The Best Way to Rule for Carpenter (Or, How to Expand Fourth Amendment Protections Without Making A Mess)

(Cross-posted from Lawfare)

Last month, the Supreme Court held argument in Carpenter v. United States, the pending case on whether the Fourth Amendment protects cell-site records. There seemed to be at least five votes sympathetic to ruling for Carpenter. At the same time, there was very little agreement about how to get there. What line should the Court draw, and based on what rationale? No clear answers emerged.

In this post, I want to identify what I think is the best way to rule for Carpenter. To be clear, I don't think this approach is the best way to rule. My amicus brief explains why I think cell phone users have no Fourth Amendment rights in their historial cell-site records. But if the Court wants to go the other way, there are better and worse ways to do that. This post identifies what I think is the best way to rule for Carpenter that would cause the fewest headaches going forward. To put it in law clerk terms, this is the thinking you do when your judge tells you, "Thanks very much for sharing your views. Now write it the other way."

Here's my bottom line: If the Court wants to rule for Carpenter, it should hold that individuals have Fourth Amendment rights in records solely useful to the government to identify that individual's physical location. In the language of modern Fourth Amendment law, government collection of location records, the sole government function of which is to reveal a person's physical location, infringes that person's reasonable expectation of privacy.

Because historical cell-site records are solely useful to the government to identify an individual's physical location, accessing those records without a person's consent would be a search. The Court should reach this result by relying on two traditional models of Fourth Amendment protection, what I have called the private facts model and the policy model.

This is a long post, so here's a road map. I'll start with an overview of the Fourth Amendment toolkit; then articulate and justify my proposed rule; next turn to the models that should be used to justify the rule; and then consider whether the Court should also decide the reasonableness issue of how much cause is sufficient to justify cell-site collection.

I. The Fourth Amendment Toolkit

I'll start with the big big picture. The Carpenter argument suggests that a majority of the Justices want to engage in what I have called equilibrium-adjustment. As I have explained, the Supreme Court sometimes responds to perceptions that technological change has greatly altered government power by adopting a new legal rule to restore the prior equilibrium. I personally don't think Carpenter is the right case for equilibrium-adjustment, as I think you need to take each technology on its facts. But the Justices seemed to want to rule more broadly. If I read the tea leaves correctly, a majority wants a new rule to account for how technology makes it easier for the government to know a person's location. They perceive that technologies like GPS surveillance and cell phones have expanded government power under the old legal rules, and they want a new legal rule to help limit that power. It's classic equilibrium-adjustment.

But how? To do that, you need to start by understanding the four basic rationales the Court has used to justify Fourth Amendment protection. Opinions draw from these four rationales—or as I have called them, four models—often mixing and matching different rationales in different cases. When the Court applies the probabilistic model, it considers the likelihood that a person's stuff or information will remain free of outside inspection. When the court applies the positive law model, it considers whether a source of positive law outside the Fourth Amendment prohibits the government's evidene collection. When the Court applies the private facts model, it makes a judgement that certain information should be deemed private (or not) and protected (or not) accordingly. Finally, when the Court applies the policy model, it justifies its holding based on the policy implications of doing so.

Okay, so that's the background. Let's move on to the two big questions. First, what should the rule be? And second, what model or models should the Court use to justify the rule?

II. The Rule—What Is Protected?

If the Court wants to rule for Carpenter, I think the best rule would be that the Fourth Amendment gives individuals Fourth Amendment rights in records solely useful to the government to identify that individual's physical location at a particular time. The question would be objective: Is that category of record something that ordinarily is solely useful to the government to determine a person's location? If the nature of that kind of record means that it is of a type ordinarily only of government use to identify a person's physical location, then it is protected by the Fourth Amendment unless there has been consent to the search. On the other hand, if government collection of that kind of record ordinarily has a non-location purpose, then this special rule would not apply and the third-party doctrine would continue to apply.

Why that rule? I think it identifies the concern the Justices have about physical location tracking and adjusts for that kind of record. Granted, it's tricky to craft a rule that is just focused on location tracking and that is relatively clear for lower courts to apply. The problem is that there are hundreds of different kinds of non-content records out there that are currently deemed unprotected under the third-party doctrine. Many of the records can allow some kind of inference to be drawn about a person's location in at least some cases. And when those records are in the hands of third parties, the third parties usually have a purpose other than to identify location.

