A striking feature of proceedings at the Foreign Intelligence Surveillance Court (FISC) is that the executive always wins. Between 1979 and 2012—the first thirty-three years of the FISC’s existence—federal agencies submitted 33,900 ex parte requests to the court.
See Evan Perez, Secret Court’s Oversight Gets Scrutiny, Wall St. J. (June 9, 2013, 7:11 PM ET),
; see also Foreign Intelligence Surveillance Act, Fed’n Am. Scientists, https://www.fas.org/irp/agency/doj/fisa (last visited Feb. 20, 2014) (collecting the FISC’s annual
reports to Congress). The judges denied eleven and granted the rest: a 99.97% rate of approval.
This “win rate,” enviable even to the Harlem Globetrotters, is almost always interpreted as evidence that that the court is failing to do its job. In the media, in legal scholarship, and in Congress, there is a widespread sense that a court in which the executive always wins can be nothing more than a rubber stamp.
See, e.g., Continued Oversight of the Foreign Intelligence Surveillance Act: Hearing Before the S. Comm. on the Judiciary, 113th Cong.
49 (2013) (statement of Laura K. Donohue, Professor, Georgetown University Law Center), available at http://scholarship.law.georgetown.edu/cong/117 (arguing that the “rather remarkable success
rate” raises a “serious question about the extent to which FISC and [the Foreign Intelligence Surveillance Court of Review] perform the function they were
envisioned to serve”); Theodore W. Ruger, Chief Justice Rehnquist’s Appointments to the FISA Court: An Empirical Perspective, 101 Nw. U. L. Rev.
239, 245 (2007) (asserting that the “government success rate [is] unparalleled in any other American court”); Spencer Ackerman, FISA Chief Judge Defends Integrity of Court over Verizon Records Collection, Guardian (June 6, 2013, 6:17 PM EDT),
(“Civil libertarians have long been alarmed that [the court] approves the vast majority of government surveillance requests.”); Carol D. Leonnig et al., Secret-Court Judges Upset at Portrayal of “Collaboration” with Government, Wash. Post. (June 29, 2013),
(“Some critics say the court is a rubber stamp for government investigators because it almost never has turned down a warrant application.”). That perception is now helping fuel legislative reforms. Both the House and Senate are considering bills to reform the FISC. 3 See Ensuring Adversarial Process in the FISA Court Act, H.R. 3159, 113th Cong. (2013); FISA Court Reform Act of 2013, S. 1467, 113th Cong.
(2013). In January 2014, President Obama added wind to their sails by endorsing a proposal to let outside advocates appear before the court. 4 See Remarks on United States Signals Intelligence and Electronic Surveillance Programs, 2014 Daily Comp. Pres. Doc. 30 (Jan. 17, 2014). And, in a nod to the fact that FISC skepticism runs deep, the President expressed openness to working with Congress on a broader set of changes at the court.
This Essay challenges the popular belief that the government’s “win rate” is, by itself, a meaningful source of evidence that the FISC is neglecting its duties. The win rate is misleading for a simple reason: The requests the FISC receives are not a random or representative sample of all cases in which the executive branch believes it would benefit from a warrant. The number and type of government requests are responsive to the level of oversight the court exercises—just as a plaintiff’s decision to litigate is responsive to changes in the law. Because it is costly to make an ex parte application (in time, resources, and reputation) and because the executive has long-running knowledge of how the FISC treats applications, there is little reason to expect agencies to submit losing requests. And while the rarity of ex parte proceedings might make this outcome seem unprecedented or extraordinary, other ex parte proceedings—like those for Title III wiretaps and delayed-notice warrants—display equally lopsided results: the government “wins” almost 100% of the time.
In offering an alternative explanation for the FISC’s high approval rate, this Essay also questions the common defenses employed by the FISC judges themselves, who—clearly embarrassed by the unflattering attention—have defended the court in ways that are awkwardly self-congratulatory or simply wrong. Indeed, there are plenty of reasonable concerns about the FISC. Its opinions are secret; its proceedings are one sided. But the emphasis on the win rate is a distraction from the real debate over the court, and dropping the misleading statistics can help critics prioritize the more genuine concerns.
In Part I, I describe the main features of FISC proceedings and sketch a simple model to explain the why we might expect an ex parte win rate to be so dramatically lopsided. In Part II, I discuss whether this simple theory of ex parte proceedings fits the available data. In Part III, I discuss how these conclusions should affect the current debate over the FISC—a debate that should more carefully consider the general tradeoffs presented by ex parte proceedings, rather than treat those tradeoffs as unique or surprising evils of the FISC.
