What Part Of "Innocent Until Proven Guilty" Don't They Understand?

by hilzoy

From today's NYT:

"The Justice Department is completing rules to allow the collection of DNA from most people arrested or detained by federal authorities, a vast expansion of DNA gathering that will include hundreds of thousands of illegal immigrants, by far the largest group affected.

The new forensic DNA sampling was authorized by Congress in a little-noticed amendment to a January 2006 renewal of the Violence Against Women Act, which provides protections and assistance for victims of sexual crimes. The amendment permits DNA collecting from anyone under criminal arrest by federal authorities, and also from illegal immigrants detained by federal agents.

Over the last year, the Justice Department has been conducting an internal review and consulting with other agencies to prepare regulations to carry out the law.

The goal, justice officials said, is to make the practice of DNA sampling as routine as fingerprinting for anyone detained by federal agents, including illegal immigrants. Until now, federal authorities have taken DNA samples only from convicted felons."

(For those of you who like looking the laws up for yourselves as much as I do: the VAWA reauthorization is here (pdf); check out sec. 1003 (p. 126.) The statute it amends is here.)

There are a lot of problems with this. Let's get the simplest one out of the way first: it will cost a lot of money, and put a huge strain on the FBI, which might have more important things to worry about. From the NYT:

"Many groups warned that the measure would compound already severe backlogs in the F.B.I.’s DNA processing. Mr. Fram of the F.B.I. said there had been an enormous increase in the samples coming to the databank since it started to operate in 1998, but no new resources for the bureau’s laboratory. Currently about 150,000 DNA samples from convicted criminals are waiting to be processed and loaded into the national database, Mr. Fram said.

He said the laboratory had added robot technology to speed the processing. But in the “worst case scenario,” where the laboratory receives one million new samples a year, Mr. Fram said, “there is going to be a bottleneck.”"

That could, of course, be solved by providing more funding. Other problems, which I'll discuss below the fold, are more interesting and less tractable.

I should say, at the outset, that I am a big fan of the forensic use of DNA. Used right, it can allow the police to identify actual criminals more accurately, which is good in any number of ways. For one thing, the more accurately we can pin a crime on a particular person, the less likely we are to imprison the innocent, which is obviously a good thing. For another, every time we accurately identify a criminal and put him or her behind bars, we spare any victims of the crimes that person would have committed had s/he remained at large from being victimized.

Moreover, if used correctly, DNA can to some extent correct for any biasses police or prosecutors might have. Suppose a rape or robbery is committed, and the police pick up someone who looks, to them, like the sort of person who might have committed that crime. DNA evidence, if it's available, can be used to prove that person's innocence; and unlike witnesses, DNA has no biasses, and thus cannot share any biasses that the police or prosecutors might have.

I am willing to take correction on this point from anyone, but I do not have a problem with keeping a database of DNA profiles of people who have been convicted of crimes. Nor do I think, as some people seem to, that it should be collected only from perpetrators of violent crimes. (This is, I think, an artifact of the days when DNA evidence was likely to be available in usable form only from crimes involving sex or blood; iirc, the original DNA databases were exclusively for sex offenders.) I see no reason why people who embezzle money from pension funds, or who defraud widows and orphans, shouldn't have their DNA on file the way burglars and rapists do.

Nor do I have a problem with collecting DNA samples from people who have been arrested. It is common practice, as I understand it, to collect DNA from all sorts of people who are involved in crimes -- for instance, from the victim, the police officers involved, etc. -- in order to exclude their DNA from any DNA evidence that might be around, and I can see no reason not to collect samples from arrestees as well, if one has either their consent or a warrant.

What's problematic, I think, is not collecting those samples from people who have been arrested, but keeping their genetic information in a database. What's wrong with it is that it is fundamentally unfair, and unfair in ways that magnify existing injustices in the criminal justice system.

