The Volokh Conspiracy

Mostly law professors | Sometimes contrarian | Often libertarian | Always independent

Volokh Conspiracy

Why the EFF—and then others—probably misunderstood the numbers on "sneak and peek" warrants


The Electronic Frontier Foundation published a report earlier this week alleging an astonishing increase in the use of sneak-and-peek search warrants. Sneak-and-peek searches are sometimes known as "covert searches" or "black bag jobs." The government breaks into a home, conducts a covert search, and leaves no sign of entry until days or weeks later. According to the EFF report, such searches have become routine in the last few years:

First, the numbers: Law enforcement made 47 sneak-and-peek searches nationwide from September 2001 to April 2003. The 2010 report reveals 3,970 total requests were processed. Within three years that number jumped to 11,129. That's an increase of over 7,000 requests. Exactly what privacy advocates argued in 2001 is happening: sneak and peak warrants are not just being used in exceptional circumstances-which was their original intent-but as an everyday investigative tool.

Elsewhere at the Post, Radley Balko relies on the EFF report to draw a host of conclusions. Among them, Balko condemns the Patriot Act that codified the sneak-and-peek standard:

Washington establishment types are often dismissive and derisive of the idea that members of Congress should actually be required to read legislation before voting on it—or at the very least be given the time to read it. There's also a lot of Beltway scorn for demands that bills be concise, limited in scope and open for public comment in their final form for days or weeks before they're voted on. If you're looking for evidence showing why the smug consensus is wrong, here is Exhibit A.

. . . . When critics point out the ways a new law might be abused, supporters of the law often accuse those critics of being cynical—they say we should have more faith in the judgment and propriety of public officials. Always assume that when a law grants new powers to the government, that law will be interpreted in the vaguest, most expansive, most pro-government manner imaginable.

TechDirt and also reported on EFF's findings and offered similar assessments.

Sounds pretty bad, right? Well, not so fast. I fear EFF's report may just misunderstand the significance of the annual "delayed notice warrant" report published by the Administrative Office of the U.S. Courts (AO). I suspect the numbers don't mean what EFF thinks they mean.

Here's the problem. In its report, EFF assumes that the numbers in the annual AO report only involve sneak-and-peek warrants. EFF's report looks at the dramatic increase in reported numbers and concludes that "sneak and peak warrants are not just being used in exceptional circumstances-which was their original intent-but as an everyday investigative tool." At first blush, that seems a plausible reading of the report. Back when the reporting requirement was established, the understanding was that the AO reports would report the number of sneak-and-peeks. But I think something else is happening: The AO numbers are not just reporting sneak-and-peek warrants. Instead, I think the numbers are combining several different kinds of warrants, only a small number of which are sneak-and-peeks.

I first came to think this in 2013, when Jonathan Witmer-Rich made the same claim as EFF in a series of blog posts over at Prawfs. I wrote a comment to one of his posts suggesting that something else might be happening: The reports might be combining sneak and peek warrants with other kinds of warrants that did not provide immediate notice.

My thinking was that federal investigators now get non-or-delayed-notice warrants for all sorts of things for which they didn't get warrants at all until a few years ago. Before the Patriot Act, for example, the FBI would routinely get e-mails with just a subpoena. There was no judicial oversight at all to install tracking devices on cars—the FBI just did it. Today, however, there is more privacy protection than before the Patriot Act in these contexts. The government has to get warrants for e-mails and to install GPS devices. Importantly, these warrants generally do not come with notice, or at least immediate notice. As I explained in my 2013 comment, that crazy spike in numbers probably tells us that the annual AO report is combining sneak-and-peek warrants with other kinds of newly-required-warrant practices. If you assume the AO report is just about sneak-and-peeks, you will conclude that the number of sneak-and-peeks has gone up exponentially. But when you consider the other practices that are probably now included because of the expanding warrant requirement, a very different picture emerges. Ironically, the increase in the reported numbers may reflect more privacy protection rather than less.

Professor Witmer-Rich's further investigation seems to confirm that something like this is happening. In the most recent draft of his forthcoming paper on delayed-notice warrants, he presents a very different understanding of the numbers published by the AO:

In the decade following congressional authorization of covert searches in 2001, the use of federal delayed notice search warrants has exploded, from less than 100 per year in 2002 to 5,601 in 2012. Just in the past six years, the rise is an incredible three thousand percent (from 174 in 2006 to 5,601 in 2012).

This data is alarming, but there is good reason to think it is misleading. Rather than showing a dramatic increase in the use of "sneak and peek" searches, the data may simply reflect a coincidental shift in the use of warrants for searches that used to be conducted without warrants: covert cell phone location tracking, covert GPS tracking, and covert searching of e-mail messages, among other things. Fundamentally, the data
does not allow us to disentangle these effects, and thus in its current form the data is not much use. In particular, the data does not tell the public, or Congress, how often investigators are conducting covert searches of homes and businesses-which is precisely what Congress sought to keep track of when it passed the reporting requirement in 2005. In short, the reporting requirement is broken, and it should be fixed.

And elsewhere:

Congress was focused on covert searches of physical spaces when it passed section 3103a and the corresponding reporting requirement. When Congress passed the reporting requirement in section 3103a(d), Congress expected to receive data showing how often investigators were covertly entering people's homes and businesses. Instead, Congress is receiving data showing how often investigators get authority to delay notice on many different types of warrants, including many that do not involve covertly entering any physical space, such as cell phone location tracking, GPS tracking, and searching e-mails. . . .

As of 2006, when the Administrative Office began compiling the 3103a Reports, there is good reason to think that the reported data was indeed capturing the number of "sneak and peek" searches. . . . In the past few years, several changes have occurred in the use of search warrants with delayed notice—changes that were largely coincidental with the passage of section 3103a, but which likely have begun to impact this data. These developments involve changes in the use of search warrants to conduct various types of searches, all involving new electronic technologies: tracking a person's location using cell phone location information, searching for the contents of e-mail messages, and the use of GPS tracking devices, among other things. There is good reason to believe that at least in the past few years, and possibly earlier, some substantial portion of federal delayed notice search warrants involve these types of searches, rather than covert searches of physical space.

It may be that the number of sneak-and-peek warrants is up, and perhaps up a lot. But, unfortunately, we don't know that either way. As Professor Witmer-Rich says, the reporting requirement appears to be broken. It seems to be combining lots of surveillance practices instead of just the one that Congress had in mind when it enacted the reporting requirement. Because the warrant requirement expanded for those other practices, it's hard to know how much of the increased number of warrants is reflecting warrants where the feds didn't get them before or the changing numbers of sneak-and-peek warrants.

Or at least that's my sense of things. If others have looked into this and have come to a different conclusion, I'd love to hear it and I'd be happy to post a correction/amendment.

UPDATE: I fiddled with the post a bit soon after putting it up to clarify two sections.