At the same time, relatively few of those records are solely useful to the government for purposes of determining a person's location at a particular time. Most records have other uses. In an investigation, obtaining records of a credit card purchase shows the purchase (even if it may also suggest the person was at the store). Obtaining records of numbers dialed from a home phone shows the call (even if it also suggests the person was home). Tailing a person in the physical world shows who they are, what they are doing, and who they are with (even if it also reveals their location). In contrast, GPS tracking from the Jones case is solely useful to the government for purposes of determining a person's location. And historical cell-site records are solely useful to the government for purposes of determining a person's location, too.

A decision that engages in equilibrium-adjustment for the category of records solely useful to the government for purposes of determining a person's physical location would be relatively focused on the problem motivating the adjustment and would be relatively clear in most cases to apply. The idea wouldn't be to say that a person has a reasonable expectation of privacy in their location in some abstract sense. That's too high a level of generality. And it raises all sorts of weird questions about public observation. Instead, the idea would be that government collection of location records—records solely useful to the government to learn a person's physical location—infringes a reasonable expectaton of privacy when the person has not first consented to it.

(As an aside, if you think there are records that reveal location but are also generally useful to the government for a secondary reason but that should also be protected, you could also imagine a rule that focuses on whether the records are predominantly useful for location information. But off the top of my head, I don't know what those records would be. And a solely useful test seems significantly clearer.)

I'll be the first to admit that there are cases that will present difficult judgment calls under my proposed rule. Part of the problem is that identifying the nature of a record can itself get complicated. Records tend to include two categories of information: The event that triggered the creation of the record, and one or more facts about that event that were expressly or impliedly in the record. So you need to consider, what determines the nature of the record: the triggering event or the fact? And if the latter, which fact? And what determines the use of the record: Use in that kind of case, use in criminal cases generally, government use in all instances? In some cases, the answers won't be obvious. But I'm making only a comparative claim. Compared to other possible rules that I can think of in a hypothetical decision ruling in Carpenter's favor, my proposed rule seems less tricky than the alternatives.

Along these lines, the Court should also squarely reject the mosaic theory of long-term versus short-term monitoring. It's totally unworkable. Any rule the Court adopts should apply across the board without distinguishing between long-term or short-term collection. Records that are protected should always be protected, at least assuming no consent or some other exception.

I gather one consequence of my proposed approach would be that the Court would likely need to overturn United States v. Knotts, the 1983 beeper case. You could maybe try to keep Knotts alive on the theory that the beeper was trying to locate a drug lab, not the location of any particular person. But if you (plausibly) see beeper signals as a category of record that ordinarily is solely useful to the government to determine a person's location, then Knotts would have to go.

If so, I think that's no big deal. Most Knotts-like cases today already come out the other way under the trespass holding of Jones, making Knotts of uncertain practical importance today. And more broadly, having to overturn Knotts is consistent with how equilibrium-adjustment often works. As technology and social practice change, and a fact pattern that was once rare becomes very common, the need for equilibrium-adjustment changes, too.

III. The Rationale—Why Is It Protected?

As I discussed above, the thinking behind my proposed rule is that it is tailored to the technological change driving the equilibrium-adjustment. With that said, the Court's opinion must still use the models to justify the rule. And in particular, it needs to choose which models to use. This is an important choice, as it determines the framework lower courts will use to apply the Fourth Amendment in related cases. The Supreme Court picks the models in the few cases, and those models become the tools lower courts use by analogy in similar cases that lower courts will encounter. As I wrote in my Four Models paper:

Fourth Amendment decision making relies heavily on analogies, which means that legal reasoning adopted in one factual setting will tend to be adopted in similar factual settings. In a system of coexisting models, this dynamic allows the Supreme Court to pick the model most likely to accurately identify police practices in need of constitutional regulation in the setting of that particular case. Lower courts faced with new cases will reason by analogy to the Supreme Court's precedents, and through analogical reasoning will incorporate and adopt the models used in the Court's most similar decisions. The result is a system of localized models: different models will control different types of cases depending on which model best draws the line in each particular type of case.

