"Not a Suicide Pact": Round Two
Debate Series: Stone Offers A Proposal
I don’t agree that the public “has already surrendered much of its communicative privacy by its profligate use” of modern means of technology that “create essentially indelible records” of our communications, purchases, etc. Certainly, it’s true that most people have embraced cells phones, email, and the internet without paying much, if any, attention to the extraordinary invasions of privacy they make possible. But this will change once people come to understand how vulnerable they are. It’s a bit like electronic bugging and wiretapping in the first half of the twentieth century. It took fifty years for courts and legislatures to begin regulating such conduct, but once people realized the danger, government electronic surveillance was declared unconstitutional and private electronic surveillance was declared unlawful. The same will happen with respect to the modern means of communication. Once people recognize the danger, they will insist on regulation. So, I wouldn’t leap to the conclusion that the public has “surrendered” its privacy. We are merely in transition.
(again, rest of Stone and response from Posner after the jump)
Judge Posner suggests that “the essential protection against governmental abuse” must lie in rules that (a) forbid use of the information “other than for national security purposes” and (b) require oversight by congressional committees and neutral agencies. Recent experience teaches that such safeguards are porous, at best. It is useful to recall that one of the bete noirs of the Bush administration was the so-called “wall” created by the Department of Justice to prevent information obtained in lawful foreign intelligence surveillance authorized by the FISA court from being used for anything other than national security purposes. The purpose of the “wall” was to allow the government special flexibility in foreign intelligence investigations without creating an incentive for the government to “cheat” by using FISA warrants for ordinary criminal investigations. The “wall” was one of the first safeguards abolished after 9/11 because it was attacked as perverse. Of course, it wasn’t perverse at all. To the contrary, it served exactly the function that Judge Posner now urges. But the “wall” crumbled as soon as pressure was placed on the system. And oversight is hardly a panacea. Congressional oversight since 9/11 has turned out to be meaningless, at least at a time when the executive and legislative branches are in the same partisan hands.
Nonetheless, there is a compelling need for some form of surveillance that is broader than what would be permissible under a strict probable cause regime. In Not a Suicide Pact, Judge Posner offers an interesting suggestion: “Suppose that the National Security Agency’s listening devices gathered the entire world’s electronic communications traffic, digitized it, and stored it in databases, where it was machine-searched for clues to terrorist activity, but the search programs were designed to hide from intelligence officers all data that contained no clues to terrorist plans or activity.”
This might be the foundation for a compromise on which Judge Posner and I could agree. Suppose the NSA were authorized by legislation to intercept (in the sense described by Judge Posner) any international electronic communication for which it has reasonable grounds to believe that a participant in the communication is associated with a terrorist group. (The terms of course would need to be defined.) All such information would be digitized and stored (for a limited period of time), but could not be examined by any intelligence officer without probable cause. Probable cause could be determined either in the conventional manner or through the use of an algorithm that enables computers that search the database to determine probable cause based on words, phrases, personal connections, etc. In a sense, this is like a drug-sniffing dog, where the algorithm is the dog.
It is easy to imagine such algorithms. For example, it would be quite possible (I would imagine) to develop an algorithm that would determine based on the content and nature of messages whether there is probable cause to believe that a particular individual is a lawyer, or a doctor, or a psychotherapist. (How often does she use phrases like res ipsa loquitur and mens rea?)Presumably, it would also be possible to develop an algorithm for terrorists. The FISA court would have the authority to approve and review the use of such algorithms after hearing expert testimony.
Such a program would both substantially increase the ability of the government to identify terrorists (relative to a conventional probable cause standard) and substantially reduce the dangers of invasion of privacy and government abuse (relative to the NSA surveillance program). Moreover, I would add to this system Judge Posner’s proposals for a “wall” and for various forms and levels of oversight. I would be very interested to hear Judge Posner’s response.
