Your DNA Is Out There. Do You Want Law Enforcement Using It?

(Bloomberg Businessweek) -- On the afternoon of Nov. 18, 1987, Jay Cook and Tanya Van Cuylenborg left Saanich, B.C., their hometown, to pick up some furnace equipment in Seattle for Cook’s father. Saanich and Seattle are a little more than 100 miles apart, but the trip takes almost five hours: a ferry into the U.S. across the Strait of Juan de Fuca, another across Puget Sound, and between them a winding coastal drive through evergreen forests and fishing towns. The young couple planned to make an overnight jaunt of it. A blurry photo snapped at the time shows them beside the bronze Ford van they took. Van Cuylenborg, 18, holds a walk-like-an-Egyptian pose; Cook, two years older and a head taller, looks off to the side half-smiling, his dark hair falling over one eye.

The next day they didn’t show up at the heating-supply store, nor did they return home that night as planned. On Nov. 24, Van Cuylenborg’s partially clothed body, hands bound by a zip tie, was found in a roadside ditch 75 miles north of Seattle. She had been raped and shot in the back of the head. Two days later, hunters spotted Cook’s body, wrapped in a torn blue blanket, under a bridge in a small town outside Seattle. He’d been beaten over the head with a rock and strangled; a pack of cigarettes was stuffed in his mouth.

Detectives from Snohomish and Skagit counties’ sheriff’s departments had a few things to go on. There was the blue blanket, which didn’t belong to either of the victims, and there was Van Cuylenborg’s missing Minolta camera—the lens turned up in a pawnshop in Portland, Ore. The killer left behind extra zip ties and a box of .380-caliber bullets, and, in what seemed like a taunt, the surgical gloves he’d used to ensure that there were no fingerprints. Investigators carefully gathered up and tagged what they found. They also gathered something else, a new type of evidence.

Only weeks before, a Florida serial rapist had become the first American convicted on the basis of DNA left at a crime scene. In the years since, DNA has come to be seen by investigators and crime-drama fans alike as unique in its infallibility and its power to turn infinitesimal human traces into telltale clues. But then as now, DNA testing had its limitations: Samples could be degraded or contaminated, the results misinterpreted by technicians. And if the DNA left at the scene of the crime didn’t exactly match another sample—one from an existing criminal database or one taken from a suspect—then it really wasn’t much of a clue. In the Cook-Van Cuylenborg case, there was DNA (authorities haven’t specified the exact nature), but there were no matches. The new technology led nowhere.

Over the years, the dead ends multiplied. Jim Scharf, the Snohomish County cold-case detective who inherited the case, included it in a deck of cards printed with unsolved crimes and distributed the information to prison inmates in the hope that they might have heard something. A series of taunting letters sent to the Cook and Van Cuylenborg families from the purported killer turned out to be from a Canadian homeless man unconnected to the crime. In 2010 detectives opened up their files to a group of top FBI and British experts gathered at a nearby forensic-science conference; they didn’t get anywhere, either.

Then last April, California police made an arrest in a decades-old series of rapes and murders. Investigators had used a combination of DNA analysis and old-fashioned genealogy to identify a former police officer named Joseph James DeAngelo as a suspect in the notorious case of the Golden State Killer. It was the first time that particular technique had been used to solve a murder. Scharf, however, had already been wondering whether something like it might work. So, it turned out, had Parabon Nanolabs Inc., a small data analytics company in the business of doing DNA work for law enforcement.

As it happened, the Snohomish sheriff’s department had already worked with Parabon on the Cook-Van Cuylenborg murders. The day after the Golden State Killer announcement, Scharf spoke by phone with the company’s chief executive officer, Steven Armentrout, who told him that, effective immediately, Parabon was offering a “genetic genealogy” service. The company could run the kind of analysis that had identified DeAngelo, and for the Cook-Van Cuylenborg case it could use the DNA file Scharf had already given them. That was a Thursday. On Saturday night, Armentrout received an email from a woman named CeCe Moore. Many of genetic genealogy’s techniques are her invention, and Parabon was in the process of hiring her. The subject line of her email read, “Solved.”

