Thoughts on the Horowitz Report, Part III: The FISA Findings

Editor’s Note: This post is part of a Lawfare series covering Inspector General Michael Horowitz’s report on the FBI’s “Crossfire Hurricane” investigation, which explored possible connections between the Trump campaign and Russian efforts to interfere in the 2016 presidential election. Part I is available here and Part II is available here.

The great irony of the public reaction to Justice Department Inspector General Michael Horowitz’s report on the FBI’s “Crossfire Hurricane” investigation is that the dire problems the president and his supporters claim Horowitz found—a supposed coup attempt against President Trump—would actually be easier to address than the problems the inspector general found in real life.

If, in fact, Horowitz had reported that there was a group of deep state conspirators at the FBI who had plotted a coup and created the Russia investigation as an “insurance policy” against Trump’s election, the remedy would be simple—and would already have been accomplished. The officials at the top level of the FBI at the time, after all, have been purged. The people the president has accused of “treason” no longer work in government. Perhaps a few more mischief-makers would need to go, but lustration is a comparatively easy project. As Julian Sanchez wrote last month in a perceptive op-ed,

If the FBI botched its applications for Foreign Intelligence Surveillance Act warrants against Mr. Page because of political bias, after all, problems of the sort Mr. Horowitz identified are most likely unique to this case. The bureau obtains about 1,500 FISA warrants each year, and an overwhelming majority have no connection to domestic politics. The solution is also similarly simple: Toss out the bad apples who acted on political motives and add a few layers of safeguards for the tiny fraction of cases that are designated “sensitive investigative matters” because they do intersect with politics.

The problem is that Horowitz, in fact, found a wholly different set of problems at the FBI—problems that are more difficult to fix. They are harder because they don’t stem from a single name or evil cabal to which blame can be affixed. They are harder because it is not obvious what caused them, or even whether they flow from a single cause. And they are harder because they may, or may not, indicate a culture of sloppiness in a process that simply can’t tolerate sloppiness. If the FISA process had been corrupted as a result of a singular act of politicization, that would have been very bad indeed. The prospect that the Horowitz report raises is, however, arguably worse: that there is nothing singular about the problems that took place in the Carter Page FISA applications and, therefore, that the FISA process on average performs significantly less rigorously than those of us who have defended it have believed.

Without summarizing the inspector general’s findings too deeply, let me say that the findings with respect to the conduct of the Carter Page FISA applications is a genuinely disturbing read. Horowitz found that agents failed to include in their initial surveillance application information about the target that would have undermined their showing of probable cause—including that Page had previously been an operational contact of the CIA. They overstated other things that tended to support the application, including the importance and reliability of Christopher Steele’s prior work with the FBI—which had the effect of presenting the material on Page in Steele’s reporting as more reliable than it actually was. As new information emerged that further cast doubt on Steele’s reporting, they failed to include this information in the renewal applications, leading to an ever-increasing list of errors and omissions in the Justice Department’s communications with the Foreign Intelligence Surveillance Court (FISC). In the most egregious incident, an FBI attorney is alleged to have altered an email before passing it on to the Justice Department with a change that negated Page’s prior source relationship with American intelligence.

As David Kris recently put it,

The errors in the FISA applications on Carter Page were significant and serious. They were not, in my experience, the kind of errors you would expect to find in every case. ... It’s not acceptable to rely on a Confidential Human Source and then not check with his FBI handler in describing his bona fides to the FISA Court. It’s not acceptable to omit some potentially exculpatory recorded statements made by the FISA target to a source. It’s not acceptable to leave unresolved credibility and perhaps factual disputes between a key source and his primary subsource. It’s not acceptable, after closing the key source, to continue to get information from him through an Office of the Deputy Attorney General (ODAG) staffer, thereby effectively treating him as a subsource of the ODAG staffer. And it’s certainly not acceptable for an FBI attorney to alter an email from another intelligence community agency as to whether the other agency had contact with the FISA target or treated him as a source. (Internal citations omitted.)

The inspector general refrains from addressing whether the Page FISA applications were in some ultimate sense legally defective. I actually suspect that at least the initial application would have likely been advanced and approved had all the appropriate information been available to Justice Department attorneys and the errors corrected—though I am less certain about the renewals.