Suppose, just for the sake of argument, that while many police officers are completely unbiassed, and only make arrests when the evidence supports them, some police officers are biassed against, oh, let's say African-Americans, and arrest them more often than the evidence against them would warrant. If we keep DNA evidence only for people who have been convicted of crimes, then this would probably result in African-Americans being disproportionately represented in a DNA database: some of those bad arrests, after all, would probably lead to convictions. Nonetheless, we might hope that the need to prove guilt beyond a reasonable doubt might provide some sort of check on police bias, and that at least the most egregious arrests would not be prosecuted. If we keep DNA evidence on everyone who has been arrested, however, then the existence of police bias against any group will lead directly to that group's being overrepresented, perhaps greatly, in the DNA database.

Well, one might ask: what's wrong with that? After all, the main difference between being in a DNA database and not being in one is that if your DNA profile is in the database, you're a lot more likely to get caught if you commit a crime. And since no one has a right to commit crimes and then get away with them, it's not clear why this would be such a bad thing.

My basic argument against this conclusion is as follows: either (1) there is nothing wrong with maintaining people's DNA profile in a database, or else (2) there is something wrong with it, in which case it's wrong to do this to people who have only been arrested, not convicted. In either case, however, keeping genetic information from everyone who has been arrested is the wrong way to go.

So: suppose that there is nothing wrong with keeping people's DNA in a genetic database. In this case, it seems to me that we should just collect samples of everyone's DNA. This would eliminate the unfairness of allowing police bias to determine how likely it is that you will get caught if you commit a crime. It would also eliminate another source of unfairness: relative searches.

It turns out that it's possible to use DNA profiles to determine not just whether the specific person whose profile it is committed a crime, but also whether any of his or her close relations did, by relaxing the search criteria slightly. (Think of this as like Googling on a few lines of an article, as opposed to Googling on the whole piece: done right, that can allow you to pick up related articles in addition to the original.) If it's unfair that biassed police officers could land you in a DNA database, it's doubly unfair that the arrest of your no-good cousin could ensure that DNA can be used to identify you.

If there's nothing wrong with being in a DNA database, then I would think that it would be best to simply make the database universal. That would completely remove the possibility of disparities due to police bias (or other sources of bias). After all, if everyone's DNA profile is included, then no one can feel unfairly singled out for inclusion.

Moreover, one of the main reasons why people support expanding DNA databases to cover people who have been arrested is that having more people in the database would make it easier to solve crimes. All of a sudden, any time a criminal left any DNA evidence at the scene of a crime, we could identify whose DNA it was. That would be great for law enforcement. But if what we want to do is expand the database, why not go all the way and include everyone? Why, that is, stop with people who have been arrested, if those people do not go on to be convicted of any crime? Wouldn't that be even greater?

I assume that the reason why no one is sticking provisions calling for a universal DNA database into obscure corners of very long bills is that it would be very unpopular; and that the reason it would be unpopular is that most people don't buy the assumption that there's nothing wrong with being included in a DNA database. People think there is something very wrong with the government having a sample of their DNA. They would Not Be Pleased with any Senator or Representative who inserted a provision for a universal DNA database into legislation in the dead of night.

But if it's bad to have your DNA profile included in a DNA database, then it seems to me obvious that this is not something we should do to people who have not been convicted of a crime. First, we try to minimize the bad things that follow simply from being arrested, and restrict them to those things that are necessary in order to actually get to the point at which someone is tried or released. If the arrestee might flee the jurisdiction, we can hold him or her in order to ensure that s/he is around for trial. But we do not get to do extraneous bad things to people simply because they have been arrested. (This is especially important given my earlier point about the possibility of police bias.) In this country, we are supposed to assume that people are innocent until proven guilty.

One way or the other, I think that a policy of putting everyone who has been arrested into a DNA database is wrong. Either there's nothing wrong with having your genetic profile in such a database, in which case everyone's profile should be included; or there is, in which case it should be limited to people who have been convicted of a crime. But including the genetic profile of everyone who has been arrested makes no sense at all, and it's an invitation to unfairness and injustice.

***

While I'm on the subject, it's worth saying a bit about what, exactly, is wrong with having your DNA profile in the federal database (whose name is CODIS), since a lot of people get it wrong. In the NYT article, for instance, we read this:

"Peter Neufeld, a lawyer who is a co-director of the Innocence Project, which has exonerated dozens of prison inmates using DNA evidence, said the government was overreaching by seeking to apply DNA sampling as universally as fingerprinting.