I. FISC Proceedings and a Theory of Ex Parte Selection
A. How the FISC Works
The FISC was established in 1978 by the Foreign Intelligence Surveillance Act (FISA) to oversee the process by which federal agencies request warrants for intelligence gathering. In its current form, the court consists of eleven federal district judges appointed by the Chief Justice of the United States to serve a maximum of seven years.
50 U.S.C. § 1803(a) (2011). Note that the political makeup of the FISA Court is also a source of some controversy: the current judges are
overwhelmingly ones appointed by Republican Presidents, and this has been the case throughout the court’s history. See, e.g., Charlie Savage, Roberts’s Picks Reshaping Secret Surveillance Court, N.Y. Times (July 25, 2013),
http://www.nytimes.com/2013/07/26/us/politics/robertss-picks-reshaping-secret-surveillance-court.html. This is a reasonable source of concern—but, as discussed below, it’s not something that necessarily has an effect on the win rate: in the Title III and
delayed-notice contexts, the federal judiciary as a whole—which is far more politically diverse—produces similarly lopsided results. Each week, one of the eleven judges is “on duty” in Washington and handles routine surveillance requests. A full-time staff of clerks and attorneys at the court also assists with these requests. The staff attorneys produce a written analysis of each application for the duty judge, who then makes a preliminary decision on the request. A staff attorney then relays this preliminary decision to the government, which has an opportunity to submit additional information and a final application for an official ruling. 6 See Letter from Reggie B. Walton, Presiding Judge, U.S. Foreign Intelligence Surveillance Court, to Senator Charles E. Grassley,
Ranking Member, Senate Comm. on the Judiciary 1-3 (July 29, 2013), available at https://www.fas.org/irp/news/2013/07/fisc-leahy.pdf. My discussion here is drawn largely
from Judge Walton’s letter. As discussed below, this might be an especially important feature of the proceedings that generates the almost-perfect success
FISA and the court’s rules of procedure dictate that only government attorneys can file applications with the court, 7 See FISC R. P. 6. and hearings on these applications “must be ex parte and conducted within the Court’s secure facility.” 8 Id. at R. 17(b) (italics omitted). Once in a blue moon, the court does receive adversarial briefing on a matter—usually in the FOIA or First Amendment contexts—but “FISA does not provide a mechanism for the Court to invite the views of nongovernmental parties.” 9 Letter from Reggie B. Walton to Senator Charles E. Grassley, supra note 6, at 7. Thus, in all of the cases in which the court receives an application from the government, the only party the court hears from is the executive branch.
B. Case Selection in the Ex Parte Context
How do such procedures affect the requests that the executive brings before the court? A central assumption of the law and economics approach to legal process is that legal actions are costly. 10 For a general discussion of the law and economics approach to litigation, see Steven Shavell, Foundations of Economic Analysis of Law 389-90 (2004). Because potential litigants are assumed to form rational estimates of the expected payoff before deciding to bring a legal action, parties will not bring cases they know they will lose—which is especially true when legal rules are clear and predictable. And because decisions to litigate are responsive to the probability of success—not just the underlying facts of the dispute—there is widespread agreement that the actions selected for litigation will “constitute neither a random nor a representative sample of the set of all disputes.” 11 George L. Priest & Benjamin Klein, The Selection of Disputes for Litigation, 13 J. Legal Stud. 1, 4 (1984).
These simple assumptions are not particularly controversial: they boil down to the rather commonsense notion that parties tend to do things when they predict that the expected benefits exceed the expected costs. Instead, the real debate is (as one scholar has put it) “about what kind of selection is taking place and why.”
Peter Siegelman & Joel Waldfogel, Toward a Taxonomy of Disputes: New Evidence Through the Prism of the Priest/Klein Model, 28 J. Legal
Stud. 101, 103 (1999).
And how does case selection work in an ex parte context? There are obvious differences between ex parte procedures and adversarial litigation. In the simplest models of adversarial litigation, both sides have incentives to avoid litigation when the future decision is certain: why waste resources litigating when settlement is available? But in ex parte contexts that presumption is turned on its head: the alternative to legal action is not a settlement that generates a surplus—it is nothing. Parties can achieve their goals only by bringing requests. But if those requests are costly, and if the requesting party knows the probable outcome of each request, we should only expect applications with a high chance of approval.
Are these assumptions—costly requests and clear rules—realistic? The assumption that the government is motivated to reduce costs shouldn’t be controversial. Costs can be broadly construed: it is time consuming and embarrassing for lawyers to make losing arguments—especially when they interact with the same set of judges and lawyers time and time again—and it is difficult for bureaucracies to justify throwing (too much) money in the fireplace.