Debate Series: Posner Responds to Stone Proposal
I am sympathetic to Geof’s proposal (though it doesn’t go as far as I would like), but before discussing it let me register disagreement with three points, closely related to each other, that he makes:
1. I don’t think the surrender of privacy to companies is based on ignorance. After all, Geof and I are among the surrenderers, and I daresay most members of the ACLU are as well. I think it’s based, first, on the fact that Americans don’t actually value privacy as much as the ACLU thinks, and, second and more interesting, that they don’t worry a great deal about companies’ invading their privacy because they realize that the purpose of the invasion is benign: it is to be able to sell consumers more products better matched to consumer preference. They do not have the same trust in the benignness of government surveillance. Nor should they, because consumer sovereignty does not describe the relation between government and the people. It is naïve to think that American-style democracy securely aligns popular preferences with the incentives of government officials and employees, whether in national security or any other domain of governmental activity. That is why I would couple extensive electronic surveillance with the two protective measures that I mentioned in my first post—forbidding use of intercepted communications for any purpose other than national security (and thus limiting, as Oren Kerr has proposed, the “plain view” doctrine of search and seizure law), and requiring that the details of each interception (who, why, what, and with what consequences) be reported periodically to watchdog committees in Congress, to the GAO, and to departmental inspectors general.
2. Would these controls be perfectly effective? Of course not. But then nothing is. I do not, however, consider the notorious “wall” that Geof mentions to be analogous. I must say first of all that I have never been able to determine whether there was a wall or, rather, whether it was simply a matter of the FBI and the CIA being unwilling, for bureaucratic reasons, to share information. The sharing of information between government agencies is impeded by the fact that one agency cannot sell its information to another; it can only swap; and barter is a clumsy method of exchange. But there is in any event a huge difference between, say, the CIA’s failing to give the FBI information about a terrorist that might have enabled the FBI to arrest the terrorist before he flew a plane into the World Trade Center, and the CIA’s (or more likely the NSA’s) failing to give the FBI information about a nonterrorist that would enable the FBI to prosecute the person for killing an animal on the endangered species list.
3. Congressional oversight of intelligence and related national security programs has always been weak. I don’t think it has made a difference whether Congress was controlled by a different party than the presidency. It is weak, as I explain in chapter 8 of my book Uncertain Shield, partly because congressional oversight of the executive branch is weak in general, partly because members of Congress tend to focus on domestic matters, partly because few of them have much knowledge about intelligence, and partly because oversight is split up among a large number of agencies, owing to the fact that the agencies themselves are distributed across a number of departments. I do not agree, however, that Congress is a mere rubber stamp, even when it in the hands of the same party as the presidency. We have seen this in recent months and weeks as Congress and the President have tussled over military tribunals and electronic surveillance.
Coming now to Geof’s proposal, I understand him to be suggesting that the NSA be authorized by new legislation to intercept international communications if there are reasonable grounds to believe that at least one of the parties to the communication is a terrorist. The restrictions I have italicized strike me as too limiting. If two terrorists inside the United States communicate with each other, the interest in intercepting their communications is as great as when one of them is overseas. And, as important, it is not possible to discover who the terrorists are if before intercepting their communications you must have reasonable grounds to believe that at least one party to a communication is a terrorist. What I think national security requires is a two-stage process. In the first, computer search programs search the world’s entire daily electronic traffic (to the extent feasible) for messages that are suspicious because of names or word clusters in the message, social security numbers or other personal identifying information besides names, the origin or destination of the message, and other suspicious characteristics. These messages, a minute fraction of all those screened by the search programs, would be listened to or read (as the case may be) by (human) intelligence officers. Computer screening is not a search, because a computer is not sentient. But listening to or reading private messages is, and, as Geof suggests, a properly configured search program might be sufficiently reliable to furnish reasonable suspicion or probable cause to read a message flagged by the computer program. Although probable cause is required for a warrant to conduct a search for evidence of ordinary crime, the Supreme Court in the Keith case suggested that the requirement might be relaxed in a search for evidence of conduct that endangers national security, even if the search involves intercepting a purely domestic communication. In such a case reasonable suspicion might suffice.