Under sustained assault, behind the sometimes obscuring hype, the human genome is giving up more and more of its secrets. Researchers are piecing together the genetic basis of cancer and Parkinson’s disease, finding tantalizing clues to human longevity, and perfecting new methods not only to read but to write DNA. The resulting technologies are finding their way into unexpected places.

Genetic genealogy might sound to an outsider like a redundant term, but it’s the offspring of two very different intellectual tribes. Genetics is the world of university labs and venture capital, high-throughput sequencers and Ph.D.s; genealogy is a community of self-taught amateurs haunting county records rooms and churchyard cemeteries. Over the past few years, the marriage of their methods has produced a powerful tool for finding previously unfindable people. Since the Golden State Killer announcement, there have been more than a dozen additional arrests using genetic genealogy techniques, and Parabon worked almost all of them. In some instances, the company has been able to produce a suspect in a few hours. “I mean, it’s incredibly powerful,” Moore says of what she does. “It’s powerful in revealing secrets.”

That is also what has begun to worry people. Genetic genealogy wasn’t developed for identifying murderers and rapists, and the question remains of who else, or what else, it could be used to find. More than 15 million people have now taken consumer DNA tests from companies including Ancestry.com Inc. and 23andMe Inc., and more than 1 million have uploaded the results to GEDmatch, the open source online genealogy database Moore and other investigators use in their work. And even if you aren’t among them, if a cousin of yours is, you can be found. A recent study in the journal Science by researchers at Columbia University and the DNA-testing company MyHeritage Ltd. concluded that, for Americans of European descent, there is already a 60 percent chance they could be identified because of a relative in a DNA database. For that number to reach 100 percent, the researchers concluded, only 2 percent of the population needs to have uploaded their DNA.

The applications for this information are sure to multiply: In recent months it’s emerged that the U.S. government performed DNA tests on migrant parents before reuniting them with the children taken from them at the Mexico-U.S. border and that Canadian immigration officials used DNA and ancestry websites to establish the nationality of migrants. Just this month, DNA testing made its entry into the 2020 presidential race, when Senator Elizabeth Warren (D-Mass.) used it to prove distant Cherokee ancestry and ignited a conversation about whether DNA is really what determines a person’s heritage. Genetic genealogy could theoretically be used to help trace mail bombs back to their maker. We’re only beginning to comprehend what can be read from the code that millions of people are sharing. Imagine a world in which DNA data are as free-flowing as your web-browsing habits or the information your phone collects about your location—human bodies, after all, constantly shed DNA. Then imagine all the parties that might find information about your health, heritage, or familial relationships interesting. Most people aren’t murderers, but everyone has skeletons.

Writing about a professional genealogist means, among other things, acknowledging the arbitrariness of starting points. One could, for example, begin Moore’s story with the Great San Francisco Earthquake, when an Australian merchant seaman dashed through the rubble to check up on a Norwegian governess of his recent acquaintance and ended up marrying her and having two daughters. Or circa the year 2000, when Moore, the descendant of one of those daughters, decided it would be neat to draw up a family tree as a wedding present for a niece.

At the time, Moore was a 30-year-old actress, singer, and model living in and around Los Angeles. She’d had some small movie and TV roles, but mostly it was commercials, corporate training videos, and, increasingly, trade shows that kept her career going. She had a regular gig for Mattel impersonating Barbie at toy fairs and store openings. She also worked auto show booths for Porsche—“Not the ones where you're wearing, like, a bikini,” she emphasizes, “the ones where you’re wearing a suit.” It was reliable, well-paying work that rewarded the ability to quickly absorb often technical information.

Moore never finished her niece’s family tree. As she would learn, in genealogy, that’s typical. “It’s a never-ending process,” she says. “I thought I already knew quite a bit about our family. But it just became an obsession, I guess.” Finding distant forebears and unknown third cousins is a series of puzzles, and the rush that comes with solving them is addictive. It’s always possible to build a tree still further back or flesh out its distal branches. Moore learned how to comb through old census and marriage records, to search obituaries and yearbooks, and to mine the new but growing trove of personal information on social media sites.

