This article is a preview from the Summer 2020 edition of New Humanist
The genetic testing company 23andMe laid off a hundred people in January, 14 per cent of its workforce. It was a dramatic turn for the direct-to-consumer pioneer whose star had been rising for over a decade – at least since 2008, when Time magazine declared its saliva-based DNA testing kit Invention of the Year.
Estimates in early 2019 placed the number of people in the United States who had had their DNA sequenced by 23andMe and its competitors at a staggering 26 million. After that, sales collapsed, for reasons that are as yet unclear. It’s possible that the popularity of the service, which at its peak saw hundreds of people uploading videos of themselves spitting into the kit to YouTube, simply faded. Or maybe the number of those willing to engage was very large, but ultimately finite.
Either way, the company’s relatively short history shines a light on issues ranging from personal health to collective surveillance, from privilege to inequality, from genealogy to race (and a great deal of racism).
It all began with a big, glitzy party during New York Fashion Week 2008. The US magazine Mother Jones called the assembled influencers who took turns depositing their saliva into 23andMe’s gleaming tubes “the spitterati”. The name stuck, as it were.
The choice of location and clientele would establish some of the key themes of the project. The early spitterati were wealthy socialites having fun, sending the message: this is who I am, and I’m taking control of it. From the outset, in interviews and promotional materials, the company has emphasised themes of personal fulfilment and empowerment. As late as last year, co-founder Anne Wojcicki was repeating this mantra:
We have over 5 million customers. People are empowered with their genetic information. Once you get this first taste of learning about yourself, it really changes how you think about your health overall, and your own sense of empowerment in your health.
It’s funny she should have said that, because health is only one component of the service, and arguably is not its main drawcard. A clue lies in the fact that 23andMe’s main competitor is called Ancestry.com, and that their entry-level product, currently priced at $99 on the US market, only provides information on “ancestry and traits”. For health information, you’ll have to fork out another $100, while $499 buys the full VIP package, with overnight shipping, priority lab processing and a “1-on-1 Ancestry results walkthrough”. The company recognises that most people are interested in their genetic information primarily for ancestry purposes. There is no health-only kit.
But if most of 23andMe’s customers are merely interested in the geographical origins of their DNA, we might wonder what makes the proposition so valuable to investors. 23andMe is a so-called unicorn, a term reserved for the rare privately held companies that have attracted over $1 billion in capital without resorting to a public offering. Its likely current valuation is around $2 billion. As Sarah Zhang quipped in the Atlantic in 2018, after the pharmaceutical giant GlaxoSmithKline pumped $300 million into the company: “You don’t make that kind of money selling $99 spit kits.”
Think of ancestry data as a lure, or a window into your genome. You might want to know where in the world you (or, more precisely, your ancestors) are from, and the market offers a very affordable solution, likely sold below cost and bundled – for no extra money, at certain times of the year – with some health information. These kits have made for popular Christmas presents from relatives. You return your spit to the lab (or a swab, for some competitors) and receive, in return, an estimate of the predominant geographical distribution of your genes, as well as access to the raw readout of your sequenced data. This very long string of numbers won’t mean anything to you, unless you’re a geneticist, but is nonetheless yours to keep. It’s also 23andMe’s to keep. A board member put it in very plain terms in 2013:
The long game here is not to make money selling kits, although the kits are essential to get the base level data. Once you have the data, 23andMe does actually become the Google of personalised health care.
The reference is more than accidental: Wojcicki was once married to Google’s co-founder Sergey Brin, who carries a mutation for Parkinson’s disease. This might well have contributed to his decision to invest in the company, as well as to 23andMe’s own interest in mapping the genetic roots of Parkinson’s. This tangle of relationships matters to the story, as do the Silicon Valley origins of 23andMe, which was headquartered in Mountain View, the birthplace of Google, before moving to Sunnyvale, also in California.