While there is not an abundance of public information about the heft of a standard FISC request, at least some of the matters before the court have
resulted in substantial briefing. See Letter from Reggie B. Walton to Senator Charles E. Grassley, supra note 6, at 4 (noting
that larger cases result in a longer timeline and additional briefing). The Department of Justice itself has said that the process “requires the intelligence community to divert scarce intelligence experts to the time-consuming process of compiling court submissions.” 14 Key Myths About FISA Amendments in the Protect America Act, U.S. Department Just., http://www.justice.gov/archive/ll/paa-dispelling-myths.html (last visited Feb.
20, 2014). Any of these rough characterizations would be broadly consistent with how agencies are traditionally modeled in other contexts, 15 See, e.g., William A. Niskanen, Jr., Bureaucracy and Representative Government (1971) (developing a model in which agencies seek to maximize
their budgets); see also Kenneth A. Shepsle & Mark S. Bonchek, Analyzing Politics: Rationality, Behavior, and Institutions 346 (1997)
(describing Niskanen’s model as a “now classic treatment” of how bureaucracies function). For a less popular description of the agency maximand, see
Richard A. Posner, The Behavior of Administrative Agencies, 1 J. Legal Stud. 305, 305 (1972) (“The agency’s goal is assumed to be to maximize the
utility of its law-enforcement activity.”). and would support the basic idea that the executive selects cases based on whether it expects to win.
The assumption of clear rules is more challenging: laws are not self-applying, and there will always be close, uncertain cases. Still, two features of the FISC process suggest that the executive branch lawyers who make the requests are unusually well informed. First, the executive is a hugely repeat player at the FISC: you don’t file 33,900 requests without developing some sense of how the court will respond to the 33,901st. The second feature is that FISC judges commonly indicate, before a government request is finalized, how they are planning to rule. Over a three-month period in 2013, for instance, Reggie Walton, Presiding Judge of the FISC, reported that “24.4% of matters submitted ultimately involved substantive changes to the information provided by the government or to the authorities granted as a result of Court inquiry or action.”
Letter from Reggie B. Walton, Presiding Judge, U.S. Foreign Intelligence Surveillance Court, to Senator Charles E. Grassley, Ranking
Member, Senate Comm. on the Judiciary 1 (Oct. 11, 2013), available at
http://www.uscourts.gov/uscourts/courts/fisc/ranking-member-grassley-letter-131011.pdf. Of course, questions remain about how this alteration process works in practice; the only available information is Judge Walton’s statement that “[t]his
assessment does not include minor technical or typographical changes, which occur more frequently.” Letter from Reggie B. Walton to Senator
Charles E. Grassley, supra note 6, at 7 n.9. But note, even if one assumes that all the altered requests would have been rejected (but for
their alterations), the FISC results would still be dramatically lopsided, since more than three-quarters of the government’s requests are approved without
alteration. In the uncertain cases, the FISC tips its hand.
This last feature is especially important for generating the close-to-100% result: it might explain why executive lawyers do not bring cases that have a high, but nonetheless uncertain, probability of success.
Consider the following intuitions: Without the preliminary feedback, we might expect the relationship between the quality of the request and the
probability of success to be continuous and monotonic. Say, for example, that a good request had an 80% chance of success, a very good request had a 90%
chance, a stellar request had a 99% chance, and so forth. If that were the case, we might expect the executive to submit more requests that were merely
“good”—but still have a high probability of success—and thus drag down the overall win rate. The supposition here is that the preliminary feedback actually
produces a discrete drop rather than a continuous distribution: either the executive learns how to make its good requests stellar, or it knows which of its
good requests will not make the cut—and thus decides not to bring them. Cf. Jon D. Michaels, The (Willingly) Fettered Executive: Presidential Spinoffs in National Security Domains and Beyond, 97 Va. L. Rev. 801, 817-28 (2011) (making a
similar point with respect to the process by which the Committee on Foreign Investment in the United States considers foreign attempts to acquire
strategically important American assets). Of course, this sketch of the FISC’s process is nonetheless a simplification—but hopefully one that captures the essential elements of the ex parte process, and provides strong reasons why government attorneys do not have incentives to file losing requests.