So which models work best in Carpenter? In my view, a ruling in favor of Fourth Amendment protection would best rely on the private facts model and the policy model. In contrast, it shouldn't rely on the probabilistic model or the positive law model.

Here's my thinking.

First, reliance on the policy model is relatively straightforward. Most Supreme Court opinions on the reasonable expectation of privacy test have relied on the policy model. This is where the Court could make arguments about equilibrium-adjustment discussed above.

Second, the private facts model is often a good choice in technology cases. In areas of new technology, the details of how the information is collected can be contingent and unstable. Focusing a rule on the kind of information that is collected, rather than the details of how it is collected, is often a more stable and consistent approach. By relying on the private facts cases (see here at pages 512-14), the Court could say that these new technologies give the government new powers to track a person and thus learn very sensitive things about them; that the general category of information solely useful to the government to determine a person's location raises special privacy concerns; and that therefore, under the private facts cases, a person has a legitimate expectation of privacy in those records.

I've already written at length why I think the positive law model and probabilistic model are the wrong ones to use in Carpenter. I argue against using the positive law model here, and I argue against the probabilistic model in part three of this post. A common theme of my criticisms is that, in areas of new and evolving technology, the positive law and probabilistic models are too contingent and unstable to rely on. Legislatures often change the positive law governing new technology. Social understandings of technology and its uses can change over time. A decision that relies heavily on the positive law model or the probabilistic model therefore risks considerable confusion.

That's true both as to the effect of changing positive law and social understandings on the facts of Carpenter and, perhaps more importantly, on the facts many other cases that are coming down the pike. If the Court rules for Carpenter, the flood gates will open for claims of similar treatment with many other records. Indeed, the gates may have opened already with this cert petition on IP addresses filed earlier this week in the Silk Road case that anticipates a ruling for Carpenter. If the Court wants expanded Fourth Amendment protections, it will want a stable basis for both Carpenter and future rulings by lower courts in related areas. The policy model and the private facts model are the best way to provide that. The probabilistic model and positive law models are not, in my view.

IV. Whether to Take on Reasonableness

A final question is whether the Justices should also answer the standard for constitutional reasonableness—and if so, how they should rule. Is the reasonable suspicion required by the Stored Communicatiosn Act sufficient or is probable cause required? If probable cause is required, is a warrant needed? In an ordinary case, you'd think the Justices would decide that set of issues if they rule for Carpenter on whether he had rights at all. The reasonableness issue was preserved below; it was within the scope of the cert grant; and it was fully briefed by the parties before the Justices.

At the same time, there are some good practical reasons not to decide that issue in Carpenter. Because the Court granted cert without a split, there has been no percolation of the issue in the lower courts. The issue barely came up at oral argument, as the Justices were heavily focused instead on the threshold question of whether the Fourth Amendment applies.

Further, it's not an easy issue. It's plausibly path-dependent on the scope of the (as of yet unknown) search rule. A narrower search rule is more consistent with a higher reasonableness standard, while a broader search rule is more consistent with a lower one. Perhaps different location records merit different reasonableness standards. And either way, there are some tricky implementation questions.

I don't have strong views on this. But it may make more sense to leave the reasonableness standard for another day if no key votes are needed on the that issue to get to a majority opinion on the search question.

Further, the scope of the good faith exception arguably makes the need for a ruling on reasonableness somewhat less pressing than it would ordinarily be. A lot of this area of law is already the subject of statutory protection. Under the good faith exception of Illinois v. Krull, the exclusionary rule does not apply to searches authorized by statute unless the authorizing statute is "clearly unconstitutional." That obviously applies to access to historical cell-site records predating any ruling in Carpenter. Carpenter won't get out of jail. And neither will anyone whose records were obtained before the Carpenter case was handed down.

But if the Court decides that Carpenter was searched but declines to rule on reasonableness, I would think the exclusionary rule won't apply ex post either. Because the ruling will leave unclear whether applying the SCA's rule to cell-site records is unconstitutional, the statute won't be "clearly unconstitutional" in that application. Some officers and agencies will be cautious and won't rely on the statute, at least if they have more cause, in light of the uncertain Fourth Amendment standard. But the exclusionary rule shouldn't apply going forward under Krull unless and until there are rulings on reasonableness saying that more cause is required.