National Day Laborer Organizing Network v. United States Immigration & Customs Enforcement Agency

OPINION AND ORDER

SHIRA A. SCHEINDLIN, District Judge:

“Sunlight is said to be the best of disinfectants; electñc light the most efficient policeman. ”¹

I. INTRODUCTION

The parties in this litigation have starkly different viewpoints about the wisdom and legitimacy of Secure Communities, which, as an aspect of national immigration policy, is a particularly sensitive and important topic. The three plaintiffs — the National Day Laborer Organizing Network, the Center for Constitutional Rights, and the Immigration Justice Clinic of the Benjamin N. Cardozo School of Law — did not file this lawsuit seeking information about the program solely out of curiosity or a commitment to government transparency. They did it as part of a major campaign calling to “End Secure Communities. Don’t Mend It. Pledge to Break ICE’s Hold on Your Community.”² Nonetheless, under the Freedom of Information Act (“FOIA”), plaintiffs are entitled to the disclosure of as much information as the law permits.

The Act is intended to facilitate transparency about the government’s policies even — or perhaps especially — when members of the public are disturbed by those policies and are fighting to end them. The Act calls on government employees to diligently and honestly respond to requests even from people with whom they disagree. And it calls upon the federal courts and the attorneys who are officers of those courts to cooperate so that the public will have access to information in an efficient, effective, and timely manner. Defendants note that they have spent thousands of hours and hundreds of thousands of dollars responding to plaintiffs’ request.³ Transparency is indeed expensive, but it pales in comparison to the cost to a democracy of operating behind a veil of secrecy. This litigation has influenced much of the public debate over Secure Communities. The Act has therefore served its purpose of engendering a more informed public and a more accountable government.

Plaintiffs bring this action for the purpose of obtaining records, pursuant to FOIA,⁴ from the United States Immigration and Customs Enforcement Agency (“ICE”), United States Department of Homeland Security (“DHS”), the Executive Office for Immigration Review (“EOIR”), the Federal Bureau of Investigation (“FBI”), and the Office of Legal Counsel (“OLC”). Specifically, plaintiffs have sought information regarding Secure Communities, a federal immigration enforcement program launched in 2008. It has long been the practice for local law enforcement agencies to send the fingerprints of individuals arrested and booked into custody to the FBI to be checked against the national criminal history database.⁵ Under the Secure Communities program, those fingerprints are also now sent to DHS to be checked against immigration records.⁶ A portion of the requested records related to the issue of whether and how state and local law enforcement agencies may “opt-out” of participation in Secure Communities.

Plaintiffs’ FOIA request was twenty-one pages long. The parties eventually negotiated a Rapid Production List (“RPL”) — a limited list of key categories that would be produced by defendants on an expedited basis. In December, 2010, after defendants failed to comply with their obligations under the agreement, I entered an order directing them to produce records relating to the opt-out question by January 17, 2011 and the remainder of RPL

documents by February 25, 2011.⁷ The defendants’ searches involved hundreds of employees and thousands of hours and resulted in the production of tens of thousands of responsive records.⁸ The parties now cross-move for summary judgment on the adequacy of those searches — the defendants arguing that the searches satisfied their obligations under FOIA and the plaintiffs arguing that the searches were legally inadequate.

In support of their motions, each of the five defendant agencies has submitted a declaration (or, in the case of the FBI and ICE, two declarations each) written by one of its FOIA officers.⁹ The declarations range from six to thirty-four pages. They describe which offices (and sometimes which custodians) conducted searches for records. And, in different levels of detail, they describe how custodians searched for records.

Plaintiffs make two broad critiques of the agencies’ searches.¹⁰ First, they argue that the agencies failed to conduct any searches of the files of certain custodians who were likely to possess responsive records. Second, plaintiffs argue that defendants have not established that the searches that they did conduct were adequate. This argument itself has two parts: First, that the agencies’ affidavits are insufficiently detailed and therefore do not permit a finding of adequacy; and, second, that even those searches that were described more fully were inadequate.¹¹ I assess each of these arguments below.

The inquiry is intensely fact-specific, particularly because it involves such a massive search. Generalizations about the quality of defendants’ searches are difficult because some of the searches appear to have been extremely rigorous, some woefully inadequate, and many simply documented with detail insufficient to permit proper evaluation. For the reasons stated below, the motions of OLC and EOIR are granted¹² and the motions of ICE, the FBI, DHS, and plaintiffs are granted in part and denied in part.

II. LEGAL STANDARD

In order to win summary judgment under FOIA, an agency must “show beyond material doubt that it has conducted a search reasonably calculated to uncover all relevant documents.”¹³ “An agency is not expected to take extraordinary measures to find the requested records, but only to conduct a search reasonably designed to identify and locate responsive documents.”¹⁴

FOIA cases are generally resolved on motions for summary judgment,¹⁵ which, as in any other context, requires that the moving party “show that there is no genuine issue as to any material fact and that the moving party is entitled to judgment as a matter of law.”¹⁶ “An issue of fact is genuine if ‘the evidence is such that a reasonable jury could return a verdict for the nonmoving party.’ A fact is material if it ‘might affect the outcome of the suit under the governing law.’ ”¹⁷ “In ruling on a motion for summary judgment, a court must resolve all ambiguities and draw all factual inferences in favor of the nonmoving party.”¹⁸

As the Second Circuit has explained,

[i]n order to prevail on a motion for summary judgment in a FOIA case, the defending agency has the burden of showing that its search was adequate .... Affidavits or declarations supplying facts indicating that the agency has conducted a thorough search ... are sufficient to sustain the agency’s burden ... [and] are accorded a presumption of good faith.¹⁹

Summary judgment is inappropriate “where the agency’s response raises serious doubts as to the completeness of the agency’s search, where the agency’s response is patently incomplete, or where the agency’s response is for some other reason unsatisfactory.”²⁰ In addition, plaintiffs may defeat summary judgment if they can show “some tangible evidence” that defendants have not satisfied their burden.²¹ Where “an agency has not satisfied its burden, a showing of bad faith is not necessary” in order to defeat a motion for summary judgment.²²

In their affidavits, agencies must “ 'identify the searched files and describe at least generally the structure of the agency’s file system’ which renders any further search unlikely to disclose additional relevant information.”²³ They must establish that they searched all custodians who were reasonably likely to possess responsive documents.²⁴ And they must “set[ ] forth the search terms and the type of search performed.”²⁵

“The adequacy of a FOIA search is generally determined not by the fruits of the search, but by the appropriateness of the methods used to carry out the search.”²⁶ However, “[ejvidence that relevant records have not been released may shed light on whether the agency’s search was indeed inadequate.”²⁷

III. DISCUSSION

A. The Agencies’ Selection of Custodians to Search

I begin by assessing the agencies’ choices to exclude certain custodians from their searches.

1. ICE

Deputy Director — ICE had not conducted a search of the Deputy Director’s records as of the date that it filed this motion. After plaintiffs highlighted this failure, ICE conducted what appears to have been a rigorous search — using search terms recommended by plaintiffs— and produced fifty-six responsive records on May 11, 2012. Although the failure to search the Deputy Director’s records for a year and a half was unreasonable and therefore inadequate as a matter of law, defendants appear to have finally conducted an adequate search of his records.²⁸

Homeland Security Investigations (HSI): With one exception not at issue here, this office did not search for records.²⁹ Other than the records of individual aliens and two sample worksheets that were produced, HSI “determined the office would not be likely to have any other responsive records.”³⁰ However, according to documents produced by other ICE offices, at least one HSI employee was involved in opt-out discussions in September, 2010.³¹ HSI played a role in enforcing Memoranda of Understanding between local law enforcement agencies and ICE³² and, in 2011, was tasked, along with other ICE programs, with “providing] an update on Secure Communities and its continued expansion.”³³ ICE has not suggested that HSI became involved in Secure Communities only after the opt-out deadline or explained the existence of these documents (including the 2010 emails) in light of HSI’s denial that it had any responsive records.³⁴ It has therefore failed to carry its burden of establishing that it was reasonable not to search HSI for records.

