The government can't get an email warrant if the account holder consents, court (wrongly) rules

Magistrate Judge James Orenstein of Brooklyn is a leader in the Magistrate's Revolt, a small group of federal magistrate judges who have often come up with unexpected theories to reject court-order applications in computer search and surveillance cases. I've found several of Orenstein's opinions in that vein very unpersuasive. Orenstein has handed down another decision earlier in the week that I find equally unpersuasive, and I thought I would explain why.

The decision goes by an amusingly generic name: In The Matter Of The Search Of Information Associated With [Redacted Email Address] That Is Stored At Premises Controlled By [Redacted Service Provider]. In the decision, Orenstein rejects the government's application for a search warrant to search a particular email account. Orenstein does not reject the application because it lacks the standard requirements of a search warrant, such as probable cause or particularity. Instead, he rejects the application because the target of the search was cooperating with the government and had consented to the search. In those circumstances, Orenstein rules, he cannot issue the warrant.

Orenstein's decision is clearly wrong, I think. I'll explain why in two parts that track the reasoning of the opinion.

Part 1: Contents of the Account

In the first part of the opinion, Orenstein argues that "the warrant procedure is inapposite" to the extent it seeks to obtain the contents of the target's emails. Orenstein starts by noting that 18 U.S.C. 2703(b) offers the government three ways to compel email: with a search warrant, with a subpoena afforded with prior notice and with a 2703(d) order afforded with prior notice. Orenstein then reasons that because the defendant has consented to the search, he must have had notice of it. And because the defendant had notice, he argues, the law bans the government from obtaining the emails with a search warrant and the government must obtain the emails another way without judicial review:

Congress has seen fit to provide for resort to the warrant procedure only in those circumstances in which the government seeks to proceed without notice to the subscriber. Compare id. § 2703(b)(1)(B), with id. § 2703(b)(1)(A). I assume that Congress understood what it was doing when it wrote the statute that way, and indeed such a structure makes sense: the distinction promotes judicial economy by imposing the burdens attendant to the warrant procedure (on both the executive and judicial branches) only in those instances in which it is necessary to preserve the secrecy of the government's investigation from the service subscriber whose communications the government seeks to access. No such necessity exists here; with the Defendant's consent, the government can secure the information it wants in the form it prefers without resort to the warrant process simply by issuing a grand jury or trial subpoena. See id. § 2703(b)(1)(A).

This argument is wrong for three reasons.

First, the provision Orenstein focuses on, 18 U.S.C. 2703(b), is not the standard way for the government to compel email. The provision was declared unconstitutional in United States v. Warshak, 631 F.3d 266 (6th Cir. 2010), because it allowed the government to get emails without a warrant. See id. at 288 ("The government may not compel a commercial ISP to turn over the contents of a subscriber's emails without first obtaining a warrant based on probable cause. … [T]o the extent that the SCA purports to permit the government to obtain such emails warrantlessly [as it does under 2703(b)], the SCA is unconstitutional."). All of the major Internet providers follow Warshak, and they won't turn over the contents of accounts with less process than a search warrant (as Orenstein appears to recognize in footnote 2). Given that the provision was declared unconstitutional, and providers don't follow it, it is difficult to use it as the standard for how to obtain email.

Second, even if you put Fourth Amendment concerns aside, the government still couldn't legally compel a provider to disclose the entire contents of an account under 2703(b). The reason is that Section 2703 requires a warrant for some kinds of email under 2703(a) and then permits the government to get other kinds of email with less than a warrant under 2703(b). Even if a provider is willing to ignore Warshak, obtaining the entire contents of an account would very likely still require a warrant under 2703(a).

Third, nothing in 2703(b) suggests that the government can't use a search warrant when a suspect has received notice. The three options in 2703(b) are just that: options. The government can use a search warrant "without required notice," or it can use less process than a warrant "with prior notice." Congress did that to allow a suspect to challenge warrantless access to his email under Section 2704. The usual rule is that the recipient of a subpoena has a pre-execution right to challenge it while the target of a warrant doesn't. The structure of 2703(b) enforces that usual rule by giving customers the notice needed to make the pre-execution challenge where such a challenge can be brought. And Congress allowed for the option of a search warrant because the consistent theme of the statute is that the government can always opt for more process than less. This is necessary to enable "one-stop shopping" with a single application, rather than multiple applications for different kinds of information sought under the statute.

