Assessing Surveillance Laws in An Era of Sunset Provisions:
In my blog post last week on the new FISA Amendments, and a follow-up on Friday, some commenters expressed strong disagreement — and in some cases, downright contempt — at my view that the most natural baseline for assessing the latest FISA Amendments was last year's FISA law, the Protect America Act. Our disagreement raises a conceptually interesting question: How should we characterize the direction of new surveillance laws in an era when so many surveillance laws are being subject to sunset provisions? And applying that to the specific case here, is the Protect America Act the right baseline for the new FISA Amendments?
I think the question is tremendously important. Statutory laws require a feedback loop: The public needs to know if their policy preferences are being enacted into law. But the details of surveillance law are a mystery to 99.99% of the population. The laws are arcane and technical, which means that the only way most people will get a sense of the direction of the law is second-hand. In my view, this puts intense pressure on those of us who want to write on these issues in a public forum to get it right. The democratic process only works correctly if we describe the law accurately. If our writings don't accurately describe what is happening, over time the law will no longer match public preferences in the way that they should in a democratic system.
Reporting on the direction of the law is particularly tricky when Congress uses sunset provisions. A sunset provision makes the new law temporary: After a period of time, the new provision elapses and we revert to the old law (pending the likely passage of some new legislation). Sunsets let Congress experiment with different rules and see how they work out; near the end of the sunset period, Congress can have hearings and then decide whether to stick with the old law or try something new.
Congress's heavy reliance on sunset provisions in surveillance law is one of the most interesting developments in the surveillance law of the post 9/11 era. As far as I know, the Patriot Act's sunset provisions were the first to use this: About half of the surveillance law changes in the original Patriot Act of 2001 sunsetted at the end of 2005. The Protect America Act of 2007 picked up the idea, sunsetting after 6 months but allowing orders granted under it to be valid for a year after they were issued. The new FISA Amendments also use sunsets: the new law expires at the end of 2012.
So what baseline should we use when reporting on these issues? Here's my best sense: The most accurate and principled way to report on changes to surveillance laws in an era of sunset provisions will usually be to compare each new provision to the prior provision Congress negotiated. With an important exception, we should ordinarily compare new rules to sunsetting or recently sunset old rules.
I think this is the case for three reasons. First, my sense is that this view matches the understanding of the institutional players in Congress and the Executive branch. Congress has pushed the use of sunsets to ensure that the baseline negotiated today can be renegotiated tomorrow; as the old law sunsets, the new law becomes the baseline for where the new law will go. The understanding of the parties is that the sunsetting law will be the future baseline.
For example, the sunsetting Patriot Act provisions became the standard for how the law would change when Congress negotiated the next step at the end of 2005. Congress had hearings on whether or how to renew or amend pretty much every single sunsetting provision, no matter how minor. Similarly, my understanding is that the Protect America Act became the baseline for the negotiation of the new FISA Amendments. Given that this is the expectation of the institutional players when they insist on (or fight against) sunset provisions, I think it makes sense to track that understanding.
Second, sunset provisions have become sufficiently common in surveillance law that any other baseline becomes rather arbitrary. As I see it, sunset provisions are here to stay: They seem to have become the new norm in surveillance law. If I'm right about that, my concern is that looking back to the pre-sunset law as the standard mostly just looks back to the era before sunsets were common rather than to any objectively "neutral" standard. In the case of surveillance of individuals overseas, for example, Congress's realization of the new technology allowing surveillance from inside the U.S. of individuals abroad led first to the Protect America Act for between 6 and 18 months (depending on how you look at it), then the new law for the next 52 months or so, and then we'll have something else new when the 2008 law expires in 2012. We may keep going on sunset provisions after that, too: we'll keep renegotiating every few years off into the future.
I think the question is tremendously important. Statutory laws require a feedback loop: The public needs to know if their policy preferences are being enacted into law. But the details of surveillance law are a mystery to 99.99% of the population. The laws are arcane and technical, which means that the only way most people will get a sense of the direction of the law is second-hand. In my view, this puts intense pressure on those of us who want to write on these issues in a public forum to get it right. The democratic process only works correctly if we describe the law accurately. If our writings don't accurately describe what is happening, over time the law will no longer match public preferences in the way that they should in a democratic system.
Reporting on the direction of the law is particularly tricky when Congress uses sunset provisions. A sunset provision makes the new law temporary: After a period of time, the new provision elapses and we revert to the old law (pending the likely passage of some new legislation). Sunsets let Congress experiment with different rules and see how they work out; near the end of the sunset period, Congress can have hearings and then decide whether to stick with the old law or try something new.