Contractors: As plaintiffs point out, multiple private contractors “played a direct role in Secure Communities [and in] discussions and decision-making on the opt out issue.”³⁵ The ICE declaration does not make clear the extent to which the records of outside contractors were searched and ICE attorneys argue that “plaintiffs incorrectly assume that discussions of staff [described in ICE’s Law Declaration] included only agency-employed staff and not outside contractors.”³⁶ According to a letter from one contractor who was publicly involved in the opt-out controversy to the ICE assistant director, the FOIA office did instruct him to search his records and he did provide a “comprehensive response.”³⁷ The extent to which outside contractors were included in ICE’s search is unclear from the Law Declaration, which makes it impossible for me to determine whether an adequate search of those contractors’ records was conducted. ICE must therefore inform plaintiffs about the extent to which contractors were included in the searches, and, if plaintiffs are dissatisfied, submit to the Court a supplemental declaration.

Office of State, Local, and Tribal Coordination (OSLTC): According to ICE, only two members of its OSLTC searched their records for responsive documents.³⁸ Plaintiffs argue that other senior OSLTC officials “were directly involved in discussions and outreach with state and local officials relating to Secure Communities and the opt-out issue.” ³⁹ For example, on January 21, 2011 (which was, admittedly, after the opt-out cut-off), the OSLTC Chief of Staff circulated to other ICE officials a summary of OSLTC’s outreach efforts; according to the document, the OSLTC met the next week to discuss how it would “leverage support for Secure Communities at the state government level.”⁴⁰ Yet the Chief of Staff did not conduct a search for responsive records. ICE argues that plaintiffs “fail to explain why [the] two custodians [who did search their records] would not have been the most likely OSLTC custodians to have responsive records in light of Law’s description.”⁴¹ But the government is not required to search only the files of the two custodians who are “most likely” to have responsive records; it must also search other locations that are reasonably likely to contain records.⁴² Because the OSLTC Chief of Staff was circulating memoranda regarding the office’s outreach efforts relating to opt-out, his or her files likely contain responsive records and should have been searched.⁴³

Privacy Office: The ICE Privacy Office informed the ICE FOIA office that it “would not likely have any [responsive] records” and it therefore did not conduct a search.⁴⁴ Plaintiffs argue that because the office’s Privacy Officer attended at least one high level Secure Communities meeting, she/he should have conducted a search for records.⁴⁵ The officer’s presence at one meeting, however, is not sufficiently probative of the existence of records in the office’s control so as to render unreasonable the government’s determination that the Privacy Office was not likely to have responsive records.

2. FBI

The Director’s Office: The government’s decision to make the FBI’s fingerprint database (called “IAFIS”) compatible with DHS’s immigration database (called “IDENT”) is a central feature of Secure Communities.⁴⁶ Plaintiffs have produced ample documentary evidence showing that the FBI Director Robert Mueller was personally involved in the decision to make IAFIS and IDENT “interoperable” and in the rollout of Secure Communities and that he was familiar with the opt-out question. Plaintiffs’ evidence includes the following: Director Mueller provided testimony to the Senate in late 2006 regarding interoperability and the first pilot projects with local law enforcement agencies;⁴⁷ a March, 2010 correspondence between DHS and the FBI notes that Mueller would have to personally approve changes to Washington, D.C.’s participation in Secure Communities;⁴⁸ on July 21, 2010, Mueller met with DHS Secretary Napolitano “to discuss information sharing”;⁴⁹ in late July, 2010, high ranking officials in the Director’s office received correspondence regarding congressional inquiries into the opt-out question and in September they received copies of Secretary Napolitano’s response to the congressional inquiry;⁵⁰ and at an August 30, 2010 FBI meeting that focused on interoperability and the opt-out question, an official stated that “the Director is aware of Secure Communities questions and is prepared to answer them.”⁵¹

However, the FBI did not search the Director’s files for Opt-Out or RPL documents from the pre-October 15, 2010 period.⁵² Defendants justify this decision on the basis that “the FBI had no reason to believe that anyone in the Director’s office — including the Director, any Deputy Director, Associate Deputy Director, Chief of Staff or Senior Counsel — contained [sic] responsive RPL or Opt-Out records.”⁵³ Defendants argue that “[i]t was only later, when the opt-out issue became more prominent, that the Director’s Office became involved in that issue” and that the FBI therefore produced records from the Director’s Office relating to opt-out from after the October 15, 2010 deadline for the opt-out and RPL production.⁵⁴

Defendants do not and cannot dispute that by November 18, 2010, the Director’s Office was involved in the opt-out issue.⁵⁵ But they do not say when the Director’s Office became involved in the issue or provide any justification for the fact that the earliest irrefutable evidence of the Office’s involvement corresponds closely to the Court’s arbitrary cut-off date for the search of opt-out records.⁵⁶’ The FBI searched for records from the Director’s Office from after October 15, 2010.and it found them; it now tries to justify not searching for pre-October 15 records based on the argument that the Office was not involved in the issue at that time. The FBI’s argument — unsupported by any evidence except the word of a FOIA officer from outside the Director’s Office and refuted by the documentary evidence marshaled by plaintiffs — is insufficient to carry its burden of showing that it conducted a search reasonably calculated to uncover all relevant records.

Office of General Counsel: The FBI FOIA office sent its search memorandum to the National Security Law Branch in the Office of General Counsel (OGC). It did not receive confirmation from OGC that a search had been conducted but nevertheless “viewed a non-response as a ‘no records’ response.”⁵⁷ This assumption was plainly improper. Agencies can satisfy their FOIA obligation by submitting “[ajffidavits or declarations supplying facts indicating that the agency has conducted a thorough search.”⁵⁸ It is absurd to suggest that they may satisfy their obligation by submitting a sworn declaration from a FOIA officer asserting that he has requested that an office perform a search, has received no response from the office, and therefore assumes that a proper search was performed and no documents were found.⁵⁹

The parties disagree over the scope of the search that OGC should have conducted.⁶⁰ Defendants argue that, as with the Director’s Office, “the FBI had no reasonable basis to believe that custodians within the Office of General Counsel ... would have had responsive records during the time period relevant to the opt-out and RPL searches.”⁶¹ However, plaintiffs point out that as early as 2009, the Privacy and Civil Liberties Unit in the OGC was involved in assessing the privacy implications of interoperability (which is responsive to RPL Item VII)⁶² and that, after October 15, 2010 and in part because of this lawsuit, it was involved in the opt-out question.⁶³

Because the National Security Law Branch of OGC never responded to the initial instructions from the FOIA office to conduct a search and because OGC was involved in the opt-out question after October 15 (and at least discussed the privacy concerns regarding interoperability before October 15), the FBI has not met its burden of showing that it conducted a reasonable search. However, plaintiffs have not yet convinced the Court that OGC should have conducted an opt-out search: it is plausible that, as defendants argue, OGC only became involved in the opt-out question after plaintiffs sought a preliminary injunction in this lawsuit. Therefore, OGC has the option of either conducting and fully documenting a new search or submitting a declaration from a supervising attorney stating that, based on her personal knowledge, it is not reasonable to believe that the office has documents responsive to the Opt-Out requests and that no search is therefore required under FOIA. The Privacy and Civil Liberties Unit is required, however, to conduct a search for documents responsive to RPL VII.

Science and Technology Branch and Office of Law Enforcement Coordination: Plaintiffs argue that the FBI should have searched the records of the Office of Law Enforcement Coordination (“OLEC”) and of the Executive Assistant Director (EAD) for the Science and Technology Branch (“STB”).⁶⁴ Given the work of these two offices (and particularly given the EAD’s statement in May, 2011 that he had been tracking the issue of opt-out “for a while”) plaintiffs reasonably assume that the two offices possessed responsive documents.⁶⁵ However, there is no documentary evidence — other than a mass email regarding a congressional inquiry into the opt-out question that OLEC and the EAD received (along with numerous other offices) on July 30, 2010 — that either office was involved in these discussions prior to October 15, 2010. Because the agencies are more familiar with their work than the plaintiffs or the Court, they are entitled to some degree of deference regarding their determination of search locations.⁶⁶ In this case, because the personnel who report to the EAD searched their files and did not have responsive records, defendants are correct that “it was reasonable to conclude that the EAD would not have had any [responsive] records.” ⁶⁷ Plaintiffs offer no evidence to refute the FBI’s presumably reasonable decision not to search OLEC’s records.

