The "Mosaic Theory" and the Aftermath of Carpenter

by Matthew Tokson

Recently, Orin Kerr has raised some interesting questions about how lower courts are using the "mosaic theory" of the Fourth Amendment after Carpenter v. United States. The mosaic theory generally refers to the idea that courts should look at police surveillance in the aggregate, rather than examining each instance of surveillance in isolation. So tracking someone's car for 3 months might require a warrant, even if tracking them for 3 minutes would not. 

Kerr notes that some lower courts have accepted the mosaic theory while others reject it, and wonders how many there are of each. As it happens, I have a coded database of every available decision citing Carpenter (for another project), and can speak to some of these questions. After doing so, I'll offer a suggestion for distinguishing two conceptions of the mosaic theory to make sense of what the lower courts have been doing.

The first notable thing is that, per Westlaw, only 7 of the over 600 Carpenter-citing decisions mention the phrase "mosaic theory" or some variant thereof. One of these only mentions it in passing. Of the other six, three reject it (two of these are different opinions in the same case), two accept it, and one criticizes it before essentially applying it. In any event, the term has not exactly caught fire in the lower courts. It's far more common for courts to simply assess the amount of data or duration of surveillance without expressly invoking the "mosaic theory." 

Speaking of which, what about the cases that accept or reject the mosaic theory without actually using that term? Here, the numbers are more lopsided. (These preliminary numbers are taken from the roughly 550 cases coded to date). Of 139 opinions that applied Carpenter to reach a novel conclusion about a search, 71 accepted analyzing the amount of data gathered (45 expressly analyzed the aggregate amount of data collected by the government, while another 26 discussed aggregate amount as a key factor under Carpenter). Zero of the other 68 cases rejected the idea of analyzing amount. Instead courts resolved these cases in cursory opinions or on other grounds, often finding no search after citing Carpenter's language about being a "narrow" decison that doesn't apply to traditional surveillance methods. 

So the lower court count is something like 73 opinions in favor of the mosaic theory vs. 3 opinions against. Of course, judicial dislike of the theory might be hard to detect--some courts might dislike the concept without going to the trouble of rejecting it. And a handful of decisons may be in tension with a robust mosaic theory, without directly contradicting it. But acceptance of some form of the mosaic theory seems to be very widespread.

Why might this be? 

Let's start with why most courts don't use the term "mosaic theory" and instead just speak in terms of the amount of data collected. For one, "mosaic theory" is a confusing term, while "amount" is clearer. The mosaic theory has both a strong form and a weak form, and the two concepts are very different in practice. The strong form mosaic theory would ask courts to consider every bit of information that the government has on a person--so if the government already knows that John sees a podiatrist and eats pretzels, it might be a search if the government also examines his utility records--because it is using all of these disparate points of information to build a mosaic of his life. The weak form mosaic theory just looks at how much information the government is seeking when it conducts some type of surveillance--so tracking a suspect's location via cellphone for 40 years might be more intrusive than doing so for 2 seconds. Kerr refers to this type of analysis as mosaic theory analysis as well. 

While the strong form mosaic theory has not been broadly adopted and would likely have serious administrability problems, assessing the amount of data collected is widely accepted by lower courts today and has likely driven Fourth Amendment outcomes since 1967's Katz v. United States

Another reason to favor amount is that the amount of data collected is just one of several factors that courts care about when assessing a case. Other factors that seem to play a substantial role in Fourth Amendment cases include the intimacy of the place or thing searched and the cost of the surveillance. My recent work on this posits that the Supreme Court has relied heavily on intimacy, amount, and cost in virtually all of its post-Katz cases. In any event, amount is just one of several factors, one that does not dictate every outcome.

When viewed this way, lower courts' overwhelming approval of amount makes more sense. Lower courts appear to be perfectly capable of distinguishing between different durations or quantities of surveillance. For example, in United States v. Monroe, the district court found that the government's collection of an IP address associated with an account that downloaded images of child pornography was not a search in part because it was "not an 'exhaustive chronicle' of [Monroe's] physical or digital activities." By contrast, in United States v. Diggs, the district court held that the government's tracking the defendant via GPS for over a month was an important factor that favored finding a Fourth Amendment search. 

Few problems have materialized, aside from the usual line-drawing "problem," which can be found in virtually every area of law. Also, Carpenter itself seems to strongly endorse assessing the amount of data gathered. It emphasized that the third-party doctrine did not apply because the government had obtained "an all-encompassing record of the [cell phone] holder's whereabouts" and an "exhaustive chronicle" of their location that effectively operated "every day for five years." Carpenter was deeply concerned with the amount of data gathered, and lower courts are likely to continue to follow along this path. 

So I'd propose a synthesis--courts should reject the strong-form mosaic theory, for all the reasons Kerr gives. But they should continue to assess the amount of data collected, for all the reasons given here.