O Brother, Where Art Thou?

It’s time for legislators to look more closely at familial searches of DNA databases.

Computerized DNA configurations

The U.S. forensic DNA database has expanded rapidly in recent years. While it was originally authorized to store the DNA profiles of only convicted violent felons, the FBI Combined DNA Index System (CODIS) now includes all federal offenders—including arrestees not yet convicted of any crime—as well as convicts from all 50 states and arrestees from many. Such expansions of the database are troubling, but at least they are explicit. More worrisome is the effective inclusion of many innocent individuals in the database, via novel and almost completely unregulated search techniques called “partial matching” and “familial searching.” By adopting one or both of these search techniques, some states are quietly expanding database coverage to “virtually” include the innocent relatives of profiled offenders—nearly always without any legislative oversight.

Familial searching and partial matching both exploit the same well-known underlying principle: Close relatives are genetically similar. Parents, children, and siblings share, on average, at least half of their DNA. Not surprisingly, similar DNA generates similar DNA profiles, which are the stripped-down numerical records stored in DNA databases. So, even if a crime-scene sample doesn’t exactly match any existing offender profile in CODIS, police may still find a partial match—an incomplete DNA match between the forensic evidence and a known offender. If this happens, police know the offender who partially matches the evidence did not himself leave the sample at the crime scene, but—and this is where it gets interesting—it’s very possible one of his relatives did. After screening those relatives with follow-up DNA testing, police may have a new lead.

This expanded process fundamentally alters the population that can be searched with DNA databases. Previously, police only matched crime-scene samples against known, profiled offenders—those who had been actually included in CODIS because of a prior arrest or conviction. (So a crime-scene sample left by a first-time offender would turn up no matches in the database.) But with these new techniques, relatives are effectively included in the database through their genetic similarity to a profiled offender. Think of it this way: If you’ve never been arrested, your DNA profile shouldn’t be in CODIS. But if your brother has been arrested and is profiled in CODIS, then whenever these new searches are used, you, too, may be searchable—and targeted for investigation—through the database. These searches render offenders’ relatives effectively searchable in CODIS, even though the relatives themselves have never been officially included.

Technically, DNA-database matches that implicate offenders’ family members can surface in one of two ways: through either “partial matching” or “familial searching.” This distinction is somewhat misleading, since these search techniques differ chiefly in intent. If police deliberately set out to search the database for relatives—usually using specialized software—the technique is called “familial search.” If the same partial hits instead arise inadvertently during a normal database search, they are referred to as “partial matches.” Of course, the underlying value of both techniques is the same—a partial match between two DNA profiles may suggest a blood relationship.

These leads sometimes bear fruit: In the United Kingdom, for instance, several high-profile cases have been solved this way. But however elegant these techniques may seem, we should think long and hard before allowing the effective expansion of our DNA databases to include the relatives of profiled offenders. Using partial matches—whether deliberately sought or not—is problematic for several reasons.

First, these matches may not be legal. Courts have held that convicts and arrestees have a lower expectation of privacy than ordinary citizens and that the state has a weighty interest in identifying criminals and preventing recidivism. On balance, courts have determined, these factors permit offenders’ DNA profiles to be stored in CODIS. But none of these reasons justifies collecting information about offenders’ kin. Directly including the DNA of innocent family members in CODIS would be unlawful under existing statutes and likely prohibited under current Fourth Amendment jurisprudence. And if actually including them in the database is prohibited, shouldn’t their effective inclusion be so, too?

The police may argue that familial searching and partial matching present no Fourth Amendment problems, since nothing has actually been seized from the relatives and they have not been personally searched. In a sense, they are correct—the family members aren’t actually in CODIS—but they are nonetheless “reachable” through a profiled relative. Their inclusion is virtual, a product of biological happenstance. But this end-run shouldn’t satisfy lawmakers: If offenders’ relatives are to be treated like offenders, there must be a good reason to single them out. There is none.

