A while back, I blogged at length about the Third Circuit’s pending case involving government access to historical cell-site records. The issue in the case is what legal standard the government must satisfy to obtain orders requiring phone companies to disclose such information. The district court had ruled that a warrant was required, and the government argued that the correct standard under the law was a “specific and articulable facts” court order under 2703(d) rather than a search warrant. Yesterday, the Third Circuit handed down its decision: In The Matter Of The Application Of The United States Of America For An Order Directing A Provider Of Electronic Communication Service To Disclose Records To The Government. In this post will explain the Third Circuit’s decision; try to figure out what it means (which turns out to be quite tricky); and then explain why I think it misreads the Stored Communications Act on an important point.
I. The Third Circuit’s Decision
The Third Circuit’s decision, written by Judge Sloviter, has three major parts. First, the court ruled that the government is right that under the Stored Communications Act, the government can obtain historical cell-site records under 2703(d) without getting a warrant. Second, the court agreed with amicus the Electronic Frontier Foundation that magistrate judges do not have to issue such orders. According to the Third Circuit, it is up to individual magistrate judges to make the decision of whether to issue the orders or else require a full probable cause warrant. Here’s the core argument:
Section § 2703(d) states that a “court order for disclosure under subsection (b) or (c) may be issued by any court that is a court of competent jurisdiction and shall issue only if” the intermediate standard is met. 18 U.S.C. § 2703(d) (emphasis added). We focus first on the language that an order “may be issued” if the appropriate standard is met. This is the language of permission, rather than mandate. If Congress wished that courts “shall,” rather than “may,” issue § 2703(d) orders whenever the intermediate standard is met, Congress could easily have said so. At the very least, the use of “may issue” strongly implies court discretion, an implication bolstered by the subsequent use of the phrase “only if” in the same sentence. . . .
Under the EFF’s reading of the statutory language, § 2703(c) creates a “sliding scale” by which a magistrate judge can, at his or her discretion, require the Government to obtain a warrant or an order. EFF Br. at 6. As the EFF argues, if magistrate judges were required to provide orders under § 2703(d), then the Government would never be required to make the higher showing required to obtain a warrant under § 2703(c)(1)(A). See id.
So according to the Third Circuit, a magistrate judge has discretion to turn down an application for an order even if the 2703(d) order is satisfied. The third and final question is, does the judge have absolute discretion to decide whether to turn down the order, or is there some guide that the magistrate judge is supposed to use?
Here things get fuzzy. Judge Sloviter instructs that magistrate judges should require warrants “sparingly,” for the reason that “Congress also included the option of a § 2703(d) order.” She also writes that magistrate judges do not have “arbitrary” discretion, and that in the rare cases that a magistrate decides to require a warrant, the judge must “make fact findings and give a full explanation that balances the Government’s need (not merely desire) for the information with the privacy interests of cell phone users.”
What exactly does that mean? That is, what is the standard? To be candid, I’m not sure. A discussion around pages 26-27 suggests that perhaps magistrates should to conduct an ex ante constitutional analysis of whether the cell-site surveillance would require a warrant under the Fourth Amendment. Here the court hints that in its view, cell-site surveillance is analogous from a constitutional standpoint to tracking device surveillance in United States v. Karo and United States v. Knotts rather than the numbers dialed in Smith v. Maryland: Because “it is unlikely that cell phone customers are aware that their cell phone providers collect and store historical location information,” a cell phone user has not conveyed that information to the phone company voluntarily and the third-party doctrine does not apply.
The analysis here is pretty murky, but perhaps the Court is suggesting that if cell-site surveillance reveals that a caller is at home during the call, then under Karo that information could be constitutionally protected and the judge should require a warrant? I’m not really sure what the court is suggesting — elsewhere the court suggests that the decision of what standard to use is entirely up to Congress, so perhaps I’m wrong about that. Ultimately the Third Circuit remands for fact-finding, so perhaps the Court is really punting the issues of what the standard is for another day. In the end, I’m not sure. (If you read the opinion differently and have ideas about the standard, please post them in the comment thread. I would very much appreciate it.)
Judge Sloviter ends with this criticism of the Stored Communications Act:
In the issue before us, which is whether the MJ may require a warrant with its underlying probable cause standard before issuing a § 2703(d) order, we are stymied by the failure of Congress to make its intention clear. A review of the statutory language suggests that the Government can proceed to obtain records pertaining to a subscriber by several routes, one being a warrant with its underlying requirement of probable cause, and the second being an order under § 2703(d). There is an inherent contradiction in the statute or at least an underlying omission. A warrant requires probable cause, but there is no such explicit requirement for securing a § 2703(d) order. We respectfully suggest that if Congress intended to circumscribe the discretion it gave to magistrates under § 2703(d) then Congress, as the representative of the people, would have so provided. Congress would, of course, be aware that such a statute mandating the issuance of a § 2703(d) order without requiring probable cause and based only on the Government’s word may evoke protests by cell phone users concerned about their privacy. The considerations for and against such a requirement would be for Congress to balance. A court is not the appropriate forum for such balancing, and we decline to take a step as to which Congress is silent.
