« (When) Was Fingerprinting Unconstitutional? | Main | If the War on Drugs Isn’t Driving Prison Population, What Is? Or, Better Put, *Who* Is? »

Wednesday, June 05, 2013

Some Reactions to Maryland v. King, and a Question for Barry Friedman and other 4A Friends

I have done only a bit more than skim Maryland v. King, and because I'm not really a Fourth Amendment guy, I'm preserving my strategic ambivalence about the outcome as a matter of doctrine. But I have a policy question for NYU's Barry Friedman along with a few other hasty reactions.

First, Barry writes the following in Slate, :

"Did the fact that Alonzo King was accused (not convicted) of pulling a shotgun on some folks provide a better reason to believe he’d committed an unrelated rape than that anyone else walking the streets had done so? Hardly."

It seems to me that Barry's posing either a specific or a general empirical question, neither of which I possess excellent knowledge about, although I bet there are good proxies out there for saying we know more than nothing.  In fact, I bet the reason law enforcement wants access to DNA of arrestees is because of their view that arrestees provide, on the margin, a better pool to capture DNA from than a random selection of the population at large.  Law enforcement interests here are driven by Big Data patterns that suggest that felony arrestees are likelier to be tied to other crimes than non-felony arrestees.  That's an empirical claim and Barry doesn't provide any links or data to suggest that we should doubt that claim's truth.

From a crime control perspective, would it be better if we had a DNA swab of every person and new baby? Sure, assuming the integrity of the collection. Indeed, Friedman thinks this would be permissible under the 4A (were Congress to pass it) but it's not likely to be authorized by Congress anytime soon.  (Btw, would a nation-wide DNA registry itself be unconstitutional under the 4A? Distributed benefits and costs might save it from the suspicionless problem. Curious for thoughts.)

But as long as a) we are not swabbing every new baby and all existing persons to create a national DNA database, and b) we are taking the time to inventory and identify felony arrestees, can anyone doubt that someone arrested for a felony is, on the margin, more likely to be guilty of some other offense that's out there?

I'm not saying there aren't doctrinal or other reasons that should restrain the DNA swabs. But simply as a matter of statistics or common experience, I'm left wondering  what supports the pretty heterodox view Barry offers that felony arrestees are not in fact more likely to have ties to other crimes than a randomly selected individual? Yes, I recognize that the value of the signal of a felony arrest is not the same as the signal of a felony conviction, but ... if we were going to block the swabs for crime-control purposes on constitutional grounds, let's at least be aware of what's being traded off in the name of constitutional fidelity.  And while we're at it, let's not forget that wide DNA access has the capacity not only to reduce Type II errors, but also to exonerate and thus redress Type I errors too. Fixing false positives is a constitutional value as well as a moral imperative for state officials. I'm not sure the 4th amendment claims advanced by the dissenters and their supporters are adequately sensitive to that, even if the majority implies this is happening already as a matter of fact.*

2.  I agree with Scalia's dissent that  the "identification" arguments on behalf of the DNA swabs are more make-weight than the straightforward though constitutionally more tricky arguments in favor of clearing cases and fixing mistakes. That's because the police could always use the DNA swab to promote their administrative needs (e.g., ensuring that the offender doesn't have a record of violence toward prison officials or communicable diseases that would have to be taken into account for housing him) without using the DNA swab to scope out possible relevance to other crimes.

3. The Court's special needs doctrine allows for suspicionless searches of the public in order to regulate safety or achieve other non-crime detection goals of certain policy weight. Here are two reasons for thinking that the majority's result is correct even if not its reasoning.

a) It's not that far a stretch to say that given the criminal justice system's interests in ensuring that the institutions of punishment are taking adequate care and precaution for the wellbeing of inmates and officials, that the population of felony arrestees is distinct from the population at large, and thus the goal of using DNA to ferret out possible dangerousness or illness is one that should pass muster on special needs grounds. But the reason I  don't love this argument is because if taken on good faith, it would not permit allowing the DNA information to be used to exonerate previously convicted offenders. That would probably be too close to the crime-detection purposes that the special needs doctrine is supposed to be attentive to. However, one might slice the constitutional baloney very thinly and say: DNA swabs are constitutional for administrative purposes pre-conviction, and they are also constitutional for purposes of exonerating others, but they can't be used as the basis to clear other cases against the defendant whose cheek is being swabbed.

b) Speaking of slicing constitutional baloney thinly, I didn't see this argument and it seems worth consideration too--though I detest it because I'm doubtful of the constitutionality and morality of the underlying practices. Here goes: Crime detection is distinct from calibrating punishment. In indeterminate sentencing regimes as well as structured sentencing that allows for "real offense" sentencing instead of (my preferred) charge offense sentencing, the admission of the DNA evidence as a tie to other crimes should be permitted for purposes of sentencing offenders on an individualized basis on the basis of conduct not proven to the jury beyond a reasonable doubt.  So, say King is in Texas and convicted of aggravated assault, which leaves him open to  a 5-99 year spread under the statute for first degree felonies.  The sentencing judge/jury/parole folks can all take into account that he's been tied via DNA to other rapes, even though not convicted of those rapes.  Poof. The sentence for the assault goes up, we don't bother with charging and convicting King for the rapes, and we rest our heads on the pillow of Williams v. New York.  Sentencing is distinct from crime-detection. Right? How awesome is that. Ick.



