Growing up, I wasn’t allowed to go places alone unless I followed certain rules. I still remember the day I broke them. 

I was about 5 years old when I walked around the corner to play at my friend’s house. My mom had told me to call as soon as I arrived. But I was excited and forgot.

Soon my mom was at the door. The look on her face said I was in trouble. She took me straight home and said I couldn’t go back to play that day. I was never to forget to call home again.

My mom had a reason to be worried. My neighborhood seemed idyllic, a tony section of Tacoma, Washington, filled with historic homes affording views of a sparkling bay and Mount Rainier. But despite the veneer of tranquility, a killer was at large.

In 1986, about two years before the day I forgot to call home, a 12-year-old girl named Michella Welch had been abducted, assaulted and murdered in a maple-filled gulch next to a playground called Puget Park. I’ve read the court papers that describe how she was found. They can’t be unread.

The crime took place just a few blocks from the house my family moved into a year later. The police investigation eventually went cold, and whoever did it was still out there. My parents, just like the parents of Michella’s friends and neighbors, feared someone would swoop down and take their daughter away forever. 

“It was terrifying,” my mom explained to me recently.

As a child, I was too young to understand how scary the crime was. It faded into the background of my carefree childhood, along with stories of other crimes and urban legends. But the sense of danger it created for my parents finally clicked for me 30 years later, when a new investigative technique identified a suspect in the Michella Welch case and led to an arrest.

Using results from the GEDmatch genetic database and public records, social media and math, investigators identified Gary C. Hartman as Michella’s killer. On June 20, 2018, Tacoma police arrested Hartman, now 67, on charges of rape and homicide. Hartman, who pleaded not guilty and intends to go to trial, lived less than two miles from Puget Park in 1986. 

The technique, pioneered by genealogists including CeCe Moore and Barbara Rae-Venter, is part of a game-changing field in police work. Called genetic genealogy, the approach starts by connecting DNA found at a crime scene with that of distant relatives who’ve created public profiles on the GEDmatch database. Experts use this information to construct a family tree of connected relatives, tracing the genetic relationships to locate a suspect. Long used to help connect family members, it’s now being used to help solve crimes.

Moore and her team at Parabon NanoLabs have enabled police to make progress in more than 50 cold cases involving murders or rapes, including Michella’s. The oldest was the 1967 murder of a police aide in Seattle. One of the most recent was in 2018, when a suspect in a rape was identified three months after the crime took place in Utah. Moore, who studied music and theater in college, began investigating cold cases in 2018 after years of honing her skills finding the relatives of adopted people.

Parabon NanoLabs is a forensics company that connects Moore to cases and then vets her results. Other genetic genealogists are working directly with police departments, like Rae-Venter, whose work identified a suspect in the Golden State Killer case and led to the arrest of Joseph James DeAngelo on multiple charges of murder and kidnapping.

Genetic genealogists are putting a dent in the high number of unsolved murders in the US, though they still have their work cut out for them. A 2010 investigation by the Scripps Howard News Service found at least 6,000 murder cases go unsolved per year. As part of the Murder Accountability Project, the same reporters found more than 300,000 unsolved murders in the US since 1965. In 2017, the most recent year with data available, 7,154 murders went unsolved. The number of rape cases that go unsolved each year is even higher, said Joe Giacalone, a professor at John Jay College of Criminal Justice and former NYPD cold case detective. 

The promise of genetic genealogy will get a lot of families’ hopes up, but not every police department will be able to conduct such a resource-intensive investigation, Giacalone says.

“It will matter to family members,” Giacalone said, but “it’s going to be very slow.”

Enthusiasm for the technique is tempered by the questions it raises about privacy. Bioethicists, civil liberties watchdogs and Moore herself have pointed out ways genetic genealogy could go wrong. 

Court cases will be the test of how viable the technique is when a suspect pleads not guilty. On Friday, a jury convicted William Earl Talbott II of two counts of aggravated murder in what’s believed to be the first trial of a suspect identified through genetic genealogy. Talbott had pleaded not guilty in the 1987 murders of Tanya Van Cuylenborg and Jay Cook in the state of Washington. Out of the suspects who were still alive years after their alleged crimes, at least five have pleaded not guilty.

At its core, genetic genealogy has the potential to become a perfect crime-solving tool that can identify suspects even if they’ve never had their DNA analyzed before they fall under suspicion. Privacy experts call this a networked privacy decision. When you share your genetic information with a service like GEDmatch, you’re making that decision for everyone you share genes with. Most of them are strangers.