There already were at that point genetic tests tailored to genealogy. The pioneer was a company called FamilyTreeDNA, founded by a serial entrepreneur who, like many genealogists, had taken on his family tree in what he’d thought would be his retirement. The company’s tests on one level worked on the same concept as traditional forensic DNA analysis. Although the vast majority of the human genome is the same from person to person—we are all one species—there is variation, adding up to about 0.1 percent of the total. That’s enough to account for the dizzying diversity of human specimens. That those variants form unique sequences in each individual is what allows DNA to function not only as a code but also as a fingerprint.

Unlike forensic DNA tests, though, FamilyTree’s were able to do more than just identify a person. In a limited way, they could also identify familial links. The tests looked at either the Y sex chromosome (Y DNA), which only men have, or the snippet of genetic material housed in a cellular organelle called the mitochondria (mtDNA). Because of the improbable molecular choreography of sexual reproduction, both of those bits of DNA happen to be inherited almost unchanged, like a genetic heirloom—Y chromosomes from fathers to sons, mitochondrial DNA from mothers to offspring of either sex. Genealogists whose relatives were willing to take a cheek swab could use their FamilyTree results to bolster their forays into historical records, tracing their lineage back along either their mother’s or father’s side of the family tree.

Then, in the mid-2000s, DNA testing started to go mainstream. The race to sequence the genome had catalyzed a host of technological and scientific advances and piqued public interest, and a new breed of company arose to capitalize on the combination. In 2006, 23andMe was founded to provide personalized genetic health information to anyone who could spit in a tube. DNA extracted from the saliva was amplified, chemically sliced into tiny segments, then placed on special microarray “chips” covered with millions of microscopic beads, each with the ability to identify a particular DNA variant. The next year, 23andMe introduced its first product, a $999 test that offered insights into a person’s disease risk, ethnic ancestry, and other traits, among them their sensitivity to certain tastes. Competitors piled into the market, most prominently AncestryDNA, a subsidiary of Ancestry.com, a genealogy company with roots in the Church of Jesus Christ of Latter-day Saints. (The church holds that dead ancestors, once found, can be posthumously baptized into the faith so the whole family may be reunited in the afterlife.)

Geneticists cautioned that the ethnicity reports could be incorrect or misleading, as could the information about health, and within a few years the Food and Drug Administration notified 23andMe that its genetic tests required federal approval. Privacy advocates warned of the risks of turning over to a corporation a file containing a large chunk of the operating system for your body—for years, AncestryDNA’s terms and conditions explicitly granted the company “a royalty-free, worldwide, sublicensable, transferable license to host, transfer, process, analyze, distribute, and communicate” users’ genetic information. “This genetic astrology could be regarded as producing no more than entertaining horoscopes,” three European geneticists wrote in a paper critical of the burgeoning consumer DNA market. “There is, however, a potential for harm and the need to consider mechanisms to ensure that these tests are evaluated and used appropriately.” In 2008, President George W. Bush, spurred by such worries, signed into law a set of protections against “genetic discrimination.”

Still, the idea that a bit of saliva could at once predict the future and unearth the past was deeply alluring, and the business took off. DNA testing, the companies suggested, could reveal a more authentic, elemental you. “Ancestry helped give me a sense of identity,” one 2015 AncestryDNA ad proclaimed. In another, a lederhosen-clad German-American explained that after a DNA test revealed he was really more Scottish and Irish, he started wearing a kilt.