In a deep sense, however, it doesn’t matter whether the errors rendered the application ultimately defective—except, of course, to Carter Page. The volume of errors is not tolerable even if the failures of candor with the court did not render the specific surveillance illegal. FISA requires a degree of rigor in agency communications with the court that this process fell far beneath. Page was entitled to have proposed surveillance of him assessed on an accurate presentation of the record. And the FISC was entitled to greater candor from the government.

All of which raises the question: How often does FBI work product in FISA cases end up so deficient?

This is not the issue the president’s defenders have been harping on. It is one of the issues traditional civil libertarians have been concerned about since the dawn of FISA. And it’s safe to say that if this is anything like what a normal FISA application process looks like, then the FISA process is a mess and the civil liberties anxieties about FISA deserve a serious new look.

But is the Page FISA application typical? Perhaps a better way to ask the question is how typical a dumpster fire like this one is. After all, in any complex human system, total failure is going to happen a certain percentage of the time—and partial failures are going to happen more often. How often does the FISA process fail to the point that the court is given an application with multiple errors and significant omissions? How often does it fail less spectacularly—like, say, an application with three or four errors, rather than the 17 Horowitz identifies in the Page FISA application? The investigation Horowitz performed does not offer much insight into this question.

There are some reasons to imagine from the information Horowitz reports that the Page case may have been genuinely unusual. For example, Horowitz notes the staffing difficulties from the fact of the investigation’s having been run out of FBI headquarters, not one of the field offices—and the consequent rotation of case agents onto and off of the case. In a typical case, individual agents would likely have a sense of whether possibly relevant information wasn’t being included. But with agents swapping in and out on a 90-day cycle, case agents might not have had sufficient knowledge of the material to know if something had been left out. As Kris summarizes,

Running a complex investigation directly from headquarters is not unprecedented, but it is not common—and for good reason. Changing agents every 90 days is obviously challenging, especially in an evolving and complex case like Crossfire Hurricane. That challenge may have been exacerbated by the intense operational tempo of Crossfire Hurricane, which is not typically the case in counterintelligence (as opposed to counterterrorism) matters.

Moreover, the FBI’s relationship with Steele—whom the bureau regarded as a confidential informant, but who understood himself as a contractor who had other clients, and who was handed off to the Crossfire Hurricane team by an agent who worked with him on other matters—was genuinely peculiar. The oddities of his handling gave rise to some, though by no means all, of the errors.

On the other hand, Horowitz also supplies ample reason to worry that the sort of comedy of errors he describes may be relatively typical. Because the Page FISA applications were politically sensitive, they received more review than was normal, he writes. People knew it was a high-profile, politically sensitive matter, so it’s reasonable to expect that agents were more, not less, apt to take care than in a run-of-the-mill case.

There’s another possibility too: that the case was, in fact, influenced by some kind of subtle political bias—albeit bias far less florid and conspiratorial than the president and his defenders imagine. Horowitz reports that he discovered no testimonial or documentary evidence that the errors in this case were driven by political attitudes, but he does not rule out the possibility—and in his testimony about the report, he left room for that interpretation. So we should consider the possibility that, though well short of the “insurance policy” conspiracy or a treasonous coup attempt, genuine alarm at the line level reinforced a kind of confirmation bias among agents who were sincerely concerned that the group of people around the Trump campaign were working with the Russians, leading them to cut corners and not to see defects in their informants and their case. As Sanchez writes,

The F.B.I.’s interest in Mr. Page—and its suspicions that he might be a Russian intelligence asset—predated his involvement in presidential politics. He had reportedly been the target of a FISA warrant in 2014 and was the focus of yet another counterintelligence investigation opened in April 2016 by the F.B.I.’s notoriously Trump-friendly New York field office, months before the bureau started an inquiry into potential links between the Trump campaign and Russia’s election interference operation. When investigators got wind of Christopher Steele’s notorious dossier, which made Mr. Page a pivotal figure in a “well-developed conspiracy of cooperation” between Mr. Trump and the Kremlin, it would have seemed like confirmation of what they already suspected.

