In a series of earlier posts (here, here, and here), I suggested that big data is going to pose problems for privacy, insofar as privacy depends on a distinction between information and data. Here, I want to look at how that problem plays out in a specific 4th Amendment case on thermal imaging devices.
In 2001, Justice Scalia, writing for a 5-4 majority in Kyllo v. U.S.,struck down the use of thermal imaging devices without a warrant. Danny Kyllo grew marijuana inside his home, an endeavor that involved the use of high intensity lamps. A police officer had used a thermal imaging device from his squad car on the street to detect the heat from the lamps. On that basis, the police obtained a warrant to search the home. The question before the Court was thus whether the original use of the thermal imaging device constituted a “search.” Scalia reasoned that the Court had consistently held that “visual surveillance” did not constitute a search. However, “the present case involves officers on a public street engaged in more than naked-eye surveillance of a home. We have previously reserved judgment as to how much technological enhancement of ordinary perception from such a vantage point, if any, is too much.” Scalia reasoned that this case crossed the line:
obtaining by sense-enhancing technology any information regarding the interior of the home that could not otherwise have been obtained without physical ‘intrusion into a constitutionally protected area,’ constitutes a search--at least where (as here) the technology in question is not in general public use. This assures preservation of that degree of privacy against government that existed when the Fourth Amendment was adopted. On the basis of this criterion, the information obtained by the thermal imager in this case was the product of a search (internal citations omitted).
Much of the concern around the case lies with the qualifying “not in general public use” – since it sounds like it ratchets legal permissibility to a rapidly changing technological standard. What interests me here is that Scalia’s argument revolves around thinking that some data is per se information, or naturally occurs as information (rather than data). There is no further work required to make that data into information, and only in those cases do we automatically consider something in “plain view” such that it can be seen without an information-producing “search.”
The problem with naturalizing vision, of course, is doing so occludes the sense in which all information involves the representation of data. This, I take it, is the basic point made by phenomenology from Husserl forward: to see is always to see “as” something, which is to say that what I see is never going to be fully separable from my capacity to represent it. We know that this representative capacity is heavily influenced by the environment in which it finds itself: this is the point of any standpoint argument, as well as the evidence describing cultural cognition. But the evidence is pretty good that even the way our brain presents seemingly neutral sensory input relies on cognitive shortcuts (a point made forcefully by Andy Clark). In short – and I don’t think there’s anything either controversial or new in making the point - there’s good reason to think that nothing is per se information: to see it as information is already to process it.
The reason the point matters in this context is that it poses a problem for any legal regime that depends on a distinction between data and information. Scalia mocks the dissent for its proposal that the thermal imaging device “makes an inference” about what it “sees,” but that seems to be more of an artifact of the machine processing things not ordinarily on the visible spectrum than any sort of legal principle. Indeed, this doesn’t work as a legal principle at all. On the one hand, as noted above, the Court had already signed off on a couple of cases (Florida v. Riley and California v. Ciraolo) which are easily read as the reductio of this principle. In those cases, the pot-growers had shielded their plants from the street (with a fence or a greenhouse). The police flew over with some optical technology (in one case at 400 feet, in the other at 1000), saw the plants, and obtained their warrants. In explaining why this was not a search, the Court proposed that the plants would be visible to anyone flying through the airspace on a commercial plane. It would be very hard to argue that there was less technological augmentation of vision in this case than in the thermal imaging case, and certainly the big lenses “represent” the view of the greenhouse as much as the thermal device. It’s only a difference in frequency.
On the other hand, the entire jurisprudence of “plain view” becomes unstable. If you’re a police officer looking for something, it’s to your advantage to argue that it’s in “plain view,” since then seeing it doesn’t require a search: plain view is per se information (see Poe’s “Purloined Letter” for literary confirmation). In a different case, however, the Court considered a situation where the police entered an apartment looking for weapons, noticed the very fancy stereo in the very un-fancy apartment, looked at the back of the turntable to get its serial number, called that in, and thereby discovered the TV was stolen. The Court ruled that the numbers were out of plain view, and so required probable cause to investigate.
The cases can be reconciled if one assumes that anything accessible to something like line of sight, no matter the technological enhancement of vision required, is in “plain view” or otherwise per se informative. Anything else, from turning the object around to detecting infrared spectrum, constitutes an act of representation that triggers the juridical determination of a “search.”
The problem is that if we admit the phenomenological point, and say that all seeing is seeing-as and therefore involves an act of representation, then either everything is a search, or nothing is. On the one hand, for me to see that you’re growing pot on the front porch requires that I know what pot looks like (and what if I normally wear contact lenses?), and perception is in any case an extremely complicated process. Acknowledging this complication could easily overwhelm the “plain view” category. On the other hand, if we move in the other direction, and decide that since vision is representative, then other technologies are like vision, it will be very hard to establish any sort of non-arbitrary limit to surveillance. In other words, if you eliminate the category of the per se informative, Scalia’s argument in Kyllo no longer works. And as odd as it is to say this, our privacy jurisprudence in an age of big data needs Scalia’s argument to work.