Last week, the House Judiciary Committee marked up its version of a bill to reauthorize and reform some expiring provisions of the USA PATRIOT Act. Plenty of interesting things happened there, and I’ll write about them elsewhere. But I want to try to draw some special attention to a rather subtle change I only just noticed. What actually appears to have happened is that, in the guise of a “clarifying” technical amendment, the Committee radically narrowed a change to PATRIOT roving wiretap powers, while conspicuously attempting to deny they’d done so. This is especially remarkable because, on other fronts, there was open and vigorous debate over how far they wanted to go in reforming surveillance powers. This gets a bit convoluted, but I hope you’ll bear with me for a bit of background.
The original House bill, H.R. 3845, sponsored by Reps. John Conyers (D-MI) and Jerrold Nadler (D-NY), renewed “roving wiretap” authority under the Foreign Intelligence Surveillance Act, to be used when targets are thought to be swapping phones or e-mail accounts or other facilities, but made some important changes to close a potential loophole. I discuss it at some length in this Cato post, but the short version is that the current language allows a “roving” warrant to be issued based only on a description of the “target,” where the “target” can be not only an individual, but “any group, entity, association, corporation, or foreign power.” Now, when you’re actually naming the facility to be surveilled—a particular phone line, account, server, physical location to bug, or whatever—this is maybe not such a big deal. If you legitimately have probable cause evidence that this place or this server or this phone are being used by (say) al-Qaeda, maybe that’s enough without further demanding that the applicant say which precise individuals are going to use it. But when you’re not naming a specific line or location, suddenly a warrant that contains nothing more than a “description” of an “entity” starts sounding an awful lot like a general warrant. Civil liberties advocates dubbed these “John Doe warrants.”
Conyers/Nadler originally fixed that by requiring FISA applications for electronic surveillance to always specify a single individual target. Now, for the reason I sketched above, I think that actually goes further than is necessary: Again, if you really have probable cause that al-Qaeda agents are operating on this server or in this warehouse, it’s probably unnecessary to ask in advance for particular individuals to be identified. But the legislative summary circulated last month made it crystal clear that, whether or not it’s wise policy, they knew this was what they were doing. According to that summary:
Sec. 101 clarifies that when the government only provides a description of the target of surveillance for purposes of obtaining a warrant (whether or not that warrant is for a regular or roving FISA warrant), that description must be sufficient to allow a court to determine that the target is a single individual.
Emphasis mine. When it came time for markup last week, a “manager’s amendment” introduced by Conyers made a number of putatively strictly “technical” changes to the original bill. The odd thing here is that the legislative summary for the manager’s amendment, dated November 3, bends over backward to make it sound as though the plainly intended and understood effect of the original language was an “unintended consqeuence.” In fact, the summary uses the phrase “unintended consequence” or “unintentionally” no less than four times. The new language, according to the summary “clarifies” that the change applies only to roving surveillance.
Well fine, you might say; the tighter “single individual” requirement only should apply to roving surveillance. That’s how it works for criminal wiretaps, where you just need to show the facility is being used (by someone) for criminal activity to get a normal warrant, but you have to give a target’s identity for a roving tap. The problem is, that’s not what the manager’s amendment actually did. What it actually did was carve away huge portions of the prohibition on John Doe warrants. In the amended House bill, there’s a restriction specifically for roving taps that requires a judge to find:
that the target of the application is a foreign power, as defined in paragraphs (1), (2), (3), and (6) of section 101(a), an agent of such a foreign power, or a specific individual
So, bear with me again, because we need definitions. The “as defined in paragraphs…” bit basically restricts “foreign power” to mean actual foreign governments and their appendages, rather than terrorist groups or other foreign political associations. But an “agent of a foreign power,” recall, encompasses any “group, entity, association, corporation, or foreign power.” So you’ve still got a version of the very same loophole you started with: Your roving warrant can identify or describe a single specific individual or an “agent of a foreign power,” understood to include any “group” or “entity,” provided the umbrella “foreign power” is a government rather than some other terrorist or political group.
The thing is, this doesn’t really diminish the problem, which fundamentally has to do with permitting surveillance rove across phones, Internet accounts, and locations, without being sufficiently moored to something more concrete than a collective description. Whether the collective description pertains to agents of al-Qaeda or the People’s Republic of China seems rather secondary. Moreover, it seems to run against the spirit of the interpretation of FISA implicit in the Senate’s report on pre-9/11 FISA implementation failures. Because that report is very clear on one point: That in the case of Zacarias Moussaoui, the FBI made the mistake of thinking a FISA application for an agent of a foreign power had to identify the foreign power as some already-recognized hostile group. The opinion of the Senate was quite clearly that you could have probable cause for identifying your target as an “agent of a foreign power” without being able to say definitively that the “foreign power” in question was al-Qaeda or Islamic Jihad or any other already-known group. That’s as much as you’ve got to show at the relatively high level of probable cause that’s your threshold for the underlying application, anyway. Is probable cause also the standard a judge is supposed to apply when he “finds” that the target group or entity is getting direction from Iran? (Can he apply the Weekly Standard and conclude that al-Qaeda is under the direction of Saddam Hussein’s ghost?) It’s not explicit, but the suggestion in context is that it’s something weaker. And if my parenthetical example above is a little frivolous, the point is serious: It’s entirely plausible that you’d have probable cause grounds for characterizing a cell linked to a terrorist foreign power and much, much more tenuous grounds tying that overarching terror group to a government. Finally, the House bill lets “lone wolf” expire, but if (as seems likely) that gets renewed, you’ve got a freestanding basis for finding a group to be an “agent of a foreign power” if it’s engaged in preparation for international terrorism without a link to any other foreign power, at which point it’s not clear how you’d read the clause.
As usual, I’ve gone on a bit, but the upshot is this: The “good” House bill looks like it closes the “John Doe warrant” loophole, but it doesn’t. At most, it tightens the loop somewhat. Whether that ends up being good enough for government work turns a good deal on how the new language would be interpreted and applied, but I’d feel a good deal better about it if the Committee didn’t seem so determined to obfuscate what it’s doing. They’re clear about wanting to leave some flexibility in the description for non-roving wiretaps, and that’s basically fine. But they’re going out of their way to permit John Doe warrants for non-terrorist groups, in a very precise and clearly intentional manner. I’d rather like to know why. And if the answer is “Well, we’ve got to do it this way, but it’s a secret—we can’t tell you why,” then we may as well just stop playing at democracy and write the statues in cipher too.