Having adopted this theory, investigators began to exhibit classic signs of confirmation bias, readily absorbing new information that fit the model they’d built, while overlooking or explaining away facts that didn’t fit.

Confirmation bias and political bias are, of course, not mutually exclusive. One can suffer from confirmation bias because of a preexisting political bias, or even a nonpolitical bias that one’s investigative subject is a bad guy.

But that sort of bias, in turn, returns us to the question of how typical these errors are: Investigators, after all, normally don’t like the subjects they investigate. The opposite is true: They typically worry that their targets are up to no good. The FBI is investigating its subjects for a reason, after all—and that reason is usually not because the subjects foster rescue puppies. So if this more subtle form of bias was at work here—the bias of being alarmed by Trump, the group of people around him and their connections to Russia—we have to ask how often that sort of alarm leads to corners being cut.

In the end, we just don’t know how typical this is. And that’s actually what’s scary about the Horowitz report.

Or, to be precise, it’s one of the things that’s scary about it. The other is that it may be very difficult to figure out how typical these sorts of errors are. Only part of the problem identified by Horowitz involves misstatements of facts known to the bureau in the applications as filed. If that were the whole of the issue, the problem would be relatively easy to evaluate. The bureau has developed procedures for checking the accuracy of statements made in FISA applications—the so-called Woods Procedures—so it would be reasonably manageable to assess how often the applications filed with the court are not adequately supported by information in the Woods files that supposedly lie beneath them. Labor intensive? Absolutely. But it’s a manageable undertaking.

But Horowitz identifies a different issue: information that should have been in the FISA applications but was not. As Sanchez wrote,

The worst misrepresentations to the court that Mr. Horowitz uncovered are sins of omission—new information the bureau obtained as the investigation progressed that should have led it to question previous representations it had made to the court.

The many layers of review FISA applications go through—laid out in a set of rules known as the Woods Procedures—were ill equipped to detect this sort of problem, because the Woods Procedures focus on confirming that facts in the application match documents in the F.B.I.’s case file. But you can’t fact check a claim that doesn’t exist—which means the process is bad at detecting important information that has been left out.

In the government’s filing the other day with the FISC laying out its plans for fixing the problems the inspector general identified, the Justice Department acknowledged this point, writing:

Admittedly, these accuracy reviews do not check for the completeness of the facts included in the application. That is, if additional, relevant information is not contained in the accuracy sub-file and has not been conveyed to the [the Office of Intelligence] attorney, these accuracy reviews would not uncover the problem. Many of the most serious issues identified by the OIG Report were of this nature. Accordingly, [the Office of Intelligence] is considering how to expand at least a subset of its existing accuracy reviews at FBI field offices to check for the completeness of the factual information contained in the application being reviewed.

(Notably, on page 15 of this filing, which is designed to persuade the court of the bureau’s commitment to ensuring the accuracy of its submissions, the FBI’s general counsel, Dana Boente, managed to spell his own name incorrectly.)

In some ways, it wouldn’t be all that surprising if there are lots of errors in FISA applications. There are, after all, lots of errors in search warrants. There are lots of errors in investigations. The suppression motion in a criminal case is so normal that we don’t think much of it. The errors the inspector general found in the Carter Page case are, with the exception of the allegedly doctored email, the kind of errors one would not be surprised to see in a suppression motion filed in any of thousands of criminal cases filed every year.

But to observe this possibility is not to excuse it. We cannot accept this level of sloppiness in FISA applications. In criminal cases, after all, there are remedies—the main remedy being the motion to suppress. Law enforcement’s conduct gets flyspecked retroactively, as well as in an ex parte fashion in advance, because the charged defendant gets to file that suppression motion. And the consequences of a judge’s granting it are great enough to operate, most of the time, as their own inhibition against misconduct. So the system is, at some level, self-correcting.

Under FISA, by contrast, the suppression motion generally doesn’t materialize—since most FISA targets, like Page, do not end up being charged. Most, in fact, do not ever learn they were targeted. (Page is an anomaly in that respect.) What’s more, even in the occasional cases when a suppression motion does get filed, the litigation does not involve the disclosure of the underlying application to the accused and his or her lawyers.