Interoperability Initiatives Unit (“IIU”): “The vast majority of the potentially responsive records” collected by the FBI’s FOIA office came from the IIU — a sub-component of the Criminal Justice Information Services Division (“CJIS”) — which is responsible for the FBI’s collaboration with DHS regarding Secure Communities.⁶⁸ Although IIU employees conducted a manual review of the unit’s shared drives for responsive documents and each conducted a search of their emails, no search was done of the files of seven former IIU employees who had worked on Secure Communities.⁶⁹ Defendants argue that “given the unit’s practice of saving records on the IIU shared drives, the manual search of shared drives would have likely retrieved most of the former IIU employee’s records.”⁷⁰ FOIA, however, requires that agencies conduct a search “reasonably calculated to uncover all relevant documents,” not “most” relevant documents.⁷¹ Furthermore, although such a search of the shared drive might have recovered most of the former employees’ word processing documents, it would not have recovered their emails and the FBI did not search for those emails. Given that the IIU only asked twenty-two active employees to search their files, the failure to search the files of seven former employees was not de minimis and made the unit’s search inadequate.

3. BHS

Secretary Napolitano: According to the declaration of its FOIA officer, DHS searched “any documents sent to or from the Secretary.”⁷² Secretary Napolitano does not have an email account, but all documents sent to and from her office, including to and from the Secretary herself, are maintained “in electronic and paper format.”⁷³ Plaintiffs argue that this search was insufficient, because “ ‘documents sent to or from the Secretary’ are likely not to encompass all the documents the Secretary possessed or created.”⁷⁴ Plaintiffs are apparently speculating that perhaps the Secretary has herself created responsive documents that she has not shared with anybody else. Given her role as the head of an agency with a quarter million employees, I find such a prospect extremely unlikely. The search of all documents sent to or from the Secretary examined a universe of documents reasonably calculated to uncover all responsive documents.

Office of General Counsel: DHS did not conduct a search of the files or communications of the General Counsel (GC), Principal Deputy GC, or of some of the Deputy GCs.⁷⁵ Emails show, however, that Principal Deputy GC David Martin was involved in the opt-out deliberations no later than September, 2010 and was circulating legal analyses of the . question within DHS.⁷⁶ As I have previously recounted at length, Martin reviewed and endorsed ICE’s (and subsequently DHS’s) analysis of the opt-out question.⁷⁷ The failure to search his records was improper. In his September 3 email (which he apparently sent “To” one person and cc’d numerous others), Martin summarized his opinion on the basic opt-out issue and said that “you (or maybe [redacted]) are probably the logical person to attend [a high-level meeting regarding opt-out] with me, but let me know your advice on that. Also, to anyone else copied on this email, please let me know if there are others within hq OGC with relevant expertise.”⁷⁸ Martin believed that the email’s recipient and the person whose name was redacted were familiar with the opt-out issue; at the very least, DHS had the responsibility to follow up on that obvious lead and search the records of those two individuals. And depending on what correspondence or other records DHS found during the searches of Martin and those two, it may also have had the duty to search the records of other people who were cc’d on Martin’s email.⁷⁹

It is unclear to the Court whether General Counsel Ivan Fong was involved in any discussions related to opt-out or the RPL prior to the cut-off date, or whether he only became involved in the discussions later.⁸⁰ DHS should ask him. If he says that he was involved or if the records from Martin, Anderson, and the two employees whose names were redacted reveal that Fong was involved, then DHS’s failure to search his records was unreasonable and it must conduct such a search.

Homeland Security Advisory Council (HSAC): No search was conducted of HSAC, which provides advice and recommendations to the Secretary. DHS has not explained why no search was conducted; it has not even declared that it made a determination that HSAC was unlikely to possess responsive records.⁸¹ Defendants have the burden of establishing that they conducted a reasonable search. In the absence of any sworn submission from defendants regarding whether or not HSAC maintained responsive records, and given that HSAC did discuss Secure Communities as early as 2009⁸² and its mission to advise the Secretary, defendants have not met their burden.

U.S. Visitor and Immigrant Status Indicator Technology (VISIT): US-VISIT searched the files of six of its employees for responsive records. Plaintiffs argue that US-VISIT should have searched the files of its Chief Information Officer (CIO) and its Deputy Director as well. Defendants say that the CIO “focuses on macro-level program issues, such as the overall architectures of the IDENT and ADIS databases, and not project-level matters or smaller efforts, such as Secure Communities.”⁸³ Plaintiffs argue that “[t]his is precisely the type of information that should have been located, ie. records related to the technical capacity of DHS (ie. the architecture of its databases) to limit information sharing with the FBI.”⁸⁴ However, plaintiffs defined their request for opt-out records as those “relating to the ability of states or localities to decline or limit participation in Secure Communities, including ... [those] referencing the technological capacity of ICE and the FBI to honor requests to opt-out, opt-in or limit participation in Secure Communities.”⁸⁵ This was a limited request made in order to expedite the documents most important to plaintiffs. It was crafted in narrow language seeking documents related to opt-out and referencing the technological capacity to honor opt-out requests. Not all documents relating to the departments’ technical capacities in general were responsive to the request. Accordingly, DHS acted reasonably in not searching records of the CIO, whom it determined would not likely have responsive records. Plaintiffs similarly fail to point to any evidence that the US-VISIT Deputy Director had responsive records prior to 2011.

4. Office of Legal Counsel (OLC)⁸⁶

OLC, which assists the Attorney General in his role as legal advisor to the President, is a small office that employs “approximately twenty attorneys at any one time.”⁸⁷ In response to plaintiffs’ request, OLC (1) searched the office’s shared drive, which contains all final OLC advice; (2) ran searches of the email files of attorneys who had departed the office; (3) asked two long-time career attorneys familiar with the office’s work whether anybody had worked on Secure Communities and; (4) sent an email to all current attorneys asking if they had worked on Secure Communities or on law enforcement information sharing. The queries revealed (1) two declarations from ICE officials that were prepared as part of the litigation over Arizona’s S.B. 1070 and that mention Secure Communities in passing; (2) no final OLC advice; (3) no responsive emails from departed attorneys; and (4) no recollection of ever having worked on Secure Communities.

Plaintiffs argue that this search was inadequate. Although the ideal search might have been conducted slightly differently, the OLC’s search was reasonably designed to uncover all responsive documents and was thus adequate. First, the seventeen search terms used to search the shared drive — which included “ICE” and “secure communities” — were reasonable. While terms such as “SC” or “interoperability” were not used, it is extremely unlikely that any responsive final opinions would not have used either the term “ICE” or “secure communities.”⁸⁸ Second, the discussions with long-time supervising attorneys wisely helped guide the office’s search. Plaintiffs’ argument that these attorneys’ memories were faulty is unsupported; the fact that two other attorneys worked on the Arizona litigation and possessed two minimally responsive documents does not mean that the supervising attorneys were wrong. Finally, the query of every attorney in the office was comprehensive. Plaintiffs are correct that the FOIA officer should have followed up with the attorneys who did not respond to the email query. However, in light of the totality of steps that it took to search for responsive documents (and the fact that none of them revealed that its attorneys had worked on Secure Communities), the OLC’s search was adequate.

B. The Search for Records By Custodians

Plaintiffs argue that, in addition to improperly excluding from their searches certain custodians, (1) the agencies provided custodians with vague search instructions and (2) the custodians used inadequate search terms or did not specify the search terms that they used.⁸⁹ As a result, plaintiffs argue, the searches were not reasonably designed to uncover all responsive records. Defendants disagree.