Given how new the technique is to law enforcement, public debate has been limited. Should it be used for all crimes? Or just the most vile? What’s more, there’s currently no academic training or certification program available for genetic genealogists, who don’t have formal rules for their profession.

“I have very high hopes for this to make society safer. Maybe it will work as a deterrent once people realize a lot of cold cases are being resolved,” Moore said in an interview. “But for us to get there, everyone needs to be very careful.”

The DNA clearinghouse

Moore has researched dozens of cases since the Tacoma Police Department announced the break in the Michella Welch case. She always starts at GEDmatch. 

The database, run by genetic genealogy enthusiast Curtis Rogers and his co-founder John Olson, serves as a kind of clearinghouse. People can upload their results from FamilyTreeDNA, Ancestry, MyHeritage and most other genetic testing services. By assembling results from the different DNA services, Rogers hoped people would find more relatives and dive deeper into their ancestries. 

GEDmatch contains DNA data, uploaded as “kits,” on about 1.2 million individuals, and is the most important tool currently available for genetic genealogy. Moore and her team used it to find distant relations of the person who stabbed 18-year-old Michelle Martinko to death, leaving her in the family car in an Iowa mall parking lot. First, Moore created a profile using DNA the presumed attacker left behind at the crime scene. Then she looked for a match on the site. 

GEDmatch only lets investigators see matches to users who’ve made their profiles public, like Brandy Jennings. Jennings, who told reporters that she’d uploaded her information to GEDmatch and forgotten about it, came up as a distant cousin of the suspect. Further investigation led to the arrest of Jerry Lynn Burns, who has pleaded not guilty. Burns’ lawyer declined to comment.

Rogers initially opted all users into police investigations as part of the site’s terms of service. But concern over how the data was being used prompted him to reconsider. So in May, he opted everyone out as a default. That effectively reset the number of people whose data was available for police investigations to zero. Users have opted back in about 85,000 of the site’s more than 1 million kits so far. 

“That was a difficult thing to do,” Rogers said of the decision. “But I think the most ethical thing to do.”

The family tree

The distant relatives found on GEDmatch are just a starting point. After genetic genealogists find those names, they take off their genetics hat and switch into genealogy mode. 

When investigators searched for the person who in 2007 assaulted and killed 39-year-old Jodine Serrin in Carlsbad, California, they built a family tree. In all genetic genealogy investigations, the family tree has to connect suspects to the people they’re related to on GEDmatch. Investigators use public records like obituaries, census documents and marriage certificates, as well as social media accounts.

If a suspect has a second cousin on GEDmatch, investigators need to go back in time to identify all eight great-grandparents. Then they move forward in time again and identify all the descendants of those great-grandparents. 

This gives them a list of the names of the GEDmatch user’s second cousins. After further research, it becomes a list of suspects. When a suspect has more than one distant cousin on GEDmatch, investigators must build out family trees for all of them.

The last step is often deduction, Moore said. That typically means identifying the distant cousins who were in the right place at the right time to commit the crime. 

In the Serrin case, genetic genealogists identified David Mabrito as a suspect. Mabrito, who died in 2011, was homeless at the time of the crime, but police told reporters he had family in the Carlsbad area. Police in another jurisdiction had Mabrito’s DNA from a separate investigation, according to news reports. When Carlsbad police compared it to the DNA from the crime scene in Serrin’s case, it matched.

The police investigation

Genetic genealogy has its limits. Police have to collect DNA from the suspects Moore and her team identify to confirm they’re the right people. And they frequently can’t narrow their findings down to one suspect. 

Police tail suspects and find a discarded item, such as a napkin, cigarette or straw, to collect a DNA sample. Sometimes they’ll have to surveil several people at once. 

That’s what happened in Moore’s investigation into the assault and murder of 18-year-old Angie Dodge in Idaho Falls, Idaho. She and her team narrowed the suspects down to a group of about half a dozen related men. Initial DNA results came back negative, until they realized there was one more man in the group, Brian Leigh Dripps Sr. 

Moore learned Dripps was related to the other men but had the last name of his stepfather, not his biological father. Dripps’ DNA matched evidence found at the crime scene and he confessed to the crime, according to Idaho Falls police. His attorney declined to comment.

The surprise benefit

The technique is especially good for catching a kind of suspect who often eludes investigation tools: white men. 

White men are believed to be underrepresented in CODIS, the FBI’s collection of DNA samples pulled from crime scenes, arrestees and criminal suspects. In GEDmatch, however, white people are overrepresented. 