Google itself went on a trajectory, from the famous company slogan “don’t be evil”, to its status today as a quasi-hegemonic global hoarder of personal data with an almost unfathomably profound influence on our everyday understanding of privacy. Most of us make myriad daily transactions that involve surrendering our personal data to the company in exchange for services or convenience.
This is the company that 23andMe models itself after. It should therefore come as no surprise that it proceeded to attempt – for the most part, successfully – to redefine what counts as a “medical device” as it moved into the analysis of DNA information for explicit health purposes.
Initially, this proved to be a bumpy road. Following a breakdown in the approval process, the US Food and Drug Administration (FDA) banned 23andMe from marketing its “Personal Genome Service” kit in late 2013. After two weeks of negotiations, the ban was lifted for ancestry and trait information but not for health information – meaning the company could continue to collect the data, just not interpret it for the customer. During this temporary blackout, 23andMe was able to make inroads with its health products in the UK and Canada, and finally resumed marketing tests for carriers of a range of genetic diseases in late 2015, with the blessing of the FDA. These were gradually expanded from carriers to gene mutations and now include a host of conditions including late-stage Alzheimer’s Disease, Parkinson’s, coeliac disease and – perhaps most notably – breast and ovarian cancer in women, and prostate cancer in men.
None of these tests have a specific diagnostic purpose – instead, they merely alert the individual to possible genetic risks by looking for specific gene mutations. This is how 23andMe has managed to avoid being labelled as a health provider, which would mean being bound to the associated restrictions and expectations of medical confidentiality. That leaves the company free to engage in proper Google-like data mining.
The issue of what the company can do with individualised as opposed to anonymised and aggregate data is regulated by terms of service that are not only opaque but also subject to change. Besides, the notion of anonymity makes less and less sense as the granularity of tests increases, and more people make use of these services and end up on public DNA databases. Take the case of Joseph James DeAngelo, who is currently standing trial for a series of crimes committed in the 1970s and 1980s in California by the so-called Golden State Killer. DeAngelo was – decades later – identified by the FBI based on DNA at one of the crime scenes. However, DeAngelo himself wasn’t on any database – a dozen or so of his very distant relatives were. These were people with whom he had great-great-great-grandparents in common. From there, the search was narrowed down to two suspects, one of whom was ruled out by a DNA test. In 2018, DeAngelo was charged with 13 counts of murder and 13 counts of kidnapping. His trial is ongoing.
On one level, 23andMe engages in the methods of web surveillance common to just about any company with an online presence, chiefly for advertising purposes. At the same time, it has spent a decade collecting information of an even more intimate nature from millions – data of obviously high value not just to law enforcement, but to pharmaceutical companies and insurers too. This is what made Charles Seife of Scientific American call the company “terrifying” as early as 2013, during the partial FDA ban. He wrote:
. . . as the FDA frets about the accuracy of 23andMe’s tests, it is missing their true function, and consequently the agency has no clue about the real dangers they pose. The Personal Genome Service isn’t primarily intended to be a medical device. It is a mechanism meant to be a front end for a massive information-gathering operation against an unwitting public.
Is the public really “unwitting”? Some have pushed back, allowing for the possibility that 23andMe is fostering genetic literacy and giving their customers greater control over their health. Ethicists Giuseppe Testa and Margaret Curnutte characterise the field of direct-to-consumer testing as “a conspicuous instance of co-production”, and further propose that “a new kind of genetic knowledge and a new kind of biological citizenship is being articulated around the figure of the genetic consumer.”
This model of the citizen-consumer, however, opens up yet another set of issues: as the early adopters at that 2008 spit party exemplify, the citizens of this new genomic nation are overwhelmingly wealthy and white. The implications for DNA mapping and health research are considerable, especially given that the size of this volunteer patient population dwarfs anything that public institutions and publicly funded research – which US law requires to be ethnically representative – have been able to draw upon. In a darkly ironic twist, minorities continue to be over-represented in burgeoning forensic DNA databases in the US. The makeup of these mirrors racial disparities in arrest practices and incarceration rates. As Peter Chow-White and Troy Duster have remarked, these digital divides “risk exacerbating legacies of inequality in bio-medical research and policing”.