Congress's heavy reliance on sunset provisions in surveillance law is one of the most interesting developments in the surveillance law of the post 9/11 era. As far as I know, the Patriot Act's sunset provisions were the first to use this: About half of the surveillance law changes in the original Patriot Act of 2001 sunsetted at the end of 2005. The Protect America Act of 2007 picked up the idea, sunsetting after 6 months but allowing orders granted under it to be valid for a year after they were issued. The new FISA Amendments also use sunsets: the new law expires at the end of 2012.
So what baseline should we use when reporting on these issues? Here's my best sense: The most accurate and principled way to report on changes to surveillance laws in an era of sunset provisions will usually be to compare each new provision to the prior provision Congress negotiated. With an important exception, we should ordinarily compare new rules to sunsetting or recently sunset old rules.
I think this is the case for three reasons. First, my sense is that this view matches the understanding of the institutional players in Congress and the Executive branch. Congress has pushed the use of sunsets to ensure that the baseline negotiated today can be renegotiated tomorrow; as the old law sunsets, the new law becomes the baseline for where the new law will go. The understanding of the parties is that the sunsetting law will be the future baseline.
For example, the sunsetting Patriot Act provisions became the standard for how the law would change when Congress negotiated the next step at the end of 2005. Congress had hearings on whether or how to renew or amend pretty much every single sunsetting provision, no matter how minor. Similarly, my understanding is that the Protect America Act became the baseline for the negotiation of the new FISA Amendments. Given that this is the expectation of the institutional players when they insist on (or fight against) sunset provisions, I think it makes sense to track that understanding.
Second, sunset provisions have become sufficiently common in surveillance law that any other baseline becomes rather arbitrary. As I see it, sunset provisions are here to stay: They seem to have become the new norm in surveillance law. If I'm right about that, my concern is that looking back to the pre-sunset law as the standard mostly just looks back to the era before sunsets were common rather than to any objectively "neutral" standard. In the case of surveillance of individuals overseas, for example, Congress's realization of the new technology allowing surveillance from inside the U.S. of individuals abroad led first to the Protect America Act for between 6 and 18 months (depending on how you look at it), then the new law for the next 52 months or so, and then we'll have something else new when the 2008 law expires in 2012. We may keep going on sunset provisions after that, too: we'll keep renegotiating every few years off into the future.
Related Posts (on one page):
- Assessing Surveillance Laws in An Era of Sunset Provisions:
- Strange Op-Ed By Chris Hedges:
- The New FISA Law -- and the Misleading Media Coverage Of It:
.
Accurately summarize the competing and/or evolving laws so the readers can conduct a policy debate instead of a framing debate.
.
The below comments describe the warrantless acquisition of international calls to which a person in the US is a party. All errors are mine.
.
FISA-1978 required a warrant for snooping international communications of US persons, neither FISA-2007/PAA nor FISA-2008 requires a warrant for this.
.
PAA and FISA-2008 are equivalent in their requirements for warrantless acquisitions: "the targeting of persons reasonably believed to be located outside the United States to acquire foreign intelligence information." FISA-2008 recites or clarifies what the government said was implied by PAA - no reverse targeting, no intentional acquisition of domestic-domestic communications.
.
PAA had a court apply "clearly erroneous" to a narrow inquiry, that the procedures to establish surveillance would result in acquisition of communications of a person reasonably believed to be located outside of the United States. I think this is an easy element to show to a much higher standard. Claiming a "tightening of independent review" by moving from "clearly erroneous" to "clear and convincing" is misleading. The change in standard of review wouldn't make any substantive difference on the question of "do these communications have a foreign component." Using a track metaphor, shifting the 100 yard dash qualifying point from 2 minutes to 30 seconds doesn't change the fact that all the contestants will qualify for the big race.
.
All three versions of the law, FISA-1978, PAA, and FISA-2008, refer to and require the DOJ/DNI to apply exactly the same minimization - 1801(h). [minimization has always applied after the acquisition of the communications]
.
Under the PAA, the "minimization" protocol was opaque to FISC unless a communications provider challenged a snoop directive. FISA-2008 inserts court review whether or not the cooperating telecom protests the snoop order.
.
Under all versions of the law, court review of minimization is "meet the definition of minimization procedures [per 1801(h)]."
.
