NEW YORK — 23andMe caused Leonard Kim not one identity crisis but two.

The first came in 2016, when Kim, who was raised believing he was 100% Korean, took the company’s DNA ancestry test and learned he was almost half Japanese. Then, earlier this month, he was out for drinks with his wife and some friends near his Los Angeles home when someone questioned his Japanese ancestry.

So Kim, 34, took out his phone and consulted the 23andMe website. That’s when he discovered his ethnic identity had changed. The site that once told him he was about 40% Japanese now pegged that figure at 5%. He was, in an instant, fully Korean again.

advertisement

Korea’s history with Japan is one stained with colonization, cultural oppression, and bloodshed. It took months of denial and soul-searching for Kim to accept a previously unknown Japanese lineage. Learning it was all a waste felt like existential whiplash.

“People in general, we have so much of our story that we use to define who we are, and when you have it all come together and then shatter apart, it puts you in this strange place,” Kim said. “It’s like, OK, so what’s my real identity?”

Kim’s story is not unique among the millions of people who have bought into consumer genetic testing for clues about their ancestry. Peter Cho, the head of design at a tech company, had roughly the same experience after submitting his saliva sample to 23andMe, yo-yoing from a shocking 58% Korean to 95%. Other customers have had less jarring changes, watching, for example, as small amounts of Spanish DNA have vanished to be replaced by genetic ties to Italy.

To the companies selling tests, the ever-evolving DNA ancestry reports are more of a feature than a bug. Each website is upfront about the fact that a given test result is only as accurate as the data behind it. With more data come more granular conclusions that almost inevitably tweak ancestry results. And because genetic datasets have long been overwhelmingly white, even an incremental update can lead to pendular shifts for customers of color. The same phenomenon holds true for genetic tests offering information on health or disease risk.

“We try to convey the notion that this is a living document,” said Robin Smith, head of 23andMe’s ancestry division. “It does change over time.”

It’s easy to see why more consumers might take note of the changes. Direct-to-consumer genetic testing has ballooned from a luxury niche into a viable market, with spit tubes available by the box at Walgreens and marked down for Christmas or Mother’s Day. The field’s two biggest companies, 23andMe and Ancestry, claim to have tested more than 25 million people combined.

“We try to convey the notion that this is a living document. It does change over time.”

Robin Smith, head of 23andMe’s ancestry division

The tests offer an educated scientific guess, but the way they’re marketed could easily convince customers that they amount to conclusive proof of identity, said Debbie Kennett, a genealogist who has written at length about the genetics of ancestry.

One 23andMe commercial follows a tourist on what appears to be a personal genetic sojourn, trekking around the globe as her DNA relation to each region  — 29% East Asian, 3% Scandinavian — flashes on the screen. (The company calls this “heritage travel” and just partnered with Airbnb to promote it.) Another ad from Ancestry features a man named Kyle discovering his long-assumed German lineage is instead linked to the British Isles, prompting a costume change from lederhosen to a kilt.

“I think the way the tests are marketed is misleading, because people are led to believe they’ll take a test and find out who they are or where they’re from, and that is not the way to view these results,” Kennett said.

A spokesman for Ancestry said its “testimonial ads” are based on customers’ “first-hand experience, in their words.”

The whole premise of spitting into a tube and learning, say, that you are French or Colombian is inherently fraught, said Arthur Caplan, a bioethicist at NYU School of Medicine. DNA is science, yes, but ethnicity is a cultural construct, and borders are political ones. To blur the lines therein is to invite trouble, he said.

“This idea that you are your genes is just false,” Caplan said. “Your genes really don’t carry racial, ethnic, and cultural markers. We do that. We divide them up. So there’s always been an underlying suspicion on my part that there’s more culture than biology in these [ancestry] reports.”

Smith, from 23andMe, said the company is mindful of the distinction between genetics and identity, both in its marketing and in how it presents test results. Each of the company’s database updates is accompanied by a blog post explaining what it does and doesn’t mean for customers, he said.

“The bedrock for us has always been the science,” Smith said. “We try to fixate on what does the DNA tell us. We’re clear up front that DNA is not identity. DNA is not culture.”

For some customers, that doesn’t sate the temptation to divine meaning from ancestry test results — and it doesn’t ease the sting of having those results shift under their feet.

Angelica Duke had a genealogical mystery. Her grandfather, since deceased, spent his life insisting that his mother was a lady in waiting to the queen of Spain. On the one hand, that claim clashed with the family’s Scottish heritage, and Duke’s grandfather had a habit of exaggerating. On the other, he spoke fluent Spanish, and his name, Frederic, was inexplicably lacking the K one would expect from a Scot.

It all seemed to come together two years ago when Duke’s 23andMe ancestry results revealed a small percentage of Iberian DNA, leading her family to conclude that, “OK, maybe there is some veracity to this claim,” she said.

Then, earlier this month, Duke signed into 23andMe and discovered that her ties to Spain had evaporated, in their place a connection to Italy. She knew her results had always been subject to change, and that trace amounts of DNA could never close the book on a family mystery. Still, the update was “a little bit of a slap in the face,” she said.

“It is serious,” Duke said. “It’s something people internalize, to a certain extent.”