“Whereas fingerprints merely identify the person who left them,” Mr. Neufeld said, “DNA profiles have the potential to reveal our physical diseases and mental disorders. It becomes intrusive when the government begins to mine our most intimate matters.”"

Not exactly. If CODIS kept a person's entire genetic sequence, then it could be mined for all sorts of medical information. But CODIS stores only small snippets of DNA information, enough to limit matches to people who have the same or very similar DNA, but not enough to contain medical information, as far as we know:

"It is also important to note that the DNA profiles generated according to national standards do not reveal information relating to a medical condition or disease. The Short Tandem Repeat (STR) core loci selected for use in CODIS were specifically selected as law enforcement identification markers because they were not directly linked to any genetic code for a medical condition."

Of course, parts of the human genome that we do not now know to contain medically relevant information might turn out to contain such information in the future. (According to this story, "the standard DNA fingerprints used by police around the world contain a subtle signature which can be linked to a person's susceptibility to type 1 diabetes.") However, it seems unlikely that the snippets of genetic information currently used in DNA databases will allow us to infer a whole lot about a person's susceptibility to disease, at least for the foreseeable future.

(If you're worried about the government having access to all your genetic information, the thing to worry about is not CODIS, but the increasing move towards having law enforcement authorities retain all the DNA samples they collect, including things like samples from the victim of a crime, which are used solely to identify the DNA a forensic investigator should not think belongs to the criminal. There are a good reasons to retain those samples, especially if they might need to be retested, but that, not CODIS, is where people's entire genome can be found.)

As I see it, there are two main problems with including people's information in DNA databases: "function creep", and the possibility of surveillance. About the first: there are times when one might legitimately want to access a DNA database for some purpose other than law enforcement. For instance, if you were trying to identify the dead after some horrible disaster, checking DNA samples through CODIS would be a great thing to be able to do. But it's easy to imagine ways in which CODIS might be used that are a lot less benign. For this reason, CODIS (and state databases) ought to have very serious limits on the ways in which it can be used written into it. (I haven't analyzed the protections it currently has; they are here, and at the linked sections.)

The second problem is surveillance. At the moment, I am not feeling too sanguine about our government's willingness to respect existing legal limits on what it can do in the course of collecting intelligence on its citizens and residents. I would therefore be alarmed by the thought that the government had the power to detect all the DNA I ever leave around -- all those discarded coffee cups, restaurant forks, wayward bits of hair or skin, and so forth -- and identify them as mine. Call me paranoid, but I do not want them to have this power.

It's worth noting that this is one of those cases in which the fact that we cannot trust our government imposes real costs on us. (And by "trust" I don't mean "trust to be perfectly virtuous and behave like angels"; I mean "trust to obey any legal requirements, and any framework of checks, that we put in place.") If we did trust our government to abide by the laws, then we could, I think, design a system that would make a universal database OK. This would allow a lot of crimes to be solved more quickly, and would therefore protect all those people who will, as things are now, be victimized by people who would have been behind bars had such a database been up and running.

So the next time you hear of someone who has been raped or beaten or assaulted, it's worth bearing in mind that it's not beyond the realm of possibility that Richard Nixon and Oliver North and J. Edgar Hoover and John Yoo bear some small part of the blame.

***

One more related note: I have not, in this post, gotten into the problem of bad DNA labs. It's a terrible problem. I could have cited all sorts of cases, but I'll stick to Josiah Sutton:

"In 1999 Josiah Sutton, then 16 years old, was sentenced to 25 years in prison for rape. The evidence seemed airtight: the victim had spotted Sutton walking down a Houston street five days after her attack, and crime lab analysis from Houston's police department showed his DNA was an exact match with semen from the crime. The DNA evidence, in particular, was convincing—the analyst testified that the probability of the match being a result of chance was 1 in 694,000. But it was wrong.

Four and a half years after Sutton was sent to prison, William Thompson, professor of criminology at the University of California in Irvine, and newspaper reporters investigated hundreds of instances of sloppy laboratory work at Houston's Harris County crime lab.