Second, partial matching and familial searching will greatly aggravate the racial inequality already embedded in offender-based DNA databases. Certain racial and ethnic populations are already overrepresented in CODIS, owing simply to the reality of crime statistics. But implicitly expanding database coverage to include relatives will grossly and unfairly amplify this bias. By applying partial matching and familial searching to a database that includes anyone ever arrested—which seems to be where the system is headed—CODIS could one day approach universal coverage for some races and not for others.

Partial matching and familial searching are clearly problematic in their own right. In practice, they are doubly worrisome because many states are quietly implementing such policies without any legislative oversight. California and Colorado are known to use familial search, and these states have at least publicly announced their new policies. But new survey results reveal that many other states also have familial-search or partial-match policies that went unannounced and were never even publicly debated. Most of these policies exist only in internal laboratory manuals, if they are written down at all. Nebraska, for instance, authorizes familial search in an internal lab manual. This policy decision evidently occurred without public discourse and has thus drawn virtually no public attention.

In fact, no state has specifically authorized familial search through legislation. Only Maryland has addressed familial search by statute—and it banned the practice. State legislators need to address these aggressive search regimes right away, both to provide guidance to state crime labs and to subject these important policies to proper legislative oversight and debate.

Moving beyond deliberate familial searching, at least 16 states have now approved the use of fortuitous partial matches in some circumstances, with a 17th—New York—poised to join this group. (Just 11 states have so far rejected partial matching.) Most of these states have drawn a line permitting the use of partial matches while at the same time prohibiting deliberate familial searching. These states presumably seek to have it both ways, occasionally exploiting the crime-solving potential of partial matching while limiting the legal pitfalls—or perhaps the bad publicity—associated with full-on deliberate familial searches.

But as we stated earlier, this distinction between the two types of searches is misleading. Both fortuitous partial matches and deliberate familial searches rely on incomplete matches to stored DNA profiles; both invade the privacy of offenders’ otherwise non-data-based relatives; and both put greater percentages of minority citizens under perpetual genetic suspicion. Every DNA-database search raises the possibility of finding a partial match, and so every database search accomplishes implicit inclusion regardless of whether, how often, or how deliberately such matches are found. Worst of all, endorsing only fortuitous partial matches may encourage state labs to loosen the requirements for “routine” database searches, so that more partial matches will conveniently crop up. States should be clear and consistent in their policies—and the line between fortuitous partial matches and deliberate familial search is neither. States should either permit partial matches to be used or forbid it. Whether these matches arise accidentally or deliberately shouldn’t matter.

The survey also reveals that at least a dozen states operate with entirely unwritten policies, including four of the 16 states that have embraced fortuitous partial matches. New Mexico, for instance, purposely decided on a policy concerning the use of partial matches but specifically chose to leave it unwritten. And even when states do commit their policies to writing, they can be difficult to access. In at least 14 states—including the aforementioned Nebraska—policies appear only in internal laboratory manuals. Upon request, some of these states simply declined to provide copies of their partial matching policy. Other state labs were unwilling to share copies of the relevant written policies absent a formal request or a request under the state’s freedom of information act. One state, Georgia, which was otherwise forthcoming and informative, declined to release a copy of the relevant policy on grounds that the lab documents are subject to outside copyright. Another state declined to say anything about its approach to partial matches, expressing concern that information about the work of the state laboratory might become public at all.

Very little is publicly known about most state practices concerning partial matching and familial searching in DNA databases. But it’s clear that policymaking has occurred, and it has largely taken place behind closed laboratory doors, with little or no public knowledge. These policies are significant. They can make innocent people investigatory targets simply by virtue of their genetic relationship with a profiled offender. They effectively expand CODIS coverage beyond the established legal limits. This is not the ordinary course of genetic identification or the traditional use of DNA databases. Partial-matching and familial-searching policies, whatever you think of them, are too serious to be left to internal laboratory manuals or wholly unwritten policies. These techniques represent major, often silent expansions of DNA-database coverage, and legislators—and the public—must address them directly.

Like  Slate on Facebook. Follow us  on Twitter.