• Democratic Institutions and Politics
• International Relations, Science and Security
• Markets and Methods
• Management and Leadership
• Social Policy

David Lazer on DNA and Law Enforcement

The use of DNA in criminal investigations is now standard practice in the U.S. and many other countries. The national DNA database contains DNA samples from more than four million individuals, and has provided critical evidence in thousands of criminal cases, helping to both convict and clear suspects. David Lazer is associate professor of public policy and director of the Program on Networked Governance. He is author of "DNA and the Criminal Justice System."

Q: Discuss how the national DNA database has grown in stature, credibility and reliability as a law enforcement tool.

Lazer: The national DNA database system began in the late 1990's; the first state system began in 1989 in Virginia. The national system has been growing dramatically over the past six years or so-it has expanded around tenfold in size. And furthermore, what we've seen is an expansion in who goes into the database. Whereas five or six years ago most states just put violent criminals into these databases, most states now put all felons into their database. And a handful of states and the federal government are putting arrestees into the databases as well. States like California and Texas are putting arrestees into the database, and this has been part of what's been fueling the growth of the DNA database in the United States.

In the meantime the United Kingdom began their database sometime before the U.S. and has had a very aggressive approach to putting people into their database. And while the UK has about the same size database as the United States (4 million people), the UK has around one-sixth the population of the U.S. Around one in six men are in their database, while one in 30 or 35 men are in the database in the US. But if you go into certain states like Virginia, the ratio is much higher, around one in 10, and I think that is the direction in which the U.S. as a whole is headed as well.

Q: What are some of the most difficult challenges posed by the use of DNA in the criminal justice system?

Lazer: There are a number of important challenges. One is around the logistics of managing the information. The systems that have been put in place in the United States have really not been designed, after centuries of honing, to deal with these DNA hits coming out, say a year after an investigation has been started. So one big issue that has arisen the past year or so is that many of these hits are not being followed up on. Front page stories about DNA matches not having been followed up on in the Massachusetts State Crime lab have been a huge black eye for public safety in Massachusetts. People who work in state crime labs talk about how they've gotten a match, they've reported it, and then six months later they report a match with the same guy, and they wonder what ever happened in that first case, and they find out it was never followed up on. So that's a huge challenge.

Another challenge is around where to draw the line about who should be in the database and who shouldn't be in the database. Any time you impose something on individuals in a society there's a question about why you impose it upon some people but not others. And I think that one clear dividing line in the US has been if you're convicted of a crime then you might be subject to additional surveillance because in some sense you violated the social contract. However, in the UK they put people in there who haven't been convicted or suspected of anything. They've done DNA dragnets where they ask you to volunteer your DNA to exclude you from being involved in a particular crime, but then your DNA is in there for life. So it raises interesting questions as to why some people should be under lifetime surveillance and not others.

What I've been working on with colleagues on familial searching raises issues along the same lines. People are in the database, but then their close relatives can be identified because of the known statistical relationship between their DNA and that of their relatives. So if Joe is in the database, now you can identify Joe's brother as the perpetrator of the crime. This means you may effectively be incorporating close family members into the database as well. So that's a second big issue.

The third issue is around use of the data. Genetic data are different from other kinds of data you might collect on individuals because genetic data can reveal many things about you-like your propensity to get certain diseases-so the question is how do you constrain the use of the data and how do you want to constrain the use of the data? For example, you potentially could use the genetic samples from crime scenes to come up with profiles of what the likely race or appearance of the suspect is. So how do you want to use that information? That's a little different from other types of data we might gather about individuals.

Q: Tell us more about the issues you've been researching around familial searching.

Lazer: Right now there exists a database of hundreds of thousands or millions of individuals, and you have a sample from a crime scene which you think may come from the perpetrator of the crime. It's an unknown sample and you search it against the database, but you don't get a direct match.

Familial searching would involve a second layer of searching the database for near misses, which might be indicative of relationships. And the key question, vis-a-vis familial searching in terms of whether it is practical, is whether you can pick the needles out of the haystack. Can you find actual family relations in a database filled with people who are not related to the actual perpetrator of the crime, or the source of the sample from the crime scene? And in the paper I did with Frederick Bieber and Charles Brenner, we found that it was in fact statistically practical to look for near misses, rank order them from most likely to least likely to be related to the perpetrator, and you wouldn't have to look too deep in that haystack to find a relative. And that obviously sets in a whole set of implications around the ethics and desirability of using that information, and whether you want to design a system to facilitate that. And there's been actually substantial national debate around that issue since we wrote our article.