When it comes to the relationship between genomic research and health, we know that the map defines the territory: as more people of European ancestry (read: white) swell up the databases, scientists and their algorithms get to know their DNA better and are able to identify more predictive patterns. This was made clear by 23andMe from its early forays into genetic risk reporting. In a 2012 blog post, for instance, the company candidly explained:
23andMe first began allowing customers to learn about their genetic risk for late-onset Alzheimer’s disease a little more than a year ago. We report on variants of the APOE gene (e2, e3, and e4) and provide a risk estimate for the disease for customers with European ancestry.
This selectivity continues to weigh on the company’s reputation. The mutations of the BRCA gene it currently reports on – which increases the risk of some cancers – are a handful out of more than a thousand. As a February 2019 New York Times editorial – possibly contributing to the decline in kit sales – explained:
[This] makes it difficult to draw conclusions from the 23andMe BRCA test. Just because you test negative for the few mutations that 23andMe screens for, doesn’t mean that you won’t get breast cancer. It doesn’t even mean that you won’t get breast cancer caused by a BRCA mutation.
The disease risk tables on 23andMe’s website – which show what conditions the company currently tests for – make especially stark reading. They comprise three columns: European, East Asian and African. The first column is always almost entirely filled out with tick marks, one for each disease. The second shows less than half. The third is nearly empty. This speaks to what Sandra Soo-Jin Lee has called “the selective promise of genomic medicine and the cost of non-participation”. The fewer ticks in the second and third column may seem like a dispassionate, objective snapshot of the current state of our scientific knowledge; but one could equally opine that we know only what we seek to know, at the expense of what we don’t.
This story has a coda, and it’s all about race. There was a widespread belief, at the time when ancestry kits hit the consumer market, that goading white supremacists into taking the test might force them to relinquish their ideology as they inevitably discovered that they were of mixed blood, just like everybody else. It was a very silly notion, of course, but it found a temporary victim in prominent US white nationalist Craig Cobb, who accepted such a challenge in 2013 during an appearance on the Trisha Goddard Show. The test found that he had 14 per cent Sub-Saharan African ancestry.
The news initially caused him some problems within the movement. He responded by taking a second test which showed his heritage to be overwhelmingly white European. Armed with these new results, Cobb denounced the makers of the kit used on the TV show as part of a ‘“Jewish conspiracy” whose “intent is to defame, confuse and deracinate young whites on a mass level – especially males”.
In a brilliant piece of research on direct-to-consumer genetic testing and its use by white nationalists, Aaron Panofsky and Joan Donovan show that Cobb was able to turn the tables and find “data and interpretations of the science (and the testing companies) that confirm[ed] his ideas of whiteness”. Studying similar cases as they played out on the Stormfront website, the two researchers examine the complex “repair” strategies used by members of that community (which only admits “non-Jewish people of wholly European descent”) to either reject the findings of the tests or reinterpret them by focusing on different aspects of the science. Either way, they conclude, it seems genetic ancestry tests “are not making white nationalists more, and certainly not less, racist”.
When the American white nationalist Richard Spencer proudly tweeted out the results of his 23andMe test, showing that he was 96.3 per cent European, he engaged in the same fallacy as Elizabeth Warren did when she quoted her results as evidence of Cherokee ancestry. This impoverished understanding of genomic information treads old colonial paths, seeking to trace minute blood quanta as if they linked the individual to a collective – and continuously transmitted – set of shared cultural meanings.
As in the case of risk tables that, by picking too few genes, can fill us with false confidence or equally false fear, the ancestry information offered by the likes of 23andMe is both precise and meaningless: it doesn’t tell our story, because that only exists in the infinitely more complex and hard-to-read text called culture.