Minimization, as defined, doesn't prevent snooping. It prevents disclosure or dissemination. It also provides some sort of barrier to retention, but under the retention barrier, the government is permitted to retain "nonpublicly available information concerning unconsenting United States persons consistent with the need of the United States to obtain, produce, and disseminate foreign intelligence information." [One year for plain text, "as long as it takes" for encoded communications]
.
The specificity and particularity of DOJ snoop descriptions / order requests (not that an order is required, but the snooper is required to describe to the Court how the surveillance directive to the telecom won't intentionally acquire domestic-domestic traffic, even if an order is not sought) changes radically going from FISA-1978 to FISA-2007, and does not change substantively when going from FISA-2007 to FISA-2008. The 2007 and 2008 versions boil down to "make sure you only obtain international calls, and apply 1801(h) minimization."
.
The "specific" [702(g)] certification made by the DOJ/DNI under FISA-2008 is reviewed by FISC for "whether the certification contains all the required elements." An inspection for presence of elements does not challenge the contents or substance of those elements.
.
PAA and FISA-2008 broadens the number of persons authorized to instigate warrantless surveillance. Under FISA-1978 that population was limited to be President and AG. Under PAA and FISA-2008 that population is all heads of an element of the intelligence community [See 50 U.S.C. 401a(4)]
.
The secret accountability reporting to FISC and Congress (I presume to a gang-of-8-like audience) is triggered by "dissemination of intelligence reports containing a reference to a United States-person identity," and "United States-person identities subsequently disseminated by the element concerned in response to requests for identities that were not referred to by name or title in the original reporting." I have no idea what those triggers constitute (meaning I can't rephrase or paraphrase the meaning), or whether or not the triggers are easily avoided in practice. In any event, the report is numerical, like the current reports of number of warrant applications, refusals, and modifications.
I think the difficulty is that the Bush Administration didn't particularly oppose adding civil liberties protections into the new law: I suspect the reason is that fears for how the PAA was being implemented were unfounded, so the Administration could add privacy protections without really touching on anything important to the Administration. I don't know if that makes the new law a "compromise bill" -- my point is only that it adds a lot of privacy protections beyond the earlier law.
.
"Above, I said I don't know what "disseminate" means. "Disseminate" is passing of information from one US government agency (e.g., NSA, but could be any agency) to another. Regulations that describe dissemination policies do not describe acquisition policies.
.
Cryptome hosts what is an embodiment of "minimization" at http://cryptome.org/nsa-fisa.htm. I think the policies stated in some of those documents "meets the definition of minimization," and would satisfy the statutory court review of "to assess whether such procedures [the minimization procedures adopted] meet the definition of minimization procedures."
What do you think Section 704 does? That's the one I was referring to.
.
e.g. See assault weapons ban.
.
In another subject at VC, the second amendment, we see a robust debate on the condition of the law in various times past - not technically changes due to "sunset," but changes as courts and legislatures regulated and approved. The public seems fairly well able to grasp the discussion without limiting itself to the most recent iteration of regulation.
.
Section 704 is similar to Section 703, but without the "and such acquisition is conducted within the United States" limitation. Section 704 covers the obtaining of a court order for targeting a United States person other than when acquisition is conducted within the United States.
.
Section 705 makes sense of the "separation" of these various non-overlapping court orders (703 &704 orders "shall cease unless the targeted United States person is again reasonably believed to be located outside ..."). IT says sections 703 and 704 are unnecessary.
.
.
On second thought, 703/704 are not unnecessary or surplussage in every case. If the government targets a US person who never sets foot in the US, orders under 703 and 704 would pertain.
.
But if a US person is targeted and a warrant is obtained to snoop on them while they are in the US, that warrant (of course) covers the snooping activity in the US, and no 703 or 704 warrant is required for when the targeted US person is out of the US.