Thompson calculated that the probability of Sutton's DNA matching the vaginal samples had been grossly exaggerated—one in eight black men would have had a similar match, he found. After an external audit produced a scathing report, the county shut down the lab and retested DNA evidence in nearly 400 cases. In March 2003 the lab found that a semen stain found in the car did not match Sutton's DNA, and Sutton was exonerated.

The incident, though shocking, is just a small part of a systemic problem, says Thompson. "Part of the problem in Harris County was indeed the lab," he says, "but a big part of the problem was the failure of the legal system to detect how atrociously bad and misleading the lab work was.""

On 60 Minutes, Thompson was even more scathing about the Houston police lab:

"“It’s the worst I’ve seen,” says Thompson. “This doesn’t meet the standard of a good junior high school science project.”

“I found consistent distortions of the statistical certainty of the DNA evidence. I found instances that looked like fudging of results, to fit the prosecution’s theory of the case. And I found a lab that consistently failed to use appropriate scientific procedures.""

Guess what? Only one person lost her job in the scandal that followed, and she was reinstated after a 28 day suspension.

Josiah Sutton spent four and a half years behind bars for that mistake.

There is no reason we can't do better in this country.

Comments

[Hartmut]

Let's not forget a potential planting of evidence (either by the criminal or the police). I have read that this has already happened with DNA and it looks far easier to me than planting convincing fingerprints.

[ThirdGorchBro]

Harris County's crime lab is a disgrace. As a Houstonian who believes in law and order, but also in justice, I wouldn't want my fate to be in the hands of that lab.

The horror stories about that lab convinced me that Texas should imitate Illinois and put a moratorium on the death penalty until we can straighten out our justice system.

[Debra Mervant]

One of the things that worries me about this scientific technology is the implicit belief that it is infallible, i.e. "objective" and does not necessitate interpretation. I suspect that the general public operates under these false assumptions, and that as a result, law enforcement officials tend to overrely on such (very expensive) techniques, slacking off on their capacity to put together an investigation based on clinical observation and reasoning (this is also a problem in medecine, by the way...)

[Gary Farber]

"an investigation based on clinical observation"

Most crimes do not actually take place in clinics, however. The phrase doesn't carry over.

And the least reliable tool of all in criminal investigations is the use of eyewitnesses.

[Jon H]

I betcha any foreigner detained at airports while entering the country will get the DNA taken.

That includes, say, businessmen and tourists.

[Ara]

On DNA labs: maybe DNA tests ought to be conducted in a kind of "double-blind" way, where prosecutors/counties do not know which labs they are sending to and labs do not know which prosecutors/cases they are working for. It just seems to me that as long as prosecutors have their choice of labs, they'll tend to pick the most compliant ones. Sometimes being a slobby scientist is a competitive advantage in going after government dollars.

I'm also assuming that most public defenders don't have access to funds enough to get an independent test done...

[Hartmut]

Good idea. And deliberately adding some "dummies" (i.e. DNA samples that are not connected to a case) would be nice too.

[BigHank53]

I read the book published about ten years ago by the FBI lab whistleblower. Among the FBI's charming habits was not calibrating their gas chromatographs. That way, the calibration records could not be subpeoned by a defense team and used to challege the FBI's lab results...

[Debra Mervant]

Gary, when I talk about clinical observation, I mean that the same investigative tools are at work in constructing a diagnosis as in constructing a crime investigation : listening to people's statements, relying on the coordination of several sources in order to construct a hypothesis/diagnosis , which is not the same thing as assuming that one eyewitness--or one test, for that matter, constitutes proof of an individual's guilt. Forensic medecine, when it is exercised properly and competently, shows just how much medecine and police investigation have in common.

[Dan]

There's the additional issue of statistical 'certainty.' Let us say that a given DNA sample--done correctly, handled responsibly--uniquely idenitifies one in a million people.

If you have two suspects, each of which has good evidence against him/her and a good alibi, then this DNA can definitely help clear up who is who.

But you don't keep a database in order to disabmiguate marginal cases. You keep a database so you can troll through it. And that "one in a million" suddenly looks like "250 Americans".

But the jury is going to be very easily convinced when you stand up and say "one in a million!"