II. Why I Think the Third Circuit’s Opinion is Incorrect
The first part of the Third Circuit’s opinion is pretty clearly right. However, I think the second part misunderstands the Stored Communications Act, and that the court’s constitutional hints are not persuasive. In this part of the post, I want to explain why.
Let me begin with the criticism that the opinion offers at the end about the Stored Communications Act, which seems to frame the Third Circuit’s statutory misunderstanding:
A review of the statutory language suggests that the Government can proceed to obtain records pertaining to a subscriber by several routes, one being a warrant with its underlying requirement of probable cause, and the second being an order under § 2703(d). There is an inherent contradiction in the statute or at least an underlying omission. A warrant requires probable cause, but there is no such explicit requirement for securing a § 2703(d) order.
There’s no contradiction or underlying omission here, however. The drafters of the SCA realized that the government would often obtain different kinds of information all at once in a single case that could be obtained individually under a range of different legal thresholds. The SCA therefore incorporates the principle that the greater includes the lesser: If the government can get particular information under a low threshold, it also always has the option of getting the information using a higher threshold. This allows the government to obtain a single higher-threshold order (such as a search warrant) to obtain several different kinds of information instead of having to obtain a warrant for the most sensitive information, a 2703(d) order for less sensitive information, and a subpoena for the least sensitive information. Here’s how I explained it in my article on the Stored Communications Act:
One interesting aspect of § 2703 is that it generally allows the government to obtain greater process when lesser process will do. If a provision of § 2703 allows government agents to compel information with a subpoena, it also allows them to obtain that information with a 2703(d) order; if it allows agents to obtain information with a 2703(d) order, then a search warrant is also acceptable. Why might the government want this option? The main reason is efficiency. Investigators may decide that they need to compel several types of information, some of which can be obtained with lesser process and some of which requires greater process. The “greater includes the lesser” rule in § 2703 allows the government to obtain only one court order–whatever process is greatest–and compel all of the information in one order all at once.
Kerr, A User’s Guide to the Stored Communications Act and Legislator’s Guide to Amending It, 72 Geo. Wash. L. Rev. 1208, 1220 (2004). See also J. Carr & P. Bellia, Law of Electronic Surveillance § 4:77, at p. 4-193 (2006)( “One feature of ECPA is that through use of greater legal process officials can gain access to any information that they could obtain with lesser process.”). Giving the government the option of satisfying a higher threshold was designed to reduce paper work at no cost to privacy, not to give magistrate judges the discretion of whether to sign court orders or to instead demand more process.
As for the court’s core argument that 2703(d) was designed to be discretionary, I find this quite unpersuasive. In fact, I admit that when I read this argument in EFF’s brief, I literally laughed out loud: I’ve been studying and working with the SCA since 1998, including working on proposed amendments to it and testifying about it, and I don’t think I have ever encountered the suggestion that 2703(d) orders are discretionary until I read it in EFF’s brief. So on one hand, kudos to my friends at EFF for making such a creative argument (BTW, hey, Kevin – congrats! And can you believe you beat Eck on an ECPA question?!?). In my view, it shows how a surprising argument just might find a home with a sympathetic majority of generalist judges. But creativity aside, I don’t think the argument works.
To understand why, it helps to start with the important background principle established in Ex Parte United States, 287 U. S. 241 (1932): Magistrate judges do not have discretion to decide whether to issue court orders if the government satisfies the legal threshold. As the Court explained in Ex Parte United States, where the magistrate judge had declined to issue an arrest warrant:
The authority conferred upon the trial judge to issue a warrant of arrest upon an indictment does not, under the circumstances here disclosed, carry with it the power to decline to do so under the guise of judicial discretion; or, as this Court suggested in Ex parte United States, 242 U. S. 27, 242 U. S. 42, the power to enforce does not inherently beget a discretion permanently to refuse to enforce.
Modern privacy law reflects that principle. The magistrate judge’s job is just to see if the legal threshold: He “must” issue the order if the legal requirements are satisfied. See Fed. R. Crim. Pro. 41 (search warrants). See also United States v. Santtini, 963 F.2d 585, 596 (3d Cir. 1992).
This makes sense if you think about the role of ex ante court order requirements. The purpose of such requirements, whether in the form of the Fourth Amendment warrant requirement or statutory requirements, is to ensure that the government has the threshold of evidence that the Supreme Court or Congress has chosen before the access to information occurs. In the case of a statute like the Stored Communications Act that lacks a suppression remedy, ex ante review ensures that the government is limited to acting when a specific statutory threshold is met. The ex ante review makes Congress’s command enforceable, substituting for ex post review that would occur if there were a suppression remedy.
A discretionary court order system, in which each judge has the discretion of whether to allow the government to proceed at that threshold or require more process, makes no sense in that setting. Indeed, off the top of my head, I have never encountered any ex ante court order requirements in any criminal setting that gave such discretion to a magistrate judge. Giving the power to individual magistrates would be a pretty remarkable departure from established practice: Without some clear language doing do, it would be a strange reading of a statute. All the more so because it would lead to serious problems in practice. If each judge has the discretion, and the government can forum-shop among magistrate judges, it creates a strong incentive for the government to go to the magistrate judge who tends to allow court orders at the lower threshold rather than the higher threshold.