Genealogists including Moore tend to describe this sort of thing, with faint scorn, as “recreational genealogy.” The tests themselves, however, turned out to have an unforeseen value. To “read” the health information encoded in our genome, companies such as 23andMe couldn’t confine themselves to small idiosyncratic bits of DNA; they had to look at hundreds of thousands of locations on the genome. Autosomal DNA—the DNA housed not on our two sex chromosomes but on the 22 other pairs—would at first seem unpromising for genealogical work. Unlike Y DNA or mtDNA, the bulk of our genetic material doesn’t get passed down unchanged. Instead, it’s reshuffled from generation to generation, as genes from each parent mix and match. That’s why traits for such characteristics as height, hair color, and cancer risk are both inheritable and variable. The unique DNA sequences bequeathed to a child from each parent steadily shrink with each reproductive recombination, and over just a few generations they fade into statistical noise.

But if you trace the DNA variants—the “polymorphisms,” in the language of genetics—you can trace the reshufflings. And what Moore and a few other biologically minded genealogists came to realize is that the recombination rate itself could provide valuable information. Like radiocarbon decaying in an ancient artifact, relatedness steps down in a predictable way: Of that 0.1 percent of DNA that varies among humans, we share roughly half with each parent or sibling; a quarter with grandparents, aunts, uncles, and half-siblings; 12.5 percent with first cousins; 6.25 percent with our parents’ first cousins; 3.13 percent with our second cousins, and so on. The ratios aren’t exact, and the further out one goes the more it’s swamped by randomness—you might not share any DNA with your great-great-great-great-grandmother at all. But there are calculable ranges, enough so that discovering you share a certain percentage of your DNA with someone—information the new testing companies could easily provide—can tell you not only that you are related but how.

And as more and more people took the tests and joined the companies’ databases, the tool grew only more valuable, as the odds climbed that each user would have at least a single third cousin or closer also in the databases. Soon 23andMe was offering a “relative finder” feature. In 2010 a genealogist and a computer-savvy engineer together started GEDmatch, where anyone can upload the files they get from the testing companies. It allows people who have sent their spit into 23andMe to find relatives who have taken a different test, such as AncestryDNA’s.

By that time, Moore was raising a young son from her earlier marriage and running a casting and production company with her longtime boyfriend, but she threw herself into the new world of DNA-enabled genealogy. She set out to turn her own extended family into a test case for its power, badgering cousins and aunts and uncles into sending their spit off to 23andMe and, in some cases, paying for the tests herself (the price had dropped to a few hundred dollars and has continued to fall since). Her family, it turned out, had been cursed with genes for the blood disorder beta thalassemia and blessed with those associated with perfect pitch, and she traced the two traits back along her tree.

Even more valuable than those insights, though, was the work she’d had to put in to identify the roughly 40 relatives she eventually got to test themselves. Traditional genealogy is concerned mostly with digging deeper into the past. Its methods are less well-suited to finding the living—in the U.S., census records, a mainstay of genealogical research, are by law kept anonymous for 72 years. Moore had to develop a new toolkit, tracing family trees outward by building them forward in time from their roots. Social media proved particularly valuable. “People post a lot on Facebook,” she says. “They worry about genetic privacy, but what they put out there on social media is much more telling than what I can tell from somebody’s DNA.”

In a field of mostly amateurs, she became a leader and an authority. From her home in Orange County, Calif., she moderated many of the message boards on which genealogists shared tips and gossip and reports of their setbacks and breakthroughs. She traveled around the country speaking at seminars put on by regional genealogical institutes. By chance, one talk this August was in Snohomish County, near where Jay Cook’s body had been found three decades ago. Her presentation, dense with technical detail, made no mention of the case or her law enforcement work. “DNA cannot often fully answer a question, but it can point us in the right direction,” she told her rapt, gray-haired audience. Afterward, she was mobbed.

Moore is petite, with long, wavy blond hair and big eyes of a not un-Barbie-like blue. Fluent in the vocabularies of both genetics and genealogy, she moves effortlessly between discussions of centimorgans (a measure of genetic linkage) and conversos (medieval Jews who converted under duress to Christianity, leaving their descendants to be surprised by DNA results suggesting Jewish heritage). She has a savant’s recall for the intricacies of the many, many family trees she’s built. Over coffee at a minimalist cafe near the high school where the conference was being held, she talked about the comfort she had originally found in genealogy. “I like learning that we’re really not that different,” she said. As she had built her own tree, Moore recalled, she kept finding evidence of infidelity: children whose genetics bore no trace of their supposed fathers (“nonpaternity events” is the term of art). “Our ancestors were so much like us,” she said. Discovering their secrets had brought them closer to her.

