New Draft Article: "Implementing Carpenter"

From my forthcoming book, The Digital Fourth Amendment.

I recently posted a new draft article, "Implementing Carpenter," on the Supreme Court's blockbuster June 2018 decision in Carpenter v. United States. The article consists of two draft chapters of a forthcoming book, The Digital Fourth Amendment, that will be published by Oxford University Press. I'd love your critical feedback. I found these chapters tough to write, and the line-drawing exercises Carpenter requires very challenging. But I did my best to make them and justify them, and I'd rather you throw rotten fruit at me now instead of when the books comes out. So if you have feedback, please send it on.

The document I posted has two chapters. The first chapter, The Carpenter Shift, explains how Carpenter takes the Fourth Amendment in a new direction and explains the new state of the law. After explaining the conceptual basis of Carpenter and why it's a considerable departure from prior law, the chapter tries to reduce Carpenter to a doctrinal test. Based on a close read of the opinion and the broader theory of equilibirum-adjustment driving it, the chapter argues that Carpenter applies to non-content Internet records otherwise left unprotected when three conditions are met. First, the records must be new kinds of records of the digital age. Second, the records must not have been generated by meaningful voluntary choice beyond what is necessary to participate in modern life. Third, the records must be of a type that can reveal an intimate window into a person's life.

The second chapter, Implementing Carpenter, applies those general principles. It first focuses on the challenging question of how to identify a Carpenter search in a particular case. How do you measure a privacy invasion? Does a search occur when an intimate fact was actually revealed in an investigation? Does it occur when the government gets enough records that the revealing of an intimate fact would be expected, the so-called Mosaic Theory? After going through the pros and cons of different approaches, the chapter concludes that the best way to measure an invasion of privacy is a source rule: Any government collection of any amount of Carpenter-protected information, no matter how small or unilluminating in a particular case, should be treated as a search.

Finally, the chapter applies Carpenter to several important cases. It identifies two kinds of Internet non-content metadata that should trigger Carpenter: to/from information about messaging services such as e-mail and text messages, and monitoring the websites a person visits. When the government wants to conduct surveillance of who a person e-mailed or messaged, or wants to install a monitoring device to see what websites a person is visiting, collecting that metadata should be a Fourth Amendment search. It also identifies a few examples of metadata collection that should not trigger a search: acquisition of voice call metadata, the IP addresses a person was assigned while connected to the Internet, and records of ride-sharing services such as Uber of Lyft. The chapter concludes by arguing that downstream analysis such as datamining should not itself trigger a search, although the prospect of downstream analysis can change whether a particular record is protected under Carprenter and can trigger the Fourth Amendment upstream for all compelled acquisition of that kind of record.

Comments very welcome. Thanks as always for reading.

(Cross-posted at Lawfare)

Comments

[Dilan Esper]

I will have to read the article, but I find the Uber / Lyft conclusion problematic. A traditional taxi can be paid in cash and leaves no record whatsoever, so I would think the default rule should be since the customer has no cash option here, the records generated by these services should be private so the government doesn't get to know everyone's movements

[NToJ]

I will read the article too.

I am as guilty of this as other readers. It saddens me that a substantive article about constitutional law gets virtually no discussion, but if you mention some hot-button partisan issue with a dash of legal thought, there will instantly be 100+ comments. I am concerned this will lead to less posting by Professor Kerr since he is (so far as I can tell) the least partisan author here. That would be a loss for Volokh's readers, generally.

I am beginning to think that this law blog has just become another boring culture war battleground.

[Rich Rostrom]

Hmm... It occurs to me that this may not be entirely new.

Back in the day, was there ever a case where investigators asked the Post Office to record and provide to them the addresses of all mail sent by a subject? (Return addresses might also be wanted, but return addresses could be omitted or easily spoofed.)

Would such surveillance constitute a search? Bear in mind that when one hands in an addressed mail item, one is supplying that address to an agency of the government, which would surely bear on whether that address could be properly supplied to another agency of the government. That consideration could carry forward into the present day: the subject's E-mail provider might be a government agency, but not the subject's employer such as a college or K-12 school.

[NToJ]

It didn't matter pre-Carpenter because the 3rd party doctrine didn't depend on the person you shared the information with being "a government agency" or even an agent. If you share information with your employer, the 4A doesn't stop the employer from voluntarily sharing it with the government.

"Would such surveillance constitute a search?"

No. Otherwise the post office would need a warrant to deliver the mail.

[Turtle Dove]

Interesting chapters.

Physically, historical cell site records are stretches of binary scattered on a few drives, assembled at the machine level, and rendered readable. But this is now an essentially arbitrary assembly and markup, one that might itself only be a part of a larger document. It exists as a simulacrum of the familiar report, much like the ones Miller had in the file cabinet, but now in an essentially arbitrary form.

Perhaps Carpenter holds that the potentially invasive uses of data have increased to the point at which the legitimate expectations that shield a Miller-type document no longer sufficiently protect the underlying data from exploitation. What is really being requested is a compilation of the millions of traces of binary on spinning silver, and a new, genuinely held legitimate expectation of privacy arises not from the form of the documents, but from the possibility of larger documents, and where the data is freely fungible, and perhaps not used in the end for the truth of the matter that the document asserts, the Carpenter data should be protected separately from the Miller documents. The problem isn't that the data is so big, but that the datum within it is so small, ubiquitous and easily repurposed.

Perhaps the scope of the search must be proportional, both on the document and the data level. The shift is from Guineas to Bitcoin -- a categorical exchange rate might not reach these qualitative distinctions between the two.