If you’re of European ancestry, the probability you have a third cousin in a database of GEDmatch’s size approaches 90 percent, according to research from genetics experts. The website’s recent decision to require all users who want to be involved in criminal investigations to opt back in has reduced that probability a lot. GEDmatch’s Rogers is optimistic users are going to keep opting back in.

“This is giving law enforcement access to this category of individuals that, prior to this, was immune to surveillance,” said Malia Fullerton, a bioethicist at the University of Washington. 

The genes that bind us

When Moore uploads the genes of a criminal suspect to GEDmatch, she uses an analysis of their autosomal DNA. Autosomal DNA is found in all of your chromosomes except the X or Y chromosome (those determine sex). It allows investigators to match DNA to distant cousins from all branches of your family.

That marks a shift in criminal investigations, because it relies on much more genetic information than police typically work with. 

While law enforcement has long collected DNA samples from suspects, they have typically only shared a stripped-down version of the DNA with CODIS. Called a DNA fingerprint, the information is reduced to the bits of meaningless genetic cruft that can be found in between genes in everyone’s DNA. 

That takes out most of the personal information your genes can reveal about you, like probable eye color, susceptibility to genetic diseases or ethnic origin. To find distant cousins on GEDmatch, genetic genealogists like Moore have to leave that information in.

Given the huge amount of personal data in DNA, privacy and criminal justice experts anticipate constitutional challenges to how investigators collect and use the information in the cases that genetic genealogists are cracking open.

The question of where to stop

Civil libertarians are concerned that genetic genealogy could be used to help solve any crime in which DNA is left behind. There’s a real chance it could be used to stop petty crimes, and the public needs to decide how far that should go, said Vera Eidelman, an attorney at the ACLU who specializes in speech, technology and privacy. 

“It is a worthwhile and important thing to solve these sorts of crimes,” Eidelman said of the murders and rapes that have been investigated so far, “But once this becomes an available resource, who’s to say it won’t be used on every crime?”

This is closer to happening than you might think. Something very similar already happens with another tool, which investigators call familial search. It’s used in Colorado, California and several other states, where investigators turn to it to find near-genetic matches in databases of DNA collected by law enforcement. For example, the DNA could be a near match with a convicted criminal. Then police focus their investigation on the convicted criminals’ relatives.

Familial search isn’t limited to murder and rape investigations. Denver police used the technique to find a suspect who had allegedly left a drop of blood behind when breaking into a car and stealing less than $2 in change. They identified him because they had his brother’s genetic information in a database.

Ashley Hall, a forensic scientist at University of Nebraska-Lincoln, says there’s a simple solution here.

“If you don’t want to be implicated,” Hall said, “don’t commit a crime.” 

Since GEDmatch first made headlines in the Golden State Killer cold case in 2018, the site’s terms of use have said law enforcement can use the site for investigations of violent crimes. Rogers, the site’s co-founder, said he hopes any regulation on genetic genealogy will mirror what GEDmatch is already doing.

The criminal suspect next door

The horror of the crimes being investigated makes a compelling case for using genetic genealogy. The technique is providing answers to victims and their families throughout the country. 

For the family of Michella Welch, her friends and the other families who grew up in Tacoma, the identification of a suspect after 30 years stirred a range of reactions. Michella’s mother told reporters she was surprised at how unobtrusive Hartman, the suspect, appeared at his arraignment in 2018. A childhood neighbor said the same thing. Somehow, you’d expect someone accused of something so evil to stand out in a crowd.

I found myself shocked to think how close he had lived to my family’s house in the 1980s. The local newspaper reported Hartman lived in the neighborhood until at least 1989, when we’d been living there for two years.

While thinking about this, I stared at a satellite view of my childhood neighborhood on Google Maps, trying to measure how close Hartman had lived to Puget Park at the time of the crime. There on the map, I suddenly saw something that made me shudder. 

I realized Hartman could have walked home from the crime scene while hidden by a two-mile-long strip of secluded woods that stretches along the Tacoma waterfront. I picked blackberries, looked for salamanders and went for long runs in those woods.

Hartman is innocent until proven guilty and, according to his lawyer, intends to defend himself against the charges on multiple fronts. His defense will likely include challenges based on privacy law. His lawyer doesn’t expect the trial to start until next year. 

If Hartman and many more of the suspects identified through genetic genealogy are found guilty, investigators will know they have a very powerful tool on their hands. 

The question will remain of how and when they should use it.