As DNA testing boomed, people started reaching out to Moore to ask whether DNA results were reliable. It was only one type of data, but for many it was proving troublingly at odds with what they thought they knew about themselves: The tests gave people new identities, but they also shattered old ones. A genealogist friend of Moore’s confided that she was going to have to throw out half of her painstakingly constructed family tree—a consumer DNA test had revealed that the man she thought was her father, in fact, wasn’t. Moore began to hear more and more versions of this story. Adult children would buy a parent a DNA test for a birthday or just to burnish Dad’s Italian American bona fides, and the results would instead drag up long-buried family secrets: out-of-wedlock births, secret adoptions. “Consumer DNA testing is uncovering all kinds of secrets,” she says. “Not just infidelities and misattributed paternities, but other, darker things.” Moore has worked several cases of babies switched in the maternity ward. She has found many instances of incest.

Moore quickly gained a reputation for skill and discretion. People with particularly difficult or improbable-sounding situations approached her: foundlings searching for the parents who abandoned them, amnesiacs trying to piece together their identities. She became an expert in connecting adoptees to their birth parents. In one notorious case, a shocked family reached out when they discovered that their daughter, conceived at a Salt Lake City fertility clinic, wasn’t actually related to her ostensible father—Moore discovered that an employee of the clinic had used his own sperm to fertilize the egg as well as untold others. Stories such as these, with their mix of mystery, identity, and Gothic melodrama, were irresistible to the media, and Moore’s skills as a sleuth, combined with her trade-show-honed presentation skills, made her a natural spokeswoman. She was hired as the genealogist for the PBS show Finding Your Roots, creating the family trees that Harvard professor Henry Louis Gates Jr. would present to his celebrity guests. She also became a mainstay on pulpier shows such as 20/20 and Dr. Oz, where she would be introduced as a “DNA detective.”

It wasn’t long before actual detectives came calling. Investigators with active serial killers and coroners with long-unidentified bodies would email. They had DNA, they would say; could she help? At the time police in some European countries had already begun to use a form of familial DNA searching to find suspects. In the U.S., a genealogist and former physicist named Colleen Fitzpatrick had begun to use Y-chromosome-based techniques to help identify both murder victims and suspects. Moore herself had worked with law enforcement on some of her foundling and amnesiac cases. She knew her autosomal DNA testing methods would theoretically work just as well for finding an unknown murderer as for finding an unknown parent, and she had done the latter dozens of times. “I knew for a very long time that the potential was there,” she says.

In 2012, Moore went to 23andMe—she had a good relationship with the company from when she helped test some of its kinship tools—and tried to get its help with a cold-case homicide. She had a series of conversations with the general counsel there, but the company said no. (Asked about the episode, 23andMe said it couldn’t confirm it, but gave this statement: “We treat law enforcement inquiries, such as a valid subpoena or court order, with the utmost seriousness. We use all legal measures to resist any and all requests in order to protect our customers’ privacy. To date, we have successfully challenged these requests and have not released any information to law enforcement.”) Moore thought GEDmatch could work—the open source site allowed anyone to upload data—but that it was still too small a pool to be promising for these kinds of searches. So she would tell the police she couldn’t help them. “But I felt terrible about it,” she says. “I definitely had sleepless nights.”

Then, a couple years ago, Moore started hearing about a data analysis company that was going to genealogy conferences and offering to pay for people’s DNA tests if they would share their data. It was Parabon. As it turned out, the U.S. Department of Defense had grown interested in using familial DNA to identify the bodies of war dead, and Parabon had won a contract to figure out how.