As for the statutory issue, I think the language of the SCA doesn’t give any sign that Congress intended such a departure from the standard practice. We need to start with 2703(c), not 2703(d). It states:
A governmental entity may require a provider of electronic communication service or remote computing service to disclose a record or other information pertaining to a subscriber to or customer of such service (not including the contents of communications) only when the governmental entity. . . obtains a warrant issued using the procedures described in the Federal Rules of Criminal Procedure [or] obtains a court order for such disclosure under subsection (d) of this section;
2703(d) then says:
A court order for disclosure under subsection (b) or (c) may be issued by any court that is a court of competent jurisdiction and shall issue only if the governmental entity offers specific and articulable facts showing that there are reasonable grounds to believe that the contents of a wire or electronic communication, or the records or other information sought, are relevant and material to an ongoing criminal investigation.
Notably, in using the words “obtains” and “shall issue,” Congress chose exactly the same terms for 2703(d) orders that it uses for search warrants. “Shall issue” is the same language that the version of Rule 41 used for search warrants back in 1994, when this language was added to the SCA. Rule 41 read: “If the federal magistrate judge or state judge is satisfied that grounds for the application exist or that there is probable cause to believe that they exist, that magistrate judge or state judge shall issue a warrant identifying the property or person to be seized and naming or describing the person or place to be searched.” Rule 41, 1994 ed. (emphasis added). The choice of the same language as the non-discretionary warrant procedure, together with the background norm that ex ante court order applications are non-discretionary, to my mind indicates that Congress intended 2703(d) orders to be just as discretionary as warrants: That is, not discretionary at all.
It’s true that in the beginning of 2703(d) the statute does say that certain kinds of courts — courts of competent jurisdiction — “may issue” 2703(d) orders. But that’s not supposed to make the procedure discretionary. “May issue” appears in 2703(d) to indicate which courts are empowered to grant 2703(d) orders. It’s the same role played in the search warrant setting by Rule 41(b), “Authority to Issue a Warrant,” which states that “a magistrate judge with authority in the district — or if none is reasonably available, a judge of a state court of record in the district — has authority to issue a warrant to search for and seize a person or property located within the district.” In the case of 2703(d) orders, there has been a legal dispute in the state courts as to whether state courts are empowered to issue 2703(d) orders. The “may issue” language was designed to explain which courts are empowered to issue 2703(d) orders, echoing Rule 41(b). In particular, courts of competent jurisdiction are empowered to issue the orders — as defined by 18 U.S.C. 3127(2).
The non-discretionary nature of 2703(d) is all the more likely given that the 2703(d) standard was added to the SCA in 1994 in order to raise the privacy level for government access to certain non-content information above a mere subpoena. Under the original 1986 Act, non-content information could always be obtained with a subpoena. The 1994 amendments raised the privacy threshold to a Terry v. Ohio– style court order in order to require the government to establish specific and articulable facts instead of the mere revelance standard of subpoenas. It would be strange if the celebrated privacy enhancement passed in 1994 establishing the 2703(d) standard instead of a subpoena were actually, and apparently secretly (with no one noticing until EFF’s brief), actually giving magistrate judges the discretion of whether to require a warrant or satisfy the 2703(d) order standard.
Finally, a brief note about the Fourth Amendment issue. I’ve explained before why I don’t think you can apply the Fourth Amendment third-party doctrine by assuming that communications technologies are a “magic box.” Notably, Smith v. Maryland did not do that: Smith presupposed a telephone user who knows how phones work. Given that approach in Smith, I don’t think you can presuppose a cell-phone user who sees cell phones as magic boxes that don’t need to communicate with cell towers. But even if you disagree with me here, the idea of applying Karo and Knotts to a cell-site information still cuts against Smith. If I understand the Third Circuit’s opinion correctly, the suggestion is that if you’re calling on a cell phone from your home, the government perhaps needs a warrant to know you’re at home at the time you’re calling. I wonder, though, if you accept that, don’t you have to overturn Smith v. Maryland on its facts?
After all, when you make a call from a land-line phone in your home, as in Smith, pen register information tells you exactly where you are: You’re inside your home, making the call. Indeed, back in the old days, the telephone company installed the phone and there were no cordless phones: Pen register information necessarily told the government you were in the home, and perhaps even what room you were in inside your home. And yet Smith v. Maryland found the location information inside the home to be irrelevant:
[T]he site of the call is immaterial for purposes of analysis in this case. . . . Regardless of his location, petitioner had to convey that number to the telephone company in precisely the same way if he wished to complete his call. The fact that he dialed the number on his home phone rather than on some other phone could make no conceivable difference, nor could any subscriber rationally think that it would.
Of course, it’s possible to distinguish the locational information conferred from cell-site information from the location information from numbers dialed on a home land-line. The former applies both when calls are made and not made, while the latter only is generated when a call is made or received. Still, I think there is some tension there given that the pen-register information in Smith indicated to the police that Smith was inside his home making the calls.