II. Do Other Ex Parte Proceedings Work the Same Way?
The previous Part makes a prediction: if requests are costly and rules are clear, we should expect the success of ex parte requests to be lopsided. Does that prediction hold up in practice outside of the FISC? This Part discusses three kinds of evidence that suggest the answer is affirmative. Subparts A and B discuss wiretapping requests made under Title III and so-called “delayed-notice” search warrants—both of which are strong analogs to the FISC process. Subpart C discusses evidence from the limited ex parte proceedings available outside the law enforcement and national security contexts: temporary restraining orders and ex parte patent hearings.
A. Other State and Federal Wiretapping Requests
The most obvious analog to FISC review is the ex parte process by which state and federal law enforcement agencies request warrants under Title III of the Omnibus Crime Control and Safe Streets Act of 1968. 18 Pub. L. No. 90-351, tit. III, 82 Stat. 197, 211-25 (codified as amended at 18 U.S.C. §§ 2510-2520 (2012)). To receive a Title III warrant, the agency must show a federal district judge that there is probable cause to believe that a crime was or will be committed. As with the FISC, the Title III process allows the judge to seek “additional testimony or documentary evidence in support of the application” as she deems necessary. 19 18 U.S.C. § 2518. And, as with the FISC, statistics on Title III wiretaps are reported annually to Congress.
Title III statistics strongly support the basic prediction that a lopsided win rate is a general feature of ex parte proceedings, not a unique characteristic of the FISC. Between 1968 and 2012, state and federal agencies made 50,419 requests for Title III wiretaps, and federal courts approved 99.93% of these requests.
See Title III Wiretap Orders—Stats, Electronic Privacy Info. Center, http://epic.org/privacy/wiretap/stats/wiretap_stats.html
(last visited Feb. 20, 2014); c.f., e.g., Kim Smith, Rejection of Search-Warrant Pleas Rare, Ariz. Daily Star (Dec. 29, 2011, 12:00 AM),
(discussing how local judges act similarly).
Of course, it might be objected that Title III hearings (and perhaps other ex parte hearings) are also rubber-stamp processes.
Though beyond the scope of this Essay, one potential difference between FISC proceedings and other wiretapping proceedings is that the latter might
have more back-end checks: for example, the evidence obtained through the wiretap will usually be introduced at trial, where the defendant can move to
suppress. This does not happen with FISA evidence—although, in theory, other back-end checks could be deployed. See Letter from Reggie
B. Walton to Senator Charles E. Grassley, supra note 6, at 8-9 (observing that “a nongovernmental party served with an order may invite an
opportunity to be heard by the Court through refusal to comply with an order,” but noting further that no parties have invoked this section of the FISA
statute). As discussed more fully below, that objection is credible. But, if made, this objection suggests that the real debate is about the general tradeoffs of an ex parte proceeding—that is, the benefits of an adversarial process versus the need for a secret one—and not the particular failings of one court or judge.
B. Delayed-Notice Search Warrant Requests
Another close cousin of FISC review is the process by which law enforcement agencies request so-called “delayed-notice search warrants”—commonly referred to as “sneak and peek” warrants. As the former name suggests, these allow law enforcement “to temporarily delay giving notice that the search has been conducted.” 22 Letter from William E. Moschella, Assistant Att’y Gen., to J. Dennis Hastert, Speaker, U.S. House of Representatives 1 (Sept. 22, 2004), available at http://www.justice.gov/dag/patriotact213report.pdf. An agency requesting a search or surveillance warrant may also request to delay notice, which the court will permit if it is found “necessary”—that is, if the court has “reason to believe that notification of the existence of the subpoena may have an adverse result” on public safety or a similar goal. 23 18 U.S.C. § 2705; id. § 3103a. Agencies can also request an extension of the delayed-notification period.
As with Title III requests and FISC applications, federal courts are required to submit annual delayed-notice statistics to Congress. Delayed-notice requests are overwhelmingly approved. In the year ending September 30, 2008, for instance, agencies made 763 applications for delayed-notice warrants, of which 760 were approved (99.6%). There were also 528 requests for extension, of which 527 were approved (99.8%). These statistics are also broken down by court and type of crime, including drug, weapons, tax, and fraud cases—all of which experience the same lopsided approval rate.
Report of the Director of the Administrative Office of the United States Courts on Applications for Delayed-Notice Search Warrants and Extensions
(2009), available at http://big.assets.huffingtonpost.com/SneakAndPeakReport.pdf.
C. Suggestive Evidence from Other Ex Parte Proceedings?
Ex parte statistics also exist outside the national security and law enforcement contexts (though with few systematic reporting requirements). Connecticut, for example, publishes statistics on the number of temporary restraining orders requested and granted, which are also lopsided: between July 2009 and July 2013, 90% of the ex parte requests for domestic abuse restraining orders were ultimately approved.