That was the case for Kim. He works in marketing, where he gets paid to help people build their personal brands. That means encouraging them to dispense with artifice and “tell their true stories,” Kim said.

Since that night at the bar, following his own advice has grown more complicated.

“Maybe I should be trying to dissociate my connection to my heritage because I can’t trust the data,” Kim said. “Maybe I should go back to being just Leonard and not Leonard who’s a Korean or Leonard who’s a Japanese person, which is where I started before I took this test.”

Leave a Comment

Please enter your name.
Please enter a comment.

  • With all the issues we have to divide us, this desire to differentiate between what ‘type’ of human being we are is, perhaps, the least relevant to our future existence. If you lucky enough to be a human, then you have already won the biggest lottery the world has to offer! Celebrate it everyday!

  • There a reason that these services are called “recreational genetics”. It is not serious research level methodology.

    Read the methodologies papers. The reference populations are based on people who trace their ancestry back to eight grandparents all from the same geographical region (approximately 1850 say). There was a lot of migration, invasion and conquest that occurred prior to 1850.

    Some of the reference populations get eliminated and removed as a reported group. I believe they are applying a statistical technique called dimensional reduction to do this. You might be able to trace all 16 of your great-great grandparents to a particular isolated island in the pacific, but if that population doesn’t survive dimensional reduction, you’ll be reported as a mix of others.

    In Europe the Vandal’s, Visagoths, Berber Moor’s, Crusaders all get mixed up and become part of the 1800s “reference” populations.

    Enjoy the results on a recreational basis, but it’s no substitute for serious genealogy.

  • There a reason that these services are called “recreational genetics”. It is not serious research level methodology.

    Read the methodologies papers. The reference populations are based on people who trace their ancestry back to eight grandparents all from the same geographical region (approximately 1850 say). There was a lot of migration, invasion and conquest that occurred prior to 1850.

    Some of the reference populations get eliminated and removed as a reported group. I believe they are applying a statistical technique called dimensional reduction to do this. You might be able to trace all 16 of your great-great grandparents to a particular isolated island in the pacific, but if that population doesn’t survive dimensional reduction, you’ll be reported as a mix of others.

    In Europe the Vandal’s, Visagoths, Berber Moor’s, Crusaders all get mixed up and become part of the 1800s “reference” populations.

    Enjoy the results on a recreational basis, but it’s no substitute for serious geneaology.

  • Part of Popu Popu: Power to the Populations
    Ancestry DNA Debate Remash Video

    Critical considerations on issues of techno-identity
    in an age of genetic ancestry testing

    An Ars Bioarctica Residency :: Bioart Research 23 and Me Mix

    https://vimeo.com/335690825

  • I took the DNA test with MyHeritage.I know that my mother was Irish and my father was half British and half Scottish.MyHeritage said that I am 76.4 percent Irish and 23.6 percent Scandinavian.I think that My Heritage is not accurate.I will believe what my parents told me.

    • Cultural heritage as a value is the basis of racism.

      You are part of a lineage going back millions of years, an unbroken line.

      Your ancestors did what they had to do to get you here.

      Sometimes, that included swallowing hard truths.

    • I am mostly of British origin but both Ancestry and 23and me found substantial Scandinavian DNA. I think this is just the influence of the huge number of Vikings who both invaded and settled in Britain including Ireland. My mitochondrial DNA is from Sweden which may be explained by the fact that in the part of Scotland where my maternal line comes from, the Northwest highlands there was a permanent Viking settlement and they brought their women over from Scandinavia and one of them was the origin of my mitochondrial DNA.

  • The questions is, what are these companies doing with our DNAs? Is it to study us–their way of getting lots of genetic samples; and to what; sell our data to businesses to advertise for potentially needed items or services based on our genetic composition; the same way banks and stores monitor our spending and how our phone listens to us and scatter related links all around; or is it something much larger than that; to produce artificial intelligence by copying our genetic make-up, or for oligarchs to eventually learn how to control human society? What are your thoughts?

    • Yes, yes, yes, and no.

      AI would be personality based, and the DNA of an artificially created human with an AI mind would ideally be without personal cultural heritage.

      The actual 4th reason is medicine, and the non-mentioned 5th reason is unsolved crimes.

      The chances of the Zodiac Killer having already taken the DNA swab for 23&Me are just as high as the Golden State Killer’s were.

  • I went through the 23 & Me process, which I now summarize as “I told them where I came from and they reported it back to me.” Fact is, what you tell 23 & Me (and others I would guess) about your ancestry becomes part of the database and your DNA profile is normalized into the next persons calculation. If ten people with related DNA report they are Country A, the database will report Country A for the eleventh person with related DNA. If the next hundred people with related DNS report they are Country B, then people 1-11 will also flip to Country B the next time they check their reports.

  • Excellent report, and timely too, for my mother and me, as we’re considering delving into this via Ancestry or a similar service. Given we’re Parsis, descendants of Zoroastrians from Persia who settled in India about a thousand years ago or so, it should be an interesting exercise.

  • It sounds as if 23andme is covering up errors. We are not talking about susceptibility to disease (even if you’re charitable and argue that the interpretation could change as new studies are published, they shouldn’t change as much as is described here).

Sign up for our Daily Recap newsletter

A roundup of STAT’s top stories of the day in science and medicine

Privacy Policy