*Scalia notes in his dissent (fn.2) that the Type I error redress option is not currently available b/c of the way the FBI runs its DNA databases. That could be fixed of course, and should be.




Posted by Administrators on June 5, 2013 at 03:54 PM in Article Spotlight, Constitutional thoughts, Criminal Law, Dan Markel | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Some Reactions to Maryland v. King, and a Question for Barry Friedman and other 4A Friends:


Thanks to Adam Liptak's piece in the NYT today, I stumbled across this interesting study already claiming to demonstrate the powerful effect DNA databases have on reducing crime.


Posted by: Dan Markel | Jun 11, 2013 10:46:26 AM

Btw, when I first blogged about this case as it was winding up the court's path, I was much more worried about the role of DNA swabs in arrestee's mouths as an encroachment on the presumption of innocence.


I'm not sure where I am in sum on this case. That said, from a scholarly perspective, there's at least a great set of regressions to be run regarding how effective this legal rule might prove to be in reducing Type II errors over the coming years. Paging Jon Klick and JJ Prescott.

Posted by: Dan Markel | Jun 10, 2013 6:00:39 PM


In addition to the way that databases can be used to exonerate that I mentioned to you offline (and which you mentioned in your last comment), I neglected to describe the most obvious way that databases can be used to correct Type I errors:

Databases have been used in postconviction DNA testing when that DNA testing of an evidentiary sample excludes the defendant as the contributor of the probative sample. But even then, the prosecution puts more weight on other evidence that it views as indicia of guilt, notwithstanding the new DNA evidence. So in that instance, the database has been used to achieve a hit to essentially solve the crime and ameliorate any lingering doubt about innocence on the part of the prosecutor, even though the exclusion itself is almost certainly enough to achieve a vacation of the conviction and exoneration. Michelle identified this above and it is the most common way databases are used in postconviction to solve Type I errors. At least in terms of the 307 known DNA exonerations in the U.S., there are stats on how often the database was used to find the true perp, which I can obtain and share if anyone is interested.

All that said, the only reason this is necessary in either of my examples is because of the concern you stated: that when there is exclusionary DNA results on a probative piece of evidence, prosecutors will too often invent an unndicted co-conspirator to explain it away, probably due to tunnel vision that doesn't allow them to effectively assimilate new facts that are contrary to their preconceived notions about the case.

Posted by: Seth Miller | Jun 6, 2013 9:02:23 AM

To anon: T1 vs t2 errors was something i had to figure out over time too, but it required as much mental work as figuring out false positive and false negs. alas.

Andrew, the 4th Amdt issues are not coextensive with the moving the needle point, but certainly, if Amar and Katyal's NYT oped is to be credited as relevant if not dispositive, there's substantial overlap re: the reasonableness of the warrantless search and the extent to which the needle is moved for reasons that cannot be reduced to other proxies. Accordingly, given the caveats I issued in the post, I'll take your point as friendly to my challenge to Barry, but perhaps i'm being obtuse.

Michelle/Orin, thanks.

One last point for now from a friend who runs one of the "innocence" legal shops. He notes that Scalia is wrong on fn. 2. Specifically, and since this goes above my pay grade, I'll just quote what he wrote and open up the floor to others to see if they agree:

"Scalia is wrong as a practical matter in footnote 2 that CODIS cannot currently be used to redress Type I errors. We have done it on a number of occasions where there is a pre-trial DNA exclusion on a probative piece of physical evidence using a pre-CODIS DNA method yet the individual is still convicted. We get retesting using a CODIS compatible method and have the lab upload it to the database to see if we get a hit, hopefully leading to the true perp after some additional fact investigation.

Given that, on balance, a more robust database, especially filled with people who empirically are more likely to be tied to crime, naturally increases the likelihood that using databases will help exonerate the innocent. I think both opinions gave this short-shrift with Scalia erroneously rejecting it altogether as a valid policy rationale for this law."

Posted by: Dan Markel | Jun 5, 2013 11:03:27 PM

Dan, your question is hard to answer because Barry's essay introduces and applies a new constitutional rule. According to Barry, the Fourth Amendment should be read to require a "really, really good reason" to subject the group to a regulatory search. It's hard to criticize Barry's application of his test because we don't know enough about that test to say if Barry is applying it correctly.