1. Summary of Agencies’ Search Terms and Instructions

The ICE FOIA office initially did not provide suggested search terms to the program offices tasked with conducting searches because the plaintiffs’ request “was so broad and covered such a wide swath of documents.”⁹⁰ However, once the parties agreed to prioritize opt-out and RPL records, the FOIA office compiled search instructions that listed eight suggested search terms and “specified that the searches were to include the full text of documents ... [but] did not address the issue of combining any of the search terms or using any connectives [such as and, or, w/10].”⁹¹

ICE describes its searches with different levels of detail, depending on the office. For example, Law says that a number of components in ICE’s Office of Principal Legal Advisor (OPLA) conducted searches using the eight recommended search terms and the term “Secure Communities” and says that other offices used fewer search terms.⁹² However,' Law does not describe the search terms used by the more than one hundred employees in the Office of Enforcement and Removal Operations — which is responsible for implementing Secure Communities — who searched their records.⁹³ Instead, he says only that they were provided with the recommended search terms. The same is true of one other office in ICE.⁹⁴

The majority of the FBI’s search was conducted via a “manual” review of the records in the IIU.⁹⁵ The FBI FOIA office, like the ICE FOIA office, issued an internal search memorandum for circulation to custodians tasked with conducting searches.⁹⁶ In contrast to ICE, however, the FBI did not submit its memorandum as part of the summary judgment briefing and there is no indication that the memorandum contained recommended or mandatory search terms. The FBI did not instruct custodians to search archived records and it is unclear if it gave any guidance about how the .search should be conducted.⁹⁷

DHS, like ICE, recommended eight search terms but did not require that they be used; supervisors “did not monitor the use of search terms, but rather relied upon the custodians to conduct appropriate searches given the requests.”⁹⁸ OLC, in contrast, provided the precise terms that its employees used to search individual and shared sources.⁹⁹ Finally, the EOIR, whose search plaintiffs do not challenge, provided specific mandatory search terms to custodians and confirmed to the Court that the custodians used those terms.¹⁰⁰

2. Analysis of Defendants’ Searches

It is impossible to evaluate the adequacy of an electronic search for records without knowing what search terms have been used. In earlier times, custodians and searchers were responsible for familiarizing themselves with the scope of a request and then examining documents individually in order to determine if they were responsive. Things have changed. Now custodians can search their entire email archives, which likely constitute the vast majority of their written communications, with a few key strokes. The computer does the searching. But as a result, the precise instructions that custodians give their computers are crucial.

At the most elementary level are simple mistakes: a search for “secure commmunities” (with three “m”s) may yield no results despite the presence of thousands of documents containing the phrase “secure communities.” Seemingly minor decisions — whether intentional or not — will have major consequences. Choosing “subject field” rather than “subject field and message body” during a search using the Microsoft Outlook email client will dramatically change its scope and results. Boolean operators are also consequential: a search for “secure communities opt-out” may yield no results while a search for “ ‘secure communities’ and ‘opt-out’ ” yields one hundred results and a search for “ ‘secure communities’ or ‘opt-out’ ” yields ten thousand. As I have previously explained, “search results will change dramatically depending on which logical connectives — such as ‘and,’ ‘or,’ ‘w/10,’ — are used.”¹⁰¹ Thus, “[i]n order to determine adequacy, it is not enough to know the search terms. The method in which they are combined and deployed is central to the inquiry.” ¹⁰²

Describing searches with this level of detail was not necessary in the era when most searches took place “by hand.” Then, as now, a court largely relied on the discretion of the searching parties to determine whether a document was responsive; but at least in that era, courts knew that the searching parties were actually looking at the documents with their eyes. With most electronic searches, custodians never actually look at the universe of documents they are searching. Instead, they rely on their search terms and the computer to produce a subset of potentially responsive records that they then examine for responsiveness.

Yet the FBI, to take one example, has given the Court no specific information about the search that it conducted beyond explaining that much (but not all) of it was “manual.”¹⁰³ For the portions that were not manual, I do not know what search terms were used, let alone how they were combined. I do not even know if any search terms were recommended.¹⁰⁴ Similar problems permeate many of the searches of DHS and ICE files, although not of OLC or EOIR files.

Defendants argue that I should grant the agencies’ motion on the adequacy of the search even though I do not know what search terms — let alone what Boolean operators, search fields, and time frames — were used by a very large portion of the custodians. As a general matter, they point out that “declarations need not ‘set forth with meticulous documentation the details of an epic search’ in order to entitle agencies to summary judgment.”¹⁰⁵ And they argue that “[i]t is also unclear why custodians could not be trusted to run effective searches of their own files, a skill that most office workers employ on a daily basis.”¹⁰⁶

There are two answers to defendants’ question. First, custodians cannot “be trusted to run effective searches,” without providing a detailed description of those searches, because FOIA places a burden on defendants to establish that they have conducted adequate searches; FOIA permits agencies to do so by submitting affidavits that “contain reasonable specificity of detail rather than merely conclusory statements.”¹⁰⁷ Defendants’ counsel recognize that, for over twenty years, courts have required that these affidavits “set[ ] forth the search terms and the type of search performed.”¹⁰⁸ But, somehow, DHS, ICE, and the FBI have not gotten the message. So it bears repetition: the government will not be able to establish the adequacy of its FOIA searches if it does not record and report the search terms that it used, how it combined them, and whether it searched the full text of documents.

The second answer to defendants’ question has emerged from scholarship and caselaw only in recent years: most custodians cannot be “trusted” to run effective searches because designing legally sufficient electronic searches in the discovery or FOIA contexts is not part of their daily responsibilities.¹⁰⁹ Searching for an answer on Google (or Westlaw or Lexis) is very different from searching for all responsive documents in the FOIA or e-discovery context.¹¹⁰ Simple keyword searching is often not enough: “Even in the simplest ease requiring a search of online e-mail, there is no guarantee that using keywords will always prove sufficient.” ¹¹¹ There is increasingly strong evidence that “[k]eyword searching] is not nearly as effective at identifying relevant information as many lawyers would like to believe.”¹¹² As Judge Andrew Peck — one of this Court’s experts in e-discovery— recently put it: “In too many cases, however, the way lawyers choose keywords is the equivalent of the child’s game of ‘Go Fish’ ... keyword searches usually are not very effective.”¹¹³

There are emerging best practices for dealing with these shortcomings and they are explained in detail elsewhere.¹¹⁴ There is a “need for careful thought, quality control, testing, and cooperation with opposing counsel in designing search terms or ‘keywords’ to be used to produce emails or other electronically stored information.”¹¹⁵ And beyond the use of keyword search, parties can (and frequently should) rely on latent semantic indexing, statistical probability models, and machine learning tools to find responsive documents.¹¹⁶ Through iterative learning, these methods (known as “computer-assisted” or “predictive” coding) allow humans to teach computers what documents are and are not responsive to a particular FOIA or discovery request and they can significantly increase the effectiveness and efficiency of searches. In short, a review of the literature makes it abundantly clear that a court cannot simply trust the defendant agencies’ unsupported assertions that their lay custodians have designed and conducted a reasonable search.

The more complicated question is this: when custodians do keep track of and report the search terms that they have used, how should a court evaluate their adequacy? As the cases cited by the parties show, the evaluation of search terms is highly context-specific: the failure to use certain search terms will sometimes be fatal,¹¹⁷ sometimes unproblematic,¹¹⁸ and sometimes improper but harmless or at least mitigated.¹¹⁹ Furthermore, even courts that have carefully considered defendants’ search terms have generally not grappled with the research showing that, in many contexts, the use of keywords without testing and refinement (or more sophisticated techniques) will in fact not be reasonably calculated to uncover all responsive material.

