagulants for a trans-Atlantic flight because her genomic analysis shows an 8% increase in developing blood clots over the population average?

Who will shoulder the blame for not anticipating the suicidal ideations of a man who takes his life for finding out that he is destined for Huntington’s disease within ten years?

Who will bear the costs for the overtesting and overdiagnosis that will inevitably result when people find out that they deviate from the population mean in the risk of various cancers?

Not 23andMe, which simply gets paid per Pandora’s box it opens.

The additional costs of investigation will be reflected in our insurance premium. The burden and legal risk will fall chiefly upon the already overworked primary care physicians (PCPs).

How will the PCPs advise? Must they advise a woman with a 22.2% increased risk of ovarian cancer differently from someone with a 6.7% increased risk? What is the threshold of increased risk of cancer for pursuing more tests? What are the tradeoffs of such pursuit?

These questions are expected when we are dealing with imperfect information. However, these questions need to be answered rationally and scientifically if genomics truly revolutionizes patient-centered medicine.

That is not to say positive externalities cannot emerge from genetic testing. My attempt at convincing my domestic comptroller to invest in an indoor mountain climber has failed for the third year. I feel I might have succeeded if I had genomic analysis that reported “6.2% higher than average risk from procrastination-induced heart disease” (of course, it might have backfired and led to tight control of my many epicurean activities).

The point is that the ban is excessive. Instead, a sort of Pigovian tax could have been imposed on 23andMe for every $99 it collected for issuing a genomic report.

Saurabh Jha is a radiologist and can be reached on Twitter @RogueRad. This article originally appeared on The Health Care Blog.