This puts enormous weight on the integrity of the initial process—which is to say the relationship between the court and the bureau, and the relationship implicit in that between the bureau and the Justice Department lawyers who represent the bureau before the court. Without those two relationships functioning properly, the FISA process very quickly becomes hard to defend.

Notably, the key breakdown in the Horowitz report was not a breakdown in the relationship between the Justice Department and the court. Indeed, Horowitz describes no failures of candor between the Justice Department Office of Intelligence (OI) attorneys and the FISC. The problems he’s identified, rather, lie in the relationship within the executive branch—that is, between the bureau and the Justice Department. These relationships are also taking place low down in the totem pole, among line and supervisory agents and a line lawyer. If there are routine problems in communications between those folks and OI attorneys on FISA matters, we have a very deep problem on our hands. To the extent that Justice Department-FBI communication is generally inadequate, and the FISA process is generally corrupted as a result, we simply have no choice but to fix the lines of communication to make them function better.

On this point, one would hope there would be political consensus—though apparently there is not. Some voices on the Trumpish right these days seem to be advocating abolishing the FISC. “Congress should consider getting rid of FISA courts altogether,” writes William McGurn in the Wall Street Journal. “Because without judges to hide behind, executive officials who order spying on their fellow citizens will have to own those decisions themselves.”

This is very dangerous nonsense. On Planet Earth, there is no realistic alternative to some kind of FISA process. Whatever McGurn may imagine, the history of executive branch surveillance authority is not a story of accountability for surveillance decisions. Conversely, treating FISA warrants like any other warrants, run through the normal Article III courts, is impracticable because of the pervasiveness in FISA warrants of material classified at a level normal courts can’t handle. So unless one intends to abolish national security surveillance altogether, one is realistically talking about mending the FISA process, not ending it.

And mending it requires realistically assessing the scope of the problem. The audit of FISA applications that Horowitz has announced is key in this regard—and regular routine audits of some FISA applications are probably important to institutionalize. The proper remedies are going to be very different if these problems crop up occasionally than if factual rigor in FISA applications is generally deficient. If the problem here was that an investigation at FBI headquarters with a complicated relationship with a confidential source led to pervasive confusion among a few Keystone Kops, we will think about the whole thing differently than we will if this sort of problem arises more frequently—say, in high-stakes investigations generally. And we will think about that problem, in turn, differently than we will think about the problem of the average FISA application being of inadequate factual rigor. So the first discipline here is learning more about the problem itself—and waiting for the results of Horowitz’s ongoing inquiry before drawing firm conclusions.

In the meantime, the bureau is tightening some procedures in a manner that seems appropriate. Some of these measures appear to be targeted at the completeness problem. FBI Director Christopher Wray, for example, in his declaration to the FISC the other day, mentioned specifically: “Adding an affirmation to the FISA Verification Form that, to the best of the agent’s and supervisor’s knowledge, OI has been apprised of all information that might reasonably call into question the accuracy of the information in the application or otherwise raise doubts about the requested probable cause findings or the theory of the case.”

One additional step that Wray did not announce, but is worth thinking about, is requiring the case agents—not just the supervisory agents—to affirm to the court the accuracy and completeness of the applications. Kris floated this idea in his article a few weeks ago, writing that “the executive branch might consider using field agents, rather than headquarters agents, as FISA affiants, or otherwise changing the way in which FISA applications and/or the Woods Procedures are implemented and managed.” The people closest to the ground in any investigation are the people who actually conduct that investigation. And if the government is looking for confidence that the application has disclosed everything the FBI has collected that is relevant to the court’s consideration of an application, the folks who collected the information in the first place may be the essential actors. Forcing them under oath to affirm the file’s completeness may be a good way to force people to rethink how they’ve characterized things, what they’ve put in and what they’ve left out.

The point here is that the problem Horowitz has identified is a really hard one. It doesn’t help to cram these issues into preexisting political categories. It also doesn’t help to assume we have enough information already to address them. The solution is going to be messy. Identifying the scope of the problem may be painful as well.