Plaintiffs have enlisted an e-discovery expert to analyze defendants’ searches. He argues that even where defendants have listed the search terms that they used, “there is no indication that [the agencies] undertook any analysis to determine whether there were other words that should have been included in their search[es], including, for example, a review of a sample set of the documents that did not contain the ... search terms.”¹²⁰ There is, he points out, an “absence of any evidence of a thoughtful process in selecting and testing search terms.”¹²¹

I accept the conclusion of plaintiffs’ expert that many of these searches were not perfect; the question, however, is whether the shortcomings on the part of the agencies made their searches “inadequate” under FOIA. Surely, the agencies have failed to establish the adequacy of the searches for which they have specified no search terms. But for those searches for which terms were specified, a determination is more difficult: on the one hand, the agencies did use search terms — like “opt out” and “secure communities” — that reflect the core topics sought by plaintiffs; on the other hand, there was no testing of the efficacy of those terms. “Since they are word-based, [keyword and Boolean searches] often fail to identify responsive documents because the author used different words to discuss the subject.”¹²² In response to these challenges, one judge has said that “[g]iven this complexity, for lawyers and judges to dare opine that a certain search term or terms would be more likely to produce information than the terms that were used is truly to go where angels fear to tread.”¹²³

Aware of the limitations of keyword searching and in the absence of evidence showing the efficacy of the terms used, it is impossible for me to assess the adequacy of most of the keyword searches used by defendants.¹²⁴ But it is also unnecessary for me to do so. In this case, which concerns the largest FOIA search in the history of ICE and an enormous search for DHS and the FBI as well, repeating vast swaths of the search in order to ensure adequacy is a waste of resources: given the level of duplication and the staleness of whatever new documents are uncovered, the costs to all parties would surely outweigh the benefits to plaintiffs, the public, and the rule of law. Nevertheless, FOIA requires the government to respond adequately to requests from the public and defendants must learn to use twenty-first century technologies to effectuate congressional intent.

Rather than fully revisit old searches, the parties will need to work cooperatively to design and execute a small number of new, targeted searches. Custodians who should have searched their records but did not will need to conduct complete searches; this order requires defendants to do no more than they should have to comply with FOIA initially.

In addition, a sample of the custodians who conducted searches but failed to provide the Court with any details about those searches will also need to conduct new, fully-documented searches; so will a smaller sample of the custodians who listed the search terms that they used but provided no evidence about the efficacy of those terms. These repeat searches will permit the parties and the Court to efficiently evaluate whether the initial searches were adequate.¹²⁵

The parties will need to agree on search terms and protocols — and, if necessary, testing to evaluate and refine those terms. If they wish to and are able to, then they may agree on predictive coding techniques and other more innovative ways to search. Plaintiffs will need to be reasonable in their demands — aware of the real cost that their massive FOIA request has imposed on the agencies — and will be restricted to seeking records from only the most important custodians on only the most important issues. Defendant agencies, in turn, will need to cooperate fully with plaintiffs. As in the past, the Court will supervise this process and provide a variety of mechanisms for resolving any disputes. Disagreements will be resolved early, before they lead to inadequate (or wasteful) searches.

Defendants shall conduct new searches of the following custodians (including searches of archived records), using a list of search terms and methodologies agreed to by the parties:

FBI:¹²⁶

1. FBI Director, Deputy Director, Associate Deputy Director, Chief of Staff, Senior Counsel.

2. General Counsel and Deputy General Counsel (unless a supervising attorney from OGC submits a declaration based on personal knowledge stating that it is not reasonable to believe that the office has documents responsive to the Opt-Out or RPL requests).¹²⁷

3. All former IIU employees whose individual records were not searched and three of the twenty-two IIU employees, selected by plaintiffs, who previously conducted “manual” searches.¹²⁸

4. Three custodians, chosen by plaintiffs, in the Advisory Groups Management Unit (AGMU).¹²⁹

DHS:

1. HSAC

2. Principal Deputy General Counsel David Martin, Deputy General Counsel Audrey Anderson, and the two custodians whose names are redacted in Martin’s September 3 email. The records of other OGC custodians may need to be searched depending on whether these four custodians’ records reveal that it is reasonable to believe that others had responsive records. Ivan Fong’s records should be searched unless he submits a sworn affidavit stating that he was not involved in any opt-out discussions prior to October 15, 2010.

3. Three US-VISIT custodians who already searched their records but did not provide the court with the

search terms that they used,¹³⁰ selected by plaintiffs.

ICE:

1. Five HSI employees, selected by plaintiffs.

2. The OSLTC Chief of Staff and the Chief Public Engagement Officer who engaged in correspondence regarding opt-out on September 21, 2010.

3. Of the approximately 106 Field Coordinators in ICE’s Enforcement and Removal Operations who conducted searches but did not provide details about which search terms were actually used, plaintiffs may select twenty custodians to conduct new, fully-documented searches.

4. One custodian from the ICE Office of the Director, chosen by defendants, who conducted a search using the eight search terms recommend by the FOIA office.

5. Two custodians from the ICE Office of the Executive Secretariat, chosen by plaintiffs, about whose search defendants gave no information.

6. In addition, ICE is instructed to inform plaintiffs about the extent to which contractors were included in the searches, and, if plaintiffs are dissatisfied, submit to the Court a supplemental declaration.

7. Finally, ICE is instructed to answer the questions about archive recovery systems and SharePoint instructions (but not the other two questions) on page twenty-nine of plaintiffs’ memorandum of law.

IV. CONCLUSION

This litigation, filed more than two years ago, has already engendered four judicial opinions — now five. I once again urge the Government to heed the now famous words of Justice Louis Brandéis with which I began this opinion. For the reasons stated above, the motions of OLC and EOIR are granted. The motions of ICE, the FBI, DHS, and plaintiffs are granted in part and denied in part. The Clerk of the Court is directed to close these motions [Docket Nos. 177 and 186], The parties are instructed to meet and confer and then to submit letters to the Court proposing a timeline for the effectuation of this decision. If the parties prefer, they may call Chambers to schedule a conference.

SO ORDERED.

1 . Buckley v. Valeo, 424 U.S. 1, 67, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976) (quoting Louis Brandeis, Other People's Money and How the Bankers Use It 62 (1933)).

2 . See www.ndlon.org/en/secure-communities. The documents produced in response to this FOIA request have been compiled at a website entitled Uncover the Truth: ICE and Police Collaborations. See http://uncover thetruth.org.

3 . See Def. Mem. at 10.

4 . 5 U.S.C. § 552 et seq.

5 . See Secure Communities, ICE, http://www. ice.gov/secure_communities/.

6 . See id.

7 . The cut-off date for the search of opt-out records was October 15, 2010. That is to say, records created after that date did not need to be produced as part of the January 17, 2011 production and are therefore not the subject of this motion.

8 . Memorandum of Law in Support of Defendants’" Renewed Motion for Partial Summary Judgment on Adequacy of Search for Opt-Out and Rapid Production List Records ("Def. Mem.”) at 1.

9 . See Declaration of Ryan Law ("Law Decl.”) and Supplemental Declaration of Ryan Law ("Supp. Law Decl.”) (ICE); Seventh Declaration of David Hardy ("Hardy Decl.”) and Declaration of Dennis Argali ("Argali Decl.”) (FBI); Declaration of David Palmer (“Palmer Decl.”) (DHS); Declaration of Crystal Rene Souza ("Souza Decl.") (EOIR); and Declaration of Paul Colborn (“Colborn Decl.”) (OLC).

10 . See Memorandum of Law in Opposition to Defendants’ Motion for Partial Summary Judgment and in Support of Plaintiffs’ Cross-Motion for Partial Summary Judgment ("Pl. Mem.”) and Memorandum of Law in Further Support of Plaintiffs’ Cross-Motion for Partial Summary Judgment on Adequacy of Defendants' Search for Opt-Out and Rapid Production List Records ("Pl. Rep. Mem.”). To support their arguments, plaintiffs have submitted seventy-seven documents that were produced by defendants during this litigation. They are indexed and attached to the Declaration of Sunita Patel ("Patel Decl.”) and are referred to here as Patel Doc. x.

11 . Defendants initially filed this motion, with supporting declarations, on January 12, 2012. See Docket No. 167. Plaintiffs objected to the declarations, arguing that they contained an insufficient level of detail and that the parties’ briefing would therefore focus on the description of the searches rather than the adequacy of the searches. I instructed defendants to refile their brief, along with more detailed declarations, which they did on March 2, 2012.

12 . The EOIR's motion for partial summary judgment was unopposed by plaintiffs.

13 . Morley v. Central Intel. Agency, 508 F.3d 1108, 1114 (D.C.Cir.2007).

14 . Id. (quotation and citation omitted).

15 . See Bloomberg L.P. v. Board of Governors of the Fed. Reserve Sys., 649 F.Supp.2d 262, 271 (S.D.N.Y.2009).

