The curious case of 23andMe

November 26, 2013, 3:11 PM UTC

FORTUNE — For $99, 23andMe will sell you a “DNA Spit Kit.” It comes in a bright green box with an opening flap that reads “welcome to you.” Inside is a barcode, which you register online, and a vial for you to hock some genetic material into. The kit is a promise, too. Mail it back to Mountain View, and six-to-eight weeks later the company will email you a login, so you can look at information — genealogical and biological — based on 23andMe sequencing a fraction of your genome. Just what the data means, what it may tell us about ourselves, our current and future in health, remains a matter of much debate. The future is very difficult to know, and besides, your genes are not your destiny.

But the trouble really begins with that spit kit: You can’t buy one right now. The FDA banned 23andMe from selling it. You can still buy one, but the FDA asked 23andMe to stop marketing their product. The question is: Why?

“This product is a device” the FDA letter states. From that point forward the terms are set. This isn’t a drug, 23andMe isn’t curing a disease or alleviating an ailment. The company is selling a thing (the kit) by saying it can provide “health reports on 254 diseases and conditions,” including categories such as “carrier status,” “health risks,” and “drug response,” and specifically as a “first step in prevention” that enables users to “take steps toward mitigating serious diseases” such as diabetes, coronary heart disease, and breast cancer …” Most of the uses “listed on your website, a list that has grown over time,” the FDA writes, “are medical device uses [for the] Personal Genome Service.” So if the spit kit is a medical device, which the FDA believes it is, it must be approved for marketing by the FDA.

MORE: Why Amazon is goosing Prime memberships

The FDA approval process for medical devices is insane. Even the FDA knows it’s way too long and way too expensive. The agency has been trying to make this better, or at least it has issued a report about making it better, but FDA approval is by its very nature anti-Valley, the opposite of moving fast and breaking things — which is why so many health trackers and similar devices (even apps) are very careful about their claims. No one wants to go the FDA approval route if they can avoid it.

Instead, the FDA finds you. In 23andMe’s case, the FDA has been calling and writing since 2009. Last year, 23andMe even started to fill out a 510(k), the FDA’s application for medical device clearance. But since May, 2013, 23andMe hasn’t called, it doesn’t write — even after “more than 14 face-to-face and teleconference meetings.” And the FDA is especially mad because, since then, 23andMe has “initiated new marketing campaigns, including television commercials.”

So far 23andMe’s official response to the FDA’s 1,485 word letter has been terse. Posted on its website, it reads, in full:

We have received the warning letter from the Food and Drug Administration. We recognize that we have not met the FDA’s expectations regarding timeline and communication regarding our submission. Our relationship with the FDA is extremely important to us and we are committed to fully engaging with them to address their concerns.

23andMe is a seven-year-old company run by Anne Wojcicki, a biotech entrepreneur who is married to Google (GOOG) co-founder Sergey Brin, who is, in turn, a big investor in 23andMe. A year ago, after the company announced it had raised $50 million in Series D funding, Wojcicki said that they were focused on lowering cost to bring in more users — a million by the end of 2013 was the company’s stated goal. More users means more data, and it is this rich dataset of genetic information that 23andMe aims to mine. “By having 1 million individual users you can get to a scale where researchers are running data queries through 23andMe, where drug studies leverage our data, and where individuals can more easily be connected to studies that could benefit them,” Wojcicki said.

Genomic medicine is still such a very young science. It stands, as David Dobbs recently wrote “in a bizarre but lovely state of confusion.” An individual’s full genome can be sequenced today in a few hours, for $1,000 — the first genome sequencing took 13 years and cost $3 billion, and that was merely a decade ago. But medicine is not governed by Moore’s law. The next decade will be far more difficult, and not just for 23andMe, but for any company that deals in genome data. Scientists are only just discovering what the data means, what it may tell us about ourselves, our current and future in health. After all, even when you get a peek at the genetic blueprint, the future is awfully difficult to predict.