Judge Orenstein reads that language instead as mandating a particular path—warrant or no warrant—depending on whether the suspect is on notice of the search. But I don't understand why he reads the statute that way. Orenstein suggests a necessity rationale by which magistrate judges are only to be bothered with warrant applications if it is necessary to preserve secrecy. But nothing in the text says that, and it make no sense because Congress also allowed delayed notice without a warrant under 18 U.S.C. 2705. Further, Rule 41 is pretty clear about a magistrate judge's obligation to issue a search warrant. Rule 41(d)(1) states that when an application is received, a magistrate judge "must issue the warrant" if probable cause is established. It's not an option. The rule says "must."

Part 2: Non-Content Records of the Account

In part two of the opinion, Orenstein denies the application for a warrant to the extent it seeks non-content records about the account. Orenstein recognizes that the statute permits the government to obtain the records with a warrant. According to Orenstein, however, there is no case or controversy allowing him to issue the warrant because the government can get the information without a warrant.

The argument seems to break down into five steps. First, because the account owner consented to the search, a provider has a legal obligation to hand over the non-content records pursuant to that consent. Second, because the provider has to hand over the account records, an order to hand over the account records doesn't make the provider do anything it doesn't have to do already. Third, if the order doesn't impose new obligations, then it has no practical effect. Fourth, if the order has no practical effect, it is not redressing any harm. And fifth, if it is not redressing harm, then there is no Article III standing allowing the judge to issue the warrant.

Here's the key passage, with a paragraph break added:

The problem is not that issuing a warrant contravenes the statute; the problem is that doing so will do nothing to alter the legal rights and obligations of any person or entity. Because the government has secured the Defendant's consent, it already has the right to access the Defendant's account records—and the Provider has a corresponding legal obligation to disclose them. Id. § 2703(c)(1)(C). "'Federal courts are without power to decide questions that cannot affect the rights of litigants before them.'" DeFunis v. Odegaard, 416 U.S. 312, 316 (1974) (quoting North Carolina v. Rice, 404 U.S. 244, 246 (1971)); Pitkin Supermarket, Inc. v. United States, 2016 WL 6879254, at *4 (E.D.N.Y. Nov. 21, 2016) (quoting same); see also Utjan v. Defs. of Wildlife, 504 U.S. 555, 561 (1992) (holding that retiressability is an element of the "irreducible constitutional minimum of standing").

Accordingly, deciding whether the government has established probable cause to believe that the Defendant's email account records will constitute evidence, fruits and instrumentalities of the Subject Offenses will not affect the government's right to compel the Provider to disclose those records. As a result, notwithstanding the SCA's applicability to a broad set of circumstances including those of this case, this court lacks the constitutional authority to determine the question of probable cause presented by the government's application.

I don't think this reasoning works. It is true that 18 U.S.C. 2703(c)(1)(C) requires a provider to disclose non-content records when the government has the consent of a subscriber to the disclosure of the records. This provision is almost never used, but it's an option in the statute. But I don't see how that matters here. First, it's not clear from the opinion that the account holder has consented to the disclosure of his records. The application only states that the account holder consented to government access to "the entire contents" of his account. It speaks of contents, not non-content records. And second, even if the account holder implicitly consented to the disclosure of the non-content records, it's not clear why the government can't use the different option of a search warrant. The statute explicitly permits the government to get non-content records in a range of ways, a search warrant among them under 2703(c)(1)(A). There is no rule that a court order can't issue if the other options are available.

Orenstein reasons that he can't issue a warrant to order a result that is supposed to happen anyway. According to Orenstein, there is no standing to issue an order if it doesn't change the government's legal rights. But I don't think this argument works for at least two reasons. First of all, a warrant is an ex parte court order rather than a resolution of a dispute. Warrant proceedings haven't traditionally fit the model of adversarial litigation, see Morrison v. Olson, 487 U.S. 654, 681 n.20 (1988). Given that, I don't think concepts like "standing" have a place in that context. When a court issues a warrant, or schedules a trial date, or grants permission to file an oversized brief, we don't ordinarily say that the court's power to issue the order is subject to the limits of adversarial litigation such as standing. Instead, such orders are generally considered part of the judiciary's inherent authority.

Second, even if a judge's ability to issue an ex parte order is subject to Article III standing limits, I don't think it makes sense to say that there is no standing just because the party seeking the order has a legal right to the information anyway. The party is seeking the order; it doesn't have the order. It is seeking the information; it doesn't have the information. Even assuming that the provider is supposed to give the government the information under the statute, it hasn't actually done that. If there was consent, the provider hasn't been told about it. Plus, having a warrant would give the government different powers to obtain the records and impose a different obligation on the provider to follow the specific terms of the warrant. The warrant would specify precisely how and when the warrant is to be executed—pretty different from the general obligation to follow the statute that doesn't have an obvious enforcement mechanism.