16 . Fed.R.Civ.P. 56(c).

17 . Roe v. City of Waterbury, 542 F.3d 31, 35 (2d Cir.2008) (quoting Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986)).

18 . McClellan v. Smith, 439 F.3d 137, 144 (2d Cir.2006).

19 . Carney v. United States Dep't of lustice, 19 F.3d 807, 812 (2d Cir.1994) (quotations and citations omitted).

20 . Exxon Corp. v. Federal Trade Comm’n, 466 F.Supp. 1088, 1094 (D.D.C.1978).

21 . Carney, 19 F.3d at 812.

22 . Families for Freedom v. Customs and Border Protection, 837 F.Supp.2d 331, 336 (S.D.N.Y.2011).

23 . Katzman v. Central Intel. Agency, 903 F.Supp. 434, 438 (E.D.N.Y.1995) (quoting Church of Scientology v. Internal Revenue Serv., 792 F.2d 146, 151 (D.C.Cir.1986)).

24 . See Banks v. United States Dep’t of Justice, 700 F.Supp.2d 9, 15 (D.D.C.2010).

25 . Iturralde v. Comptroller of Currency, 315 F.3d 311, 313-14 (D.C.Cir.2003).

26 . Id. at 315.

27 . Weisberg v. United States Dep’t of Justice, 705 F.2d 1344, 1351 (D.C.Cir.1983).

28 . See Law Supp. Decl. This declaration was submitted after the close of briefing. In their 4/30/12 Letter to the Court (which was submitted before the declaration but after the parties had communicated about the supplemental search), plaintiffs’ criticism centered on the failure to search records beyond email, the failure to use the search term "interoperability,” and the failure to more fully describe the Deputy Director’s file storage system. The first of these concerns was addressed by Law's declaration, which states that the agency searched for "additional electronic or hard-copy files,” id. ¶ 19, and the second two concerns are, given the totality of the agency's search, relatively minor.

29 . See Law Decl. ¶ 24 n. 3. The parties agreed to a special process for addressing records in the possession of HSI's Law Enforcement Service Center that relate to individual aliens.

30 . Id.

31 . 9/10/10 Email, Patel Doc. 24.

32 . See Undated Memorandum of Understanding Between DHS, ICE, and the Colorado Department of Public Safety, Patel Doc. 38, at 7.

33 . 5/3/11 Email, Patel Doc. 60, at 1.

34 . See Memorandum of Law in Opposition to Plaintiffs’ Cross-Motion for Partial Summary Judgment and in Further Support of Defendants’ Motion for Partial Summary Judgment on Adequacy of Search for Opt-Out and Rapid Production List Records (“Def. Rep. Mem.”) at 8.

35 . PL Mem. at 22. See 4/15/11 Email, Patel Doc. 8, at 3 (conveying ICE's comment to the New York Times that "ICE retained a contractor to assist in the implementation of Secure Communities”) and 4/12/11 Letter from Dan Cadman, ICE contractor, to Marc Rapp, Acting Assistant Director for Secure Communities ("Cadman Letter”), available at http:// big.assets.huffingtonpost.com/LofgrenFollow Up.pdf (describing author’s role in developing and implementing opt-out policy).

36 . Def. Rep. Mem. at 8.

37 . Cadman Letter at 2 n. 2.

38 . See Law Decl. ¶ 49 (stating that only the Deputy Assistant Director and a Senior Public Engagement Officer conducted searches).

39 . PL Mem. at 22. See 9/27/10 Email from Richard Rocha, Patel Doc. 50 (proposing a conversation to discuss how "OSLTC can help make [ICE’s] position on participation [in Secure Communities] clear”); 9/9/10 Recommendations for Outreach on Required Activation of Secure Communities, Patel Doc. 53 (including OSLTC in the list of offices responsible for clarifying opt-out questions); 1/26/11 Email Chain, Ex. B to Declaration of Sunita Patel in support of PL Rep. Mem. ("Patel Rep. Decl.”) (discussing a meeting in Boston regarding opt-out set up by OSLTC).

40 . 1/21/11 Email, Ex. A to Patel Rep. Decl. In addition, Patel Doc. 37 contains the September 21, 2010 correspondence of OSLTC's Chief Public Engagement Officer regarding opt-out. According to ICE's declaration, this officer did not search her/his records.

41 . Def. Reply Mem. at 8-9.

42 . See Oglesby v. United States Dep't of Army, 920 F.2d 57, 68 (D.C.Cir.1990) (although one location may be the "most likely” to turn up records, “the agency cannot limit its search to only one record system if there are others that are likely to turn up the information requested”).

43 . Plaintiffs have not presented evidence showing that the Assistant Director for OSLTC was involved in discussions related to opt-out. (The meetings including the Assistant Director that were cited in the 1/21/11 Email did not relate to opt-out).

44 . Law Decl. ¶ 55.

45 . PL Mem. at 23.

46 . Under Secure Communities, fingerprints sent by state and local law enforcement to the FBI's IAFIS database are automatically forwarded to DHS’s IDENT database for an immigration check. The central questions regarding “opt-out” was whether state and local law enforcement would be permitted to use the FBI database without having the information that they input forwarded to DHS and, if the answer to that question was no, whether the local agencies could chose not to receive the resulting immigration information back from DHS.

47 . 12/6/06 Testimony to the Senate Judiciary Committee, Patel Doc. 64. In 2008, FBI Deputy Director John Pistole described the Bureau's involvement in Secure Communities as an example of federal-local collaboration that helps ICE achieve its goals. See 3/28/08 ICE News Release, Patel Doc. 62.

48 . 3/26/10 Email from Secure Communities Branch Chief for Deployment to Redacted Recipients, Patel Doc. 27.

49 . 7/8/10 DHS Weekly Report, Patel Doc. 39, at 1528. The document said that Napolitano and Mueller were scheduled to meet on July 21. Defendants do not suggest the meeting never occurred. See Def. Rep. Mem. at 13.

50 . See 7/30/10 Email Chain, Patel Doc. 28 (congressional inquiry circulated by the Director's Office to the FBI Chief of Staff, Deputy Director, Associate Deputy Director, and Office of General Counsel); 9/16/10 Email Chain, Patel Doc. 29 (Secretary Napolitano's response to congressional inquiry).

51 . Secure Communities Internal Meeting Minutes, Patel Doc. 5 at 1893.

52 . Within the Director’s Office, only the Office of Public Affairs and the Office of Congressional Affairs conducted Opt-Out and RPL searches. See Hardy Decl. ¶¶ 16, 17, 26.

53 . Argali Deck ¶ 6.

54 . Def. Rep. Mem. at 12. These records, like many others, were produced on a less rapid basis during 2011 and 2012.

55 . See 11/18/10 Email Chain, Patel Doc. 17 (FBI staffers circulating drafts of a memo to the Director on opt-out). See also Patel Docs 19, 30, 32 (documents showing involvement by Director in opt-out after October 15, 2010).

56 . See Def. Opp. Mem. at 13-14; Pi. Rep. Mem. at 6-7.

57 . Hardy Decl. II 17.

58 . Carney, 19 F.3d at 812.

59 . In a previous declaration, Hardy swore that OGC "reported having no documents responsive to plaintiffs' request.” 11/12/10 Declaration of David M. Hardy in Support of Defendants’ Opposition to Plaintiffs' Motion for Preliminary Injunction [Docket No. 15 Ex. Bl]. The truth, according to Hardy’s most recent declaration, is that OGC never reported back at all.

60 . Following the parties’ lead, I exclude from this discussion the OGC representative from the Access Integrity Unit, who did conduct a search.

61 . Def. Rep. Mem. at 13.

62 . See Policy Initiation and Coordination Memo, Patel Doc. 10.

63 . See, e.g., 12/1/10 Email, Patel Doc. 18.

64 . See Pl. Mem. at 14-15 and Pl. Rep. Mem. at 7.

65 . See Patel Docs. 63, 65 (describing the mission of the OLEC) and Patel Doc. 20 (5/11/11 email from Executive Assistant Director saying that he has "a pretty good feel for [the opt-out issue] and ha[s] been tracking it for a while”).