In particular the key ethical question is: if you have a standard, like we seem to have had in the U.S., that a person needs to have been convicted of a crime in order to go into the database for life-to be under lifetime genetic surveillance because you violated the social contract-does it follow that you should be in the database simply because you are related to someone who has been convicted of a crime? And furthermore, since people who are convicted of crimes don't come from a random slice of society, this will certainly have an uneven impact upon parts of our society. For example, of this class of people to be incorporated into the database, a disproportionate share of them will come from minority communities. So you'll have a set of people who are in the database who haven't been suspected of a crime or convicted of a crime, but will be in the database and will be disproportionately minorities. The policy question is: is this a desirable thing to do? The plus side-and a very powerful plus side-is that certainly some crimes will be resolved as a result of using this technique. We conservatively estimate that it would increase the number of 'hits'-matches you would get between individuals and crime scenes-by 40 percent. That is a really big increase and it is hard to say that you shouldn't pursue an investigative technique that is cheap and will resolve many cases. So that is the basic conundrum.

Q: How do you foresee future uses of DNA technology by law enforcement?

Lazer: There is a set of interesting issues about information that the government possesses about individuals because of the malleability of information. Information can be re-used in ways not conceived of and that would not have been approved of by the original designers of the information system. Let me give you an extreme example from the Netherlands in the 1930s. A population database was designed for reasons of administrative efficiency, and one of the bits of data they collected was religious identification of Dutch citizens. That was done for completely benign reasons. However, when the Netherlands was conquered by the Nazis, those data were then used to identify Dutch Jews and to exterminate the Dutch Jewish population. In fact, the Netherlands suffered the highest rate of elimination of its Jewish population of all of Western Europe, and I think in significant part because of the population database.

So the concern one might have, especially vis-a-vis genetic data, is that we don't know what we'll be able to find out from these data ten years from now, twenty years from now. Even something like familial searching hadn't really been contemplated by most lawmakers. Certainly some people in-the-know discussed this possibility at the beginning of the use of DNA in the criminal justice system in the early 1990s, but most state legislators, I'm sure, didn't think about this. There is no state legislation that I have seen in this country that has talked about familial searching, and has either prohibited familial searching or encouraged familial searching.

Q: There is a much larger public debate going on about privacy in general. And that brings pressure onto policy makers, and this is a piece of that discussion. How do you see privacy concerns weighing on the minds of policymakers as they consider the use of DNA and DNA databases in the criminal justice system?

Lazer: I have a demo that I do when I talk about privacy issues which shows a map of Cambridge and people walking around Cambridge based on their cell phone data. Our lives are increasingly transparent. In Boston we've shifted to a new way of paying for mass transit that involves a card with a device that you put into a reader. But my card, for example, is actually linked to my identity, which means I can be identified in the system during the times I am taking the bus or the subway.

There are increasing advances in face recognition systems. We can identify the movement of individuals through cameras located strategically through a city. Certainly in London there are an unbelievable amount of cameras. In New York there are more and more cameras photographing the movement of people. Right now the utility of those cameras is somewhere limited because of the limitations in face recognition, but all that is going to change in coming years with things like face and voice recognition becoming quite accurate. Of course, law enforcement can already tap into phone conversations, but it's not practical to do that on a large scale until you can digitize and analyze the content of those conversations. We can do that now, so we could in principle, from an informational point of view, record and track and manage the information of the phone calls of millions of people, and that just simply would not have been possible even in the most totalitarian of states a decade ago.

So the broader collective question here is: what informational capacities do we want to give the government? And the answer is, probably more than it's had because the reason it hasn't had certain capacities in years past is that it was technologically impossible. But just because something is technologically possible does not mean that it is desirable. So the question that confronts us is: where should we draw the line between the capacities of the government to get information about us, as compared to limiting the data the state can get, which at times will limit the capacity of the government to do good things? And obviously this is particularly acute in this age with concerns around terrorism-that there may be more and more pressing reasons for the government to track what individuals are doing. But there have to be certain limits.