The company hadn’t originally been interested in genetics. Armentrout is a computer scientist who’d worked both in finance and at the CIA before founding Parabon in 1999. His original product was software that borrowed the unused processing power of thousands of desktop computers sitting idle in homes or offices, then coordinated them into a network. It was a tool for tackling supercomputer-scale problems such as climate modeling or how proteins fold—calculations that involve “some sort of combinatorial explosion,” as Armentrout puts it. After a few years, Parabon starting designing applications of its own to use the tool. Working with DNA fit the bill: With 3 billion nucleotide base pairs in the chromosomes in each human cell, genetics is full of combinatorial explosions.

Parabon is based outside Washington, in one of the office parks sprawled around Dulles International Airport. Its first foray into forensics was a 2011 Pentagon contract to see whether it was possible to predict a person’s physical appearance—initially just eye, hair, and skin color—from a DNA sample. Armentrout was given to understand that it was to help identify the makers of improvised explosive devices. He attacked the problem with the standard machine-learning playbook, creating a giant data set of genetic and physical information and using his software to find the correlations. “It’s trying to get algorithms to tease out relationships that are difficult for humans to tease out and then building models on top of that,” he says. That work would lead to the company’s first law enforcement offering, a DNA phenotyping service called Snapshot, which provided a digital sketch—coloring but also facial structure and ethnicity—based on crime-scene DNA.

Eventually, Moore, who took it upon herself to be up-to-date on all things genetic genealogy, reached out to Parabon. She quickly bonded with Armentrout over their shared interest in genetics and the anonymous dead. By this point she had decided that identifying Jane and John Doe homicide victims was a less contentious test case for forensic genetic genealogy than identifying suspects—no one would get sent to prison based on the work. Other citizen scientist genealogists had reached a similar conclusion: Fitzpatrick had co-founded a nonprofit called the DNA Doe Project to do just that. Moore and Parabon were in talks about how they might collaborate when the Golden State Killer arrest was announced in April. Without notifying GEDmatch’s administrators, the investigators had uploaded the crime-scene DNA file to the site.

The blaze of publicity engulfed the field in ethical arguments. “I am not at all sure how I feel about my DNA potentially being used in a criminal investigation on the other side of the Atlantic,” the English genealogist Debbie Kennett wrote in MIT Technology Review. “All of us have the right to decide how our DNA and our own data is being used. The police force in a foreign country should not have the right to make that decision for me.” Others in the community, however, saw the arrest as evidence of genealogy’s power to do good. Moore came down on this side. Just as important, she argued, the wall-to-wall coverage of how DeAngelo had been identified had put everyone on notice: Within two weeks, GEDmatch users would be greeted by a message clearly stating that law enforcement was using the database. Moore, for her part, decided she didn’t need to wait any longer to wade into the new field. Three days after the Golden State Killer announcement, she was at work for Parabon on Detective Scharf’s cold case.

When she’s teaching, Moore likes to talk about the importance of building “speculative trees”—quick and dirty pedigrees sketching out how the various puzzle pieces fit together. The ratios themselves can tell you only so much: Someone who shares about a quarter of your genes might be your grandparent, your half-sibling, your niece, or your aunt. So you make a guess and then you iterate, moving people around and progressing from the most likely to the least likely possibilities, until things begin to solidify. For someone with her experience, the ratios become a sort of muscle memory, the way a professional poker player intuits odds.

It took Moore less than a day to find the suspect in the Snohomish case. The day after Scharf talked to the company, a Parabon researcher uploaded the DNA file to GEDmatch. The site runs those files through a relative-finder algorithm of its own, suggesting users who might be related to each other based on the ratios of shared autosomal DNA. The next morning, a Saturday, Moore woke up to the results and took her laptop to her couch. GEDmatch had found two users on the site who appeared to be related to whoever had left his DNA at the crime site 31 years earlier. They were good matches, too. The amount of shared DNA suggested that each was a second cousin of the suspect.