66 . See Johnson v. Executive Office for U.S. Attorneys, 310 F.3d 771, 776 (D.C.Cir.2002) (noting that compliance with FOIA requires "both systemic and case-specific exercises of discretion and administrative judgment and expertise”).

67 . Argali Decl. ¶ 7.

68 . Hardy Decl. ¶¶ 17-18.

69 . Id. ¶ 20.

70 . Id.

71 . Morley, 508 F.3d at 1114.

72 . Palmer Decl. ¶ 30.

73 . Id.

74 . PL Rep. Mem. at 8.

75 . See Palmer Decl. ¶ 24.

76 . See 9/6/10 Email Thread, Patel Doc. 23.

77 . See National Day Laborer Org. Network v. United States Immigration and Customs Enforcement Agency (“NDLON”), 827 F.Supp.2d 242, 258-59 (S.D.N.Y.2011).

78 . 9/6/10 Email Thread, Patel Doc. 23.

79 . Agencies have an obligation to "follow through on obvious leads to discover requested documents” Valencia-Lucena v. United States Coast Guard, 180 F.3d 321, 325 (D.C.Cir.1999) and "the court evaluates the reasonableness of an agency’s search based on what the agency knew at its conclusion rather than what the agency speculated at its inception” Campbell v. United States Dep’t of Justice, 164 F.3d 20, 28 (D.C.Cir.1998). According to emails on 4/27/11, the Office of General Counsel had submitted comments regarding the October 2 Memorandum in October, 2010. See Patel Doc. 45. Because of redactions, it is unclear to the Court whether the files of the woman referenced in the first paragraph of the 4:26 p.m. email were searched, but they should have been. It also appears likely that Deputy GC Audrey Anderson was involved in discussions about the October 2 Memorandum prior to the October 15, 2010 cut-off date. Thus, a reasonable search would also have examined her records.

80 . See Patel Docs. 45-46.

81 . See Palmer Decl. ¶¶ 6, 17 (listing the offices that DHS determined were likely to maintain responsive records but not asserting that the list was exhaustive and not asserting that DHS made any determination regarding HSAC).

82 . See 9/30/09 HSAC Meeting Minutes, Patel Doc. 69.

83 . Palmer Deck ¶ 23.

84 . Pl. Mem. at 18.

85 . Id. at 1.

86 . Because OLC is so much smaller than ICE, the FBI, and DHS, I address both its choice of custodians and its method of search here.

87 . Colborn Decl. ¶ 3.

88 . OLC’s search of departed attorneys’ emails — which included only the terms "secure communities” and "interoperability”— should have used additional terms (such as, e.g., opt-out, opt out, ICE, IDENT, and IAF-IS). If the entirety of OLC’s search had consisted of searches using only those two terms, it would have been inadequate. However, in combination with the additional steps that the office took, its limited search of departed attorneys’ emails does not render the entire search inadequate.

89 . Plaintiffs also argue that the agencies did not sufficiently describe the geography of their file storage systems. This may be true. In general, however, the important shortcomings in defendants' declarations involve their inadequate description of searches. A fuller description of searches would make less consequential any gaps in defendants' description of their file storage systems.

90 . Law Deck ¶ 20.

91 . Id. ¶ 36. See How-To-Search for Opt-Out Records, Ex. B to Law Deck The recommended search terms were: "opt-out,” “voluntary,” "opting-out,” “mandate,” "mandatory,” "participation,” “choosing,” and "opt out.” The instructions said, in bold letters, "however, please do not be limited by the suggested terms if you believe that you may have responsive documents that can be located with other search terms.” Id.

92 . See Law Deck ¶¶ 44-47 (describing the searches of three components within the OPLA and noting that a fourth component (Homeland Security Investigations) conducted a search using only the terms "opt-out” and "opt out” because "other terms were likely to produce large numbers of unresponsive documents unrelated to Secure Communities”); ¶ 49 (two staff members in the OSLTC conducted searches using the terms "opt-out,” "voluntaiy,” and "mandatory”); ¶ 51 (Office of Public Affairs searched using the eight recommended terms plus "Secure Communities”); ¶ 52 (Office of the Director used the eight recommended terms).

93 . See id. ¶ 37-43.

94 . See id. ¶ 50 (no indication of which search terms were used by the Office of Congressional Relations).

95 . The FBI’s FOIA officer explains that although his office "did not provide specific instruction as to how the individual searches [of paper files, e-mail, and personal drives] were to be performed, each employee conducted a manual review of his or her records to locate potentially responsive records.” Hardy Decl. ¶ 19. Although Hardy does not define his use of the phrase "manual review,” defense counsel construe Hardy's statement to mean "a search involving review of every individual document for potential responsiveness.” Def. Mem. at 16. "The FBI largely did not rely on search terms, but instead relied on the knowledge of its custodians” who "sift[ed] through and review[ed] tens of thousands of pages of records.” Def. Rep. Mem. at 15. If in fact the IIU employees who conducted this search looked through every one of their email messages and other files, then they will have conducted an adequate search. Hardy, however, does not say that; rather, that is defense counsels’ construction of the phrase "manual review.” It seems likely that some IIU custodians only reviewed certain categories of documents (i.e., emails from certain people or documents in certain folders) or narrowed the number of emails that they examined by first using search terms. In any event, the FBI's declarant should have provided some specificity about what this manual search entailed. Such a description need not "set forth with meticulous documentation the details of an epic search,” Perry v. Block, 684 F.2d 121, 127 (D.C.Cir.1982), but it does need to contain "reasonable specificity of detail rather than merely conclusory statements,” Grand Cent. P’ship v. Cuomo, 166 F.3d 473, 478 (2d Cir.1999).

96 . See Hardy Decl. ¶ 16.

97 . See id.

98 . Palmer Decl. ¶ 20. Compare id. ¶ 25 ("exact manner in which searches were conducted was left to the discretion of the individual custodians” in US-VISIT office and their search terms were not reported to the Court (or, presumably, the FOIA office)) with id. ¶ 32 (custodians in the Office of the Secretary used the eight recommended terms).

99 . See Colborn Decl. ¶¶ 6, 8 and Ex. A (listing search terms).

100 . See Souza Decl. 114.

101 . Families for Freedom v. United States Customs and Border Protection, 837 F.Supp.2d 331, 335 (S.D.N.Y.2011).

102 . Id.

103 . Given the tedious nature of the assignment of examining every single document and the difficult and subjective nature of deciding what is and is not responsive, it would have been wise of the FBI to run a few verification tests using sophisticated search techniques to ensure that the manual review was actually capturing the universe of responsive documents. Such tests would have given the Court significantly more confidence regarding the adequacy of these manual reviews. See Maura R. Grossman & Gordon V. Cormack, Technology-Assisted Review in E-Discovery Can Be More Effective and More Efficient Than Exhaustive Manual Review, XVII Rich. J.L. & Tech. 11 (2011). See generally the Text Retrieval Conference (TREC) Legal Track at http://trec-legal.umiacs.umd.edu/.

104 . The FBI ran a search for "secure communities” in one of its central databases, which did not return any responsive records. See Hardy Decl. ¶ 14. The FBI has not submitted information about the search terms used in the many other searches that employees conducted of their individual and shared records. After I ordered defendants to produce opt-out and RPL documents on an expedited basis, the FBI searched for the terms "opt-out” and "opt out” in the documents that it had already collected in response to plaintiffs’ complete FOIA request. See id. ¶ 25. But I do not know what search terms it used to compile those documents in the first place.

105 . Def. Rep. Mem. at 4 (citing Perry, 684 F.2d at 127).

106 . Id. at 10.

107 . Grand Cent. P’ship, 166 F.3d at 478.

108 . Iturralde, 315 F.3d at 313 (quoting Ogles-by, 920 F.2d at 68). See also Def. Rep. Mem. at 4 (reiterating this requirement).

109 . See Pension Comm. of the Univ. of Montreal Pension Plan v. Banc of Am. Sec., LLC, 685 F.Supp.2d 456, 473 (S.D.N.Y.2010) ("I note that not every employee will require hands-on supervision from an attorney. However, attorney oversight of the process, including the ability to review, sample, or spot-check the collection efforts is important. The adequacy of each search must be evaluated on a case by case basis.”).