Protective Order / Restraining Orders, Judicial Branch Statistics, Conn. Jud. Branch,
http://www.jud.ct.gov/statistics/prot_restrain/Prot_Restrain_Order.pdf (last visited Feb. 20, 2014). And the U.S. Patent and Trademark Office keeps records on ex parte requests for patent reexaminations, which require showing that a “substantial new question of patentability” exists. 26 See 35 U.S.C. § 257 (2011); Dennis Crouch, Ex Parte Reexamination Statistics I, Patently-O (June 5, 2008), http://www.patentlyo.com/patent/2008/06/inter-partes-re.html. Since 1981, patent examiners and directors have approved 92% of the 12,874 requests for reexamination, a percentage that seems to be growing over time. 27 See Reexamination Information, U.S. Pat. & Trademark Office 1,
(last updated Sept. 30, 2013).
These ex parte proceedings differ from federal warrant requests in important ways. Perhaps most crucially, these proceedings are not permanently ex parte: the judge can hold an adversarial hearing to decide the same questions. While this is good reason to be skeptical of an easy analogy to the FISC, these data offer some support for the basic idea that ex parte proceedings produce a lopsided result even when the executive isn’t the sole party making requests.
III. Reorienting the FISC Debate
The fact that ex parte proceedings are lopsided in a broad range of contexts—and that there are good reasons to expect them to be lopsided—has several implications for the current debate over the FISC. First, it gives us reason to be skeptical of other explanations for why the FISC approves so many requests. For example, former FISC Judge James Carr has written that the “low FISA standard of probable cause—not spinelessness or excessive deference to the government—explains why the court has so often granted the Justice Department’s requests.” 28 James G. Carr, Op-Ed., A Better Secret Court, N.Y. Times (July 22, 2013), http://www.nytimes.com/2013/07/23/opinion/a-better-secret-court.html. But if this were correct, we should expect different approval rates when the requesting party needs to clear a higher bar—say, by justifying the warrant (probable cause) and delayed notice (necessity) in Title III cases. We don’t see this. It also casts doubt on the relationship between the lopsided win rate and the political affiliation of FISC judges—a common media suggestion. 29 See, e.g., Savage, supra note 5. If Republican deference to the executive produced the FISC win rate, we should also expect to see variation in win rates based on political affiliation in the Title III and delayed-notice contexts (where data on individual judges are available). Again, we don’t.
A second set of conclusions involves consistency. There are good reasons to be skeptical of ex parte proceedings: they are in tension with basic norms of due process—the right to notice, to a hearing, to confront adversaries, and so forth. But if the FISC’s ex parte proceedings are troubling to some critics, those critics should explain whether and why that skepticism extends to the Title III and delayed-notice contexts. The government’s win rate can’t help distinguish among these cases.
Third, thinking about the lopsided win rate as a general consequence of ex parte proceedings, rather than as a specific problem of the FISC, offers perspective on the current FISC reform proposals. In January 2014, President Obama suggested creating an adversarial process for “significant cases” heard by the FISC
See Remarks on United States Signals Intelligence and Electronic Surveillance Programs, supra note 4, at 5.
—a proposal that mirrors roughly similar proposals under consideration in Congress.
See FISA Court Reform Act of 2013, S. 1467, 113th Cong. (2013).
These proposals highlight an important difference between the FISC’s ex parte work and the ex parte work of federal courts responding to delayed-notice and Title III requests: FISC cases sometimes generate precedent on novel questions of law. But these cases are also rare—it has been said that novel legal questions come before the FISC only “[o]nce in a very great while.”
See Strengthening Privacy Rights and National Security: Oversight of FISA Surveillance Programs: Hearing Before the S. Comm. on the Judiciary, 113th
Cong. 1 (2013) (statement of Judge James G. Carr), available at http://www.judiciary.senate.gov/pdf/7-31-13CarrTestimony.pdf. If this proposal is adopted, critics of the FISC should be at peace with the fact that the judges will probably continue to approve the vast majority of the executive’s routine requests; in other words, adding an occasional adversary for the most significant cases is unlike to affect the win rate.
But the simplest lesson of this analysis is that public discourse should stop focusing on the rate with which the FISC approves applications. Ultimately, we should care about the substance of what the court approves, not the frequency with which it does so. And if the substance of what the court does affects how the government selects applications, there’s no reason to think FISC reforms will change the win rate. Even an ex parte process that has satisfied the FISC’s most fervent critics might continue to produce an outcome in which the government always wins.