Moore was especially excited to see that the matches didn’t share stretches of DNA between them: They were related to the suspect but not to each other. This meant that each cousin was from a different branch of the suspect’s family; if Moore could find where the two matches’ family trees intersected, she would have arrived at the suspect’s immediate family. Second cousins, by definition, share great-grandparents, so Moore set about building the family trees of each of the suspect’s cousins back three generations. When done, she started to come forward in time, mapping all the descendants of all the great-grandparent pairs, looking for something—probably a marriage—that would connect their two clans. It was straightforward work, except for one spot where the DNA ratios and the genealogical record didn’t at first agree—what appeared to be a nonpaternity event that turned out to be a remarriage and an adoption.

Within hours, Moore had found a newspaper obituary for a woman who was descended from one family and carried a last name from the other. That was where the two families joined. The suspect had to be one of their children. They’d had four, but only one son. By Saturday afternoon, Moore had his name, and she wrote Armentrout telling him to give her a call.

She spent the rest of the weekend trying out alternative hypotheses, but she kept arriving at the same place. On Monday they gave Scharf the name—the detective was off that day, and out back feeding his horses when he saw Armentrout’s email. Two and a half weeks later, on May 17, police arrested William Earl Talbott II, a 55-year-old truck driver. At the time of the murders he’d been living with his parents in Woodinville, Wash., only seven miles from where Cook’s body was found, and he still lived nearby. After Moore had found his name, police placed him under surveillance, and when a paper coffee cup fell out of his truck at a traffic light, they tested the saliva on it. It matched the DNA from the crime scene.

Armentrout flew out to Washington for the press conference to announce the arrest. He stood on the dais beside Scharf, the sheriffs, and the victims’ still heartbroken relatives. “Like most of you, I feel a range of emotions about the events that led us here today,” he said in a tone of voice that betrayed none of them. “I’m angry when I think about the crimes. I’m proud that my company, Parabon, was able to assist in the investigation. I’m sad for the families of the victims, for their pain, for their loss, and hopeful that they will soon find some peace.” From her home, Moore appeared on a video screen behind him, her magnified face looming over the stage. Genetic genealogy, Armentrout said, was the future of criminal investigations. “Most people will applaud the use of these technologies,” he said. “Some will not.”

Most of Parabon’s cases have proved more difficult than Snohomish. In some, Moore has been able to narrow the search to a list of names rather than just one; in others, GEDmatch hasn’t found any usable genetic matches at all. But the process will only get easier as the number of people offering up their data grows. Some users took their data off GEDmatch right after the Golden State Killer announcement, presumably angered about the way the database was now being used, but others joined for the first time, presumably willing to be a potential genealogical link in the chain leading to a criminal. Curtis Rogers, GEDmatch’s octogenarian co-founder, says the reservations he originally had about police using his database have eased with each additional arrest. “We’re not violating privacy, I’m absolutely convinced of that,” he says. “We may be violating people’s expectations, even though we have made it clear matches might be used in some other way than genealogy.” He stressed that an individual’s actual genetic code isn’t visible on GEDmatch, just who he or she is related to. Moore and Armentrout both promise that more arrests are forthcoming; indeed, there has seemed to be a new one every week. Barbara Rae-Venter, the initially anonymous genealogist who helped find the Golden State Killer suspect, is working on further cases, too. And law enforcement agencies are fielding calls from other genealogists offering their services.

But the history of forensic science is full of infallible techniques that eventually proved fallible. Ballistics, polygraphy, blood splatter and burn pattern analysis, fingerprinting: All have been revealed to be vulnerable to varying degrees to the human biases they were meant to inoculate against. Even DNA, supposedly airtight in its ability to place someone at the scene of the crime, can mislead. Genetic genealogy has yet to be challenged in a court of law, but similar techniques have already led police astray in at least one case. In 2015, Idaho police, looking for leads in a decades-old murder, combed through a Y and mtDNA consumer database owned by Ancestry.com. They landed on a man named Michael Usry, whose Y chromosome matched 34 of 35 markers in the crime-scene DNA. The match was close enough to suggest that even though Usry wasn’t a suspect, someone closely related to him might be. That led police to Usry’s son, also named Michael. The younger Usry was eventually cleared—his DNA wasn’t a match, either—but not before he was interrogated and publicly identified. The Y and mtDNA database has since been shut down.

