Standing in the 2nd Circuit Bulk Metadata Case

by Michael Dorf

Last week, in ACLU v. Clapper, the U.S. Court of Appeals for the 2nd Circuit ruled that the government's collection of bulk metadata (information about all calls placed and received in the U.S. but not the contents of the conversations) violates Section 215 of the PATRIOT Act. The Court did not reach the question whether the program (which was revealed by Edward Snowden) violates the Fourth Amendment. As Professor Orin Kerr noted on Volokh/WaPo last week, the direct practical effect of the merits decision (with which he agrees) is limited, due to the pending expiration of the Act. But as Professor Kerr also acknowledges, the ruling could affect the congressional debate over reauthorization.

Here I want to linger over the standing analysis. The opinion by (my former colleague) Judge Jerry Lynch had to overcome the initial obstacle of the Supreme Court's 2013 decision in Clapper v. Amnesty Int'l USA, which held that people whose speech was chilled by the fear that the government was monitoring their electronic communications had only suffered "self-inflicted injuries," and thus lacked standing to sue in a federal court. Judge Lynch pointed to what one might think is a powerful distinction between the current suit and the 2013 case: By contrast with the Amnesty Int'l USA plaintiffs, whose case was decided by the SCOTUS before the Snowden revelations, here the plaintiffs know for certain that they are being monitored because the bulk metadata program scoops up everybody's metadata.

But the government argued that scooping up the metadata is not by itself an injury because the government ignores nearly all of the metadata it collects. Only when the government has "reasonable articulable suspicion" does the government query the database about particular calls. The odds that any random individual's metadata will be examined are quite low, certainly below the probablistic threshold for injury in Amnesty Int'l USA.

Judge Lynch has two responses to the government's argument. First, he says that the collection of the data itself is an injury because it is, in the terms of the Fourth Amendment, a "seizure." As Professor Kerr notes, it's not entirely clear why Judge Lynch is talking about the Fourth Amendment for standing purposes. The relevant question for standing is whether the metadata collection is an "injury." Accordingly, I think Judge Lynch can be best understood to be saying that the inclusion of "seizures" in the Fourth Amendment establishes that a seizure, even unaccompanied by a search, injures the party whose person or property is seized.

That's fair enough, but I'm not sure that a metadata seizure is the same as a seizure of the person or of tangible property. Suppose the government seizes Joe's briefcase. Even if the government never opens the briefcase to see what is inside, the seizure of the briefcase deprives Joe of the use of the briefcase during the period that the government holds it. Not so with metadata. When the government copies the metadata from the Verizon computers to its own computers, it doesn't even deprive Verizon of the metadata, much less Verizon's customers.

To be sure, when the government takes someone's metadata, that person feels anxiety because of the risk that the government will look at the metadata. If I were writing on a clean slate, I would probably say that this anxiety is enough to establish standing. But given Amnesty Int'l USA, that line of reasoning appears to be foreclosed.

Judge Lynch has a second argument, though. He says that whenever the government performs a suspicion-based search on a particular suspect
its computers search all of the material stored in the database in order to identify records that match the search term.  In doing so, it necessarily searches appellants’ records electronically, even if such a search does not return appellants’ records for close review by a human agent.  There is no question that an equivalent manual review of the records, in search of connections to a suspect person or telephone, would confer standing even on the government’s analysis. That the search is conducted by a machine might lessen the intrusion, but does not deprive appellants of standing to object to the collection and review of their data.
That last move strikes me as dubious. In Illinois v. Caballes, the SCOTUS held that being subject to a dog sniff is not a search for Fourth Amendment purposes because the dog only alerts in the presence of contraband. The dog may learn that the suspect hasn't bathed recently or ate garlic for dinner the previous night, but the dog does not convey this information to any human, and so there is no invasion of privacy. Even if one thinks that the Fourth Amendment should protect people against being forced to divulge private (non-criminal) information to dogs as well as to humans, presumably that's because dogs have minds. Given current technology, computers do not. The computer doesn't learn anything about someone whose metadata it sorts because computers are not capable of learning in the relevant sense. Computers can "learn" in the sense that they can be programmed and even in the sense that they can, with experience, get better at a task (such as predicting what songs or movies you'll like), but outside the world of science fiction there is no evidence that modern computers subjectively experience the world.

Of course, Caballes only establishes that being sniffed by a dog isn't a search for Fourth Amendment purposes. It can still be an injury for Article III purposes. Likewise, we might think that having one's metadata sorted as part of an effort to get at someone else's records is not a search for Fourth Amendment purposes, but is still an injury for Article III purposes. But why? In what way does someone suffer any sort of setback to his interests when a computer program sorts through his data in the course of identifying someone else's data for examination by humans?

Some people might feel injured by that kind of sorting, I suppose. After all, people have varying intuitions about what counts as a violation of privacy. Consider a provocative example offered by my soon-to-be colleague Professor Andrei Marmor in a recent article in Philosophy & Public Affairs. (You need a subscription to follow the link.) He asks the reader to suppose
that we came to learn that there is an alien civilization, galaxies away, that is watching us closely; they cannot establish any contact with us, and never will, but they are watching. Some people might find it disturbing, others might not; either way, it would be very difficult to point to any interest of ours that might be affected here. If we cannot establish any contact with the aliens, then we cannot have an interest in how we present ourselves to them. How would our lives go less well, in any sense whatsoever, by knowing that the aliens are watching? I take it as a confirmation of my argument that it results in the conclusion that the aliens are not violating our right to privacy.
I do not share Professor Marmor's reaction to this hypothetical example. To me, the mere knowledge that the aliens are watching could be harmful in itself. Indeed, one might even think that a person thus observed is harmed--injured for Article III purposes--by the alien observation even if the person never learns that the aliens are watching. A variant of that particular puzzle is presented by the underrated 1993 thriller Sliver, in which the voyeuristic owner of an apartment building secretly observes his tenants via hidden cameras. If the tenants never learn of the voyeurism, has their privacy been violated?

When I have discussed examples like these with students in my constitutional law and federal courts classes, reactions tend to vary. People might also have diverse intuitions about the question of whether having one's data sorted by a computer counts as an injury on privacy or other grounds. As one who thinks that the SCOTUS standing doctrine is overly restrictive, I would like to see the Court uphold the Second Circuit's standing decision. But if I had to bet, I'd guess that a majority would be inclined the other way.