Posted by: Orin Kerr | Jun 5, 2013 9:17:59 PM


Of course hundreds of Type I errors have been fixed by DNA, but these usually occur through a convict managing to secure the right to use newly available technology to test the crime scene DNA in his case against his own DNA. I take it, in (a), you're asking instead about the power of offender databases (as opposed to crime scene and convict DNA) to correct Type I errors, that is, to exonerate. I don't have numbers, but when I was researching these cases ten years ago, at least, the fact pattern usually worked something like this: After the convict is permitted to have himself tested against the crime scene DNA and no match results, then law enforcement will usually test the newly processed crime scene DNA against the offender database, and often they will get a hit for someone else already in the system. In some cases, the no-match for the convict is enough by itself to overturn the conviction, and the hit in the offender database is gravy. Here is such a case: http://arstechnica.com/civis/viewtopic.php?f=23&t=628338

But in other cases, as you suggest in (b), prosecutors aren't convinced, for a variety of reasons, that a no-match alone exonerates the convict. Whether they are convinced or not depends on both the details of the case (e.g., how likely it is that there were multiple assailants; how likely it is that crime scene DNA could have been innocently left) and also on the individual prosecutor's discretion and willingness to admit s/he was (probably) wrong.

For instance, in 2001, it was revealed that sperm left on the body of the Boston Strangler's final victim did not match that of Albert DeSalvo, who had been suspected of being the strangler (and was killed in jail while serving time for an unrelated offense). Prosecutors responded by saying that this didn't posthumously exonerate DeSalvo, who could have killed the final victim without raping her, but that it did suggest that at least one other person was present when she was killed. Compare that case with that of Jeffrey Grant, incarcerated awaiting trial in 1999 for being the Long Beach rapist. The same DNA was left at two rape scenes, and prosecutors were shocked when Grant's DNA didn't match. Still, they dropped all charges, saying, "You can speculate that whoever did these crimes might have had an accomplice, but it is only speculation, and you can't prosecute someone based on speculation."

In these circumstances, the convict's case for postconviction relief usually becomes stronger if his no-match is coupled with a hit in the database to someone else -- preferably, someone with an MO that fits the instant crime and with no ties to the convict. That's more probative than a no-match alone, and so this is where I see offender databases, as currently constructed, sometimes correcting (as opposed to preventing) Type I errors.

Posted by: Michelle Meyer | Jun 5, 2013 8:58:31 PM

I always get confused by type I v type II and have to look it up. Is there a reason to prefer those terms over false positive and false negative? I ask because the idea is important, but I would prefer to stick with language I can understand.

Posted by: Anon | Jun 5, 2013 8:26:20 PM


When taken literally, Barry's statement does seem to be refuting an empirical claim. If we grant that as an empirical matter, someone pulling a shotgun had a better chance as being tied to a past crime, however, and all we're looking at is a "moves the needle" standard, it really doesn't shed any light on the 4th Amendment question. A person who lives in the same town as a crime scene is surely more likely to have been a part of it than someone the next town over, yet no one would use that fact to suggest that they can be searched for it on that basis alone. (Of course, high-crime neighborhoods *are* used as a factor, but not the only factor, and "high crime" is more than "moves the needle".)

Posted by: Andrew Selbst | Jun 5, 2013 7:11:52 PM

Thanks for those links, Michelle. Very interesting stuff.

Your first post about the universal database made me think about something somewhat related, though to some extent, it diminishes without undermining my point about fixing type 1 errors. These questions/challenges are directed generally, not Michelle specifically.
a) Do we know how many Type 1 errors have been fixed by these DNA databases so far?
b) Is there a substantial risk that in many of the situations, prosecutors will argue that the presence of someone else's DNA only shows that the crime may have been committed by more than one person, and thus, we might not see as many type 1 error fixes as we think/hope?

Posted by: Dan Markel | Jun 5, 2013 6:31:01 PM

Hi Dan,

I have a post (http://www.thefacultylounge.org/2013/06/maryland-v-king-low-stringency-dna-searches-and-the-case-for-a-universal-database.html) exploring the possibility of a universal database, especially in light of the fact that regularly-conducted familial searches of CODIS indirectly include arrestees' relatives in the database.

I also touch there on the Type I error issue. You're right, of course, that crime scene samples from purportedly "solved" crimes can and should be maintained in the database, but even if "offender" databases (scare quotes, given that so many who are included in them have not necessarily committed any offense) don't today very often correct Type I errors, they already serve to prevent such errors from occurring in the first place (and a universal database would even better accomplish this).

(Although your post doesn't raise this issue, and so this pimping out of my other post is truly gratuitous, as long as I'm at it, I also have a post (http://www.thefacultylounge.org/2013/06/looking-for-the-next-maryland-v-king.html) on variation among state and federal statutes authorizing DNA collection from arrestees along several dimensions (triggering offenses; whether arraignment, indictment, or judicial determination of probable cause is required prior to collection and/or uploading of samples; expungement policies; juveniles; and non-law enforcement uses of the database) that King suggests may make a constitutional difference.)

Posted by: Michelle Meyer | Jun 5, 2013 5:31:56 PM

The comments to this entry are closed.