For privacy advocates, DNA’s newfound ability to implicate relatives of ours we may never have met (or to implicate us through our relatives) is only one of the many things we don’t realize we’re exposing ourselves to. Recent revelations about the access Facebook Inc.’s advertisers had to users’ personal content are a case study in what happens when private information is fed into the relentless maw of monetization, and genetic information has the potential to be far more valuable. “We’re still learning what exactly the genome can tell us and with what certainty it tells us those things,” says Erin Murphy, a law professor at New York University and author of Inside the Cell: The Dark Side of Forensic DNA. A working paper recently posted online by computer scientists at the University of Washington explains how one could create fake GEDmatch profiles and use them to blackmail people or fool criminal investigators—the study’s authors conclude that companies might mitigate these risks by creating ways to authenticate DNA data. The recent Science paper on genetic exposure made a similar recommendation. (In a letter written in response to that paper, Parabon and Moore describe it as “misleading” and based on faulty premises. “[D]etermining an individual’s identity” from a match in GEDmatch, they counter, “is extraordinarily complex.”)

Several states have passed strict laws governing familial searches in the databases law enforcement builds of criminals’ DNA; Maryland and Washington, D.C., have forbidden it altogether. “When you look at the early stories on the Golden State Killer investigation, there were advocates saying the practical limitations of all this genetic sleuthing will protect from too much use of this technique, but then Parabon uploaded hundreds of crime-scene samples,” says Stephen Mercer, the former Maryland public defender who helped get his state to pass its ban on familial searches in criminal databases. Once the power of Moore’s techniques is clear, he fears, the temptation to use them for lesser transgressions such as trespassing will be overwhelming. Others have raised the specter that, in a post-Roe v. Wade world, they might be used to identify women who’ve had illegal abortions. At present, the only thing stopping Parabon or anyone else from using genetic genealogy to investigate other kinds of crimes is the GEDmatch terms of service.

Moore has heard these concerns. Her response is that she’s just pointing police in the right direction. At the very least, they have to go and get a DNA sample from the suspect to ensure that it matches. “You can’t arrest somebody based on me saying their DNA looks like the right fit,” she says. “It’s a really highly scientific tip, and then they have to build their case. Which is why it’s a misnomer when people say I solve these cases. I didn’t solve these cases.” Scharf agrees. As a cold-case detective, he is a connoisseur of tips. Moore’s work “is the most powerful kind of tip you can get,” he says, “but it’s not evidence.”

In general, Moore tends to worry less about privacy than about secrecy. She believes in revelation, even if that means exposure. In her unknown parentage cases—even the foundling ones—the parents she tracks down usually end up embracing their previously unacknowledged children. “We’ve always had some kind of happy ending,” she says. Even if the parents themselves didn’t want to connect with their children, a sister did, or an aunt. “There’s always something really amazingly positive that comes out of these stories.”

It’s certainly true that widespread DNA testing has forced secrets into the open. People shocked by the results of genetic tests now gather in “nonexpected parent” Facebook groups, some with more than a thousand members. At 23andMe, customer-support staffers sometimes have to play the role of comforting friend or therapist. “You may learn information about yourself that you do not anticipate,” the company wrote in an early version of its terms of service. “This information may evoke strong emotions and has the potential to alter your life and worldview.”

Moore prides herself on her ability to keep confidences, but all in all, she’d rather live in a world with fewer of them. The threat of detection, she suggests, might dissuade future murderers, rapists, and con artists, perhaps even philanderers.

“I’ve seen all kinds of secrets come out, and I’ve been shocked to learn what our society is really about, what people are really doing,” she says. “I don’t really see it as potentially bad that things are out there.”

To contact the editor responsible for this story: Daniel Ferrara at dferrara5@bloomberg.net

©2018 Bloomberg L.P.