110 . Defense counsel in this litigation have noted, correctly, that search obligations under FOIA are not identical to those under the Federal Rules of Civil Procedure. But under FOIA, to prevail on an adequacy of the search motion, a defendant "must show beyond material doubt that it has conducted a search reasonably calculated to uncover all relevant documents.” Morley, 508 F.3d at 1114. Defendants are of course correct to note that " 'the adequacy of a FOIA search is generally determined not by the fruits of the search, but by the appropriateness of the methods used to carry out the search' because 'particular documents may have been accidentally lost or destroyed, or a reasonable and thorough search may have missed them.'" Def. Rep. Mem. at 3 (quoting Iturralde, 315 F.3d at 315). The failure to actually uncover all documents is not fatal to the adequacy of the government's search; but the failure to design a search that is reasonably calculated to uncover all documents is fatal. Therefore, much of the logic behind the increasingly well-developed caselaw on e-discovery searches is instructive in the FOIA search context because it educates litigants and courts about the types of searches that are or are not likely to uncover all responsive documents.

111 . Shira A. Scheindlin, Daniel J. Capra, & The Sedona Conference, Electronic Discovery and Digital Evidence: Cases and Materials at 327 (2d ed.2012). Accord Victor Stanley, Inc. v. Creative Pipe, Inc., 250 F.R.D. 251, 257 (D.Md.2008) ("there is a growing body of literature that highlights the risks associated with conducting an unreliable or inadequate keyword search or relying exclusively on such searches”).

112 . Maura R. Grossman & Terry Sweeney, What Lawyers Need to Know About Search Tools: The Alternatives to Keyword Searching Include Linguistic and Mathematical Models for Concept Searching, Nat. L.J. (Aug. 23, 2010) ("What Lawyers Need to Know") (citing three studies showing that Boolean keyword search identifies between twenty and twenty-five percent of relevant documents).

113 . Moore v. Publicis Groupe & MSL Group, No. 11 Civ. 1279, 287 F.R.D. 182, 190-91, 2012 WL 607412, at *10 (S.D.N.Y. Feb. 24, 2012) (quoting Ralph C. Losey, Child’s Game of 'Go Fish’ is a Poor Model for e-Discovery Search, in Adventures in Electronic Discovery 209-10 (2011) and citing the following cases criticizing keyword search: United States v. O’Keefe, 537 F.Supp.2d 14, 24 (D.D.C.2008); Equity Analytics, LLC v. Lundin, 248 F.R.D. 331, 333 (D.D.C.2008); Victor Stanley, Inc., 250 F.R.D. at 262; William A. Gross Constr. Assocs., Inc. v. American Mfrs. Mut. Ins. Co., 256 F.R.D. 134, 134, 136 (S.D.N.Y.2009)).

114 . Prominent among these are the publications of the Sedona Conference. See, e.g., Sedona’s Commentary on Achieving Quality in the E-Discovery Process (2009) and Commentary on Search & Retrieval Methods (2007), available at www.thesedonaconference.org/ publications.

115 . William A. Gross, 256 F.R.D. at 136. See Victor Stanley, Inc., 250 F.R.D. at 262 ("Selection of the appropriate search and information retrieval technique requires careful advance planning by persons qualified to design effective search methodology. The implementation of the methodology selected should be tested for quality assurance; and the party selecting the methodology must be prepared to explain the rationale for the method chosen to the court, demonstrate that it is appropriate for the task, and show that it was properly implemented. In this regard, compliance with the Sedona Conference Best Practices for use of search and information retrieval will go a long way towards convincing the court that the method chosen was reasonable and reliable.”).

116 . See What Lawyers Need to Know (excerpted in Electronic Discovery and Digital Evidence at 328-29).

117 . See, e.g., Fox News Network, LLC v. United States Dep’t of the Treasury, 678 F.Supp.2d 162, 166 (S.D.N.Y.2009) (finding that failure to use an obvious acronym made the search inadequate); Hasbrouck v. United States Customs and Border Protection, No. 10 Civ. 3793, 2012 WL 177563 (N.D.Cal. Jan. 23, 2012) (finding failure to search spelling variants as requested was inadequate); Habeas Cotpus Resource Ctr. v. United States Dep’t of Justice, No. 08 Civ. 2649, 2008 WL 5111224 (N.D.Cal. Dec. 2, 2008) (finding declarations that specified two or three search terms were insufficient to establish adequacy).

118 . See Media Research Center v. United States Dep't of Justice, 818 F.Supp.2d 131 (D.D.C.2011) (defendant was not required to use certain search terms); Vest v. Department of Air Force, 793 F.Supp.2d 103 (D.D.C.2011) (search for spelling variants was not required); Anderson v. United States Dep’t of State, 661 F.Supp.2d 6 (D.D.C.2009) (same).

119 . See Amnesty Int'l v. Central Intelligence Agency, No. 07 Civ. 5435, 2008 WL 2519908, at *15 (S.D.N.Y. June 19, 2008) (“a search that is designed to return documents containing the phrase 'CIA detainees’ but not 'CIA detainee’ or 'detainee of the CIA’ is not reasonably calculated to uncover all relevant documents .... Nevertheless, that failure does not render the Government’s searches unreasonable in this case [because] the vast majority of [document databases] were searched by hand ... [and] such hand searches are not tainted by any deficiencies in the Government’s computerized searches.”) (some quotations omitted).

120 . Declaration of Daniel L. Regard II in Support of Plaintiffs' Cross-Motion for Summary Judgment and in Opposition to Defendants’ Motion for Summary Judgment on Adequacy of Search ("Regard Decl.”) ¶ 28.

121 . Id. ¶ 42.

122 . What Lawyers Need to Know (excerpted in Electronic Discovery and Digital Evidence at 328). Plaintiffs have urged defendants to conduct searches using multiple variants of the terms "opt out” and "secure communities.” See Law Supp. Dec. ¶ 9 (listing plaintiffs’ recommended terms as "opt-in,” "opt in,” "opt-out,” "opt out,” "opting out,” "opting in” and "secure communities,” "secure -communities,” "scomm,” "s-comm,” “sc,” "circa,” and "scmpo”).

123 . O’Keefe, 537 F.Supp.2d at 24.

124 . This does not apply to OLC. Given the size of that office and the various approaches that it took to searching for records, I am confident that the broad search using the terms "ICE” and "Secure Communities” was reasonably designed to uncover all responsive documents.

125 . Defendants were given three opportunities on this motion to submit detailed declarations (the initial January declarations that were insufficiently detailed, the March moving declarations, and the reply declarations). With a few exceptions noted below, defendants will now need to rectify their failure to establish adequacy by conducting new searches rather than only by submitting new declarations.

126 . Plaintiffs ask the Court to order the FBI to explain “how searches of electronic records were conducted by individual custodians and what search terms were used.” Pl. Mem. at 28. I do not know if the FBI ever collected that information, although it should have. Revisiting the past searches in that level of detail is less efficient or important than conducting thorough and well-documented searches of specific custodians going forward, and therefore it is not necessary.

127 . In either event, however, the Privacy and Civil Liberties Unit is required to conduct a search for documents responsive to RPL VII.

128 . A search of these three test employees' records using a negotiated search methodology will permit plaintiffs and the Court to evaluate whether the earlier manual review was in fact adequate.

129 . Plaintiffs believe that the AGMU may have documents relating to a key recommendation made by the Advisory Policy Board (which is technically not comprised of FBI employees) to the FBI Director, which he accepted, to make information sharing between the FBI and DHS mandatory. Defendants searched the AGMU and produced responsive records, but have provided no details about the nature or scope of that search. See Hardy Deck ¶ 17 n. 6. A renewed, targeted search for responsive records (including for opt-out records from February through October, 2010) is therefore appropriate.

130 . Compare Palmer Deck ¶ 25 (stating that specific search terms were “recommended” to US-VISIT custodians) with ¶ 32 (stating that specific search terms were "used” by Office of the Secretary custodians).