As one of your critics on this issue, though not a contemptuous one, I suggest a compromise: When, as here, a newsworthy disclosure involving national security, civil liberties, and secretive action by the government is made (the Times story), followed by legislative reaction perceived (perhaps wrongly) by the public as temporary (the PAA), and then followed by legislative change viewed by the public as permanent (the FISA 2008 amendments), all within about 2 1/2 years, why not treat the series of events as one developing story rather than looking for a "natural" baseline? Prior to 9/11 and the events thereafter, I think many Americans, even civil liberties activists, were unaware that there even was a FISA or FISA court. The whole system is new to them, even though of course FISA was enacted in the wake of Watergate 30 years ago. Why is the system new to them? It's complicated and it supposedly never involved Americans until the Times revealed that it did. A thorough, helpful accounting of how the FISA system affects the lay public is incomplete without an assessment of the balance sheet of what rights we have now v. what rights we had then (then being the point when it all started to change). There's a reason many intelligent people are upset about FISA 2008. It's not that they are (necessarily) raving leftists. It's that the "compromise" between the Democratic Congress and the Bush administration over the course of time since the Times' revelation appears to be the Democrats giving away the store to Bush and the telecoms, without giving civil liberties due consideration. And in a complete accounting, that's essentially what it is, however it looks on a micro level to an expert legal technician such as yourself--isn't it? If not, tell us why not, since you know more about the issue than most people, but please don't scold the readership or the press for being naive enough to think that the Democrats should have put up more of a fight when it came to replacing an expiring law with something permanent.
I think this is the source of the disagreement. I see it exactly the opposite way -- the point of the sunset clause is to function like a tolling agreement. It enables the parties to reconsider the issue against the original baseline after obtaining more information. That's what Congress is supposed to do.
In a larger sense, I think this gets us into some of the same considerations that we face in trying to describe "settled" law or "stare decisis". The concept we're trying to express is that the public has considered an issue over time, heard the arguments, and reached a conclusion. Here are some factors I'd consider in deciding whether we've reached that point:
1. The issue was of notable importance and widely debated nationwide. George Washington once said that “It is on great occasions only and after time has been given for cool and deliberate reflection that the real voice of the people can be known.”
2. Congressional votes on the issue were considered, deliberate, and reasoned.
3. The law was reaffirmed over time by different majorities.
4. Both the President and the Supreme Court agreed with the law (or at least the Court doesn’t disagree).
5. The law was consistent with the text of the Constitution, the fundamental principles of republican government, and settled practices in related areas.
I'd note, with some irony, that my approach is more conservative than yours, at least in the Burkean sense.
The more cynical suspect that the reason is that the Administration has no intention of being bound by the restrictions, so doesn't care what they are. That is, after all, exactly what the Administration did with the restrictions in FISA.
If a surveillance law with a sunset provision expires and Congress tries to reenact it, they should be able to justify the new law. If the old law was really flawed in light of new circumstances, it should not be difficult to make the argument for an extension of the recently expired law or something similar.
Over time, if similar provisions are reenacted, I can see the presumption changing. After a reenactment or two it makes sense to view a sunset clause as a way to help ensure liberty, without it being a statement that the law is temporary. This deals with the issue that sunset provisions are becoming the norm in surveillance law. It does not freeze the baseline back in the 90s, it merely requires a certain degree of time and deliberation to pass before the establishment of a new baseline.
But in the present context, the old law is the better baseline. This is especially true for the public debate on the issue. Whatever sunset provisions might mean to Congress and the President, to the average American they probably mean something like "temporary emergency measure."
To the contrary, the sunset provision was clearly a temporizing measure to delay debate for 6 months.
As such, the proper baseline is pre PAA.
Say B is the law as it was before the first sunsetting amendment, S is the law plus the current sunsetting amendment, and F is the law as it will be after the sunsetting amendment is either renewed, modified or jettisoned. Orin, as I understand it, wants commentators to state something like "F confers more (fewer) civil liberties (or whatever) than does S." But this formulation seems preferable (not all that more complicated and gives fuller information): "B conferred more civil liberties than does S and fewer than will F."
I suppose the problem is that you might have a whole series of laws that fall in between B and S. Call Q the set of all {the law plus sunsetting amendments that have already sunsetted}. In this instance it is still probably best not to revert to Orin's formula. Instead, exclude from the universe of discourse any of B through "the member of Q that sunsetted most recently" that are overwhelmingly unworkable, and then flesh this out appropriately: "F falls in between x and y in terms of the civil protections it affords, and is thusly related to S" (or "F affords the fewest civil protections of all members of the (appropriately modified) set {B, the member of Q that sunsetted most recently} and S," etc.).
The PAA came along, and it did some things right, but it appeared to leave some holes open, like the infamous possibility for reverse targeting. So, allowing the PAA as the baseline allowed Congress to tweak the law a bit more, fixing problems that the PAA had either opened up or not solved.
Your argument works for the non-controversial portions of the PAA. The controversial aspects were put off for 6 months. No new baseline was estbalished in these areas.
Yes, if only.