Long v. Immigration & Customs Enforcement

• Court: District Court, District of Columbia
• Decided: March 30, 2026
• Docket: Civil Action No. 2022-2655
• Status: Published

Opinion

UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA

_________________________________________ ) SUSAN B. LONG, et al. ) ) Plaintiffs, ) ) v. ) Case No. 22-cv-02655 (APM) ) IMMIGRATION AND CUSTOMS ) ENFORCEMENT, ) ) Defendant. ) _________________________________________ )

MEMORANDUM OPINION & ORDER

I.

Before the court in this Freedom of Information Act (FOIA) case are the parties’ cross-

motions for summary judgment regarding the adequacy of Defendant Immigration and Customs

Enforcement (ICE)’s search. See Def.’s Mot. for Summ. J., ECF No. 34 [hereinafter Def.’s Mot.];

Pls.’ Cross-Mot. for Summ. J., ECF No. 38 [hereinafter Pls.’ Mot.]. For the reasons that follow,

the court denies both motions without prejudice.

II.

This case arises out of related proceedings in Long v. ICE (Long I), No. 14-cv-109-APM

(D.D.C.) [hereinafter Long I Docket]. In 2010, Plaintiffs requested “a complete set of

documentation” on two of Defendant’s databases: the Enforcement Integrated Database (“EID”)

and Integrated Decision Support Database (“IIDS”). Long v. ICE, No. 14-cv-109, 2025 WL

3459086, at *1–2 (D.D.C. Dec. 2, 2025) (internal quotation marks omitted). The former contains

information about Defendant’s law enforcement operations. Id. at *1. The latter “is used primarily

for reporting purposes” and contains a subset of the information found in the EID. Id. at *2. For each database, Plaintiffs sought records of both the “Data Dictionary”—which “lists the names of

the tables and fields contained in each database”—and code lookup tables—which contain plain-

English translations of some of the databases’ codes. Id. at *4, *7 (internal quotation marks

omitted).

As Long I carried on and the court ordered Defendant to produce additional responsive

documents, the parties reached an impasse as to whether the court’s orders required Defendant to

produce documents from current versions of the databases. See Joint Status Report, Long I Docket,

ECF No. 96, at 2–3. Instead of litigating that issue in Long I, Plaintiffs submitted renewed FOIA

requests and filed this suit. The FOIA requests sought “all descriptive information—including the

name and description of each table and field of information along with the definition of all codes

used—contained in the current version of” both the EID and IIDS. Compl., ECF No. 1 [hereinafter

Compl.], ¶ 8. Plaintiffs specified that “current version” means “as of the date of this request, or

the date of the search if it is later.” Id.

Upon receiving Plaintiffs’ request, Defendant extracted the data dictionaries for each

database from its “online data dictionary repository know[n] as Matrix.” Decl. of Patrick Lewis,

ECF No. 34-4 [hereinafter Lewis Decl.], ¶ 7. The data dictionaries in this repository are initially

populated from each database’s source code. Id. ¶ 9. Defendant then adds information like plain-

English translations, reformats the data, and uploads the newly generated data dictionaries to the

Matrix repository. See Second Decl. of Patrick Lewis, ECF No. 42-2 [hereinafter Second Lewis

Decl.], ¶¶ 5–6. The Matrix dictionaries are periodically updated “as the databases evolve to meet

changing ICE operating conditions and needs.” Lewis Decl. ¶ 9.

For the code lookup tables, Defendant first looked to the IIDS data dictionaries to identify

any such tables. See Decl. of Timothy Gibney, ECF No. 34-3 [hereinafter Gibney Decl.], ¶ 8.

2 Defendant then used the resulting list to extract the code lookup tables directly from the databases.

See id. ¶ 9; Lewis Decl. ¶ 10. Defendant provided responsive information to Plaintiffs over four

separate releases. Pls.’ Mot., Decl. of Susan B. Long, ECF No. 38-1 [hereinafter Long Decl.], ¶ 7.

After reviewing Defendant’s productions, Plaintiffs informed Defendant that they believed

the productions were incomplete. See Lewis Decl. ¶ 11; Long Decl. ¶ 7. In response, Defendant

performed a supplemental search and made five additional releases to Plaintiffs. Lewis Decl. ¶ 11;

Long Decl. ¶ 8. Still, some information has not been provided to Plaintiffs because, although

present in the EID at the time of the searches, it was not present in the Matrix dictionaries.

See Lewis Decl. ¶ 11c.iii. Thirteen tables were not produced because they “were added to Matrix

as part of the annual data dictionary update cycle which was performed after the Supplemental

Search.” Id.; Def.’s Combined Reply in Further Supp. of Def.’s Mot. & Mem. in Opp’n to Pls.’

Mot., ECF No. 42 [hereinafter Def.’s Reply], Def.’s Resp. to Pls.’ Stmt. of Additional Undisputed

Material Facts, ECF No. 42-1 [hereinafter Def.’s Resp. Stmt.], ¶ 38. Other tables were not

produced because they did not meet the criteria for addition to the data dictionaries to begin with,

so they are “currently outside the scope of the Matrix data dictionary repository.” Lewis Decl.

¶ 11c.iv; Second Lewis Decl. ¶ 10. On this basis and others, Plaintiffs now challenge the adequacy

of Defendant’s search. See Pls.’ Mot. at 6–11.

III.

A court “shall grant summary judgment if the movant shows that there is no genuine

dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed. R.

Civ. P. 56(a). A “genuine” dispute is one in which a rational trier of fact could find for the

nonmoving party based on the record before it. Scott v. Harris, 550 U.S. 372, 380 (2007) (quoting

Matsushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 586–87 (1986)).

3 An agency meets its burden to show a search is adequate if it demonstrates it “made a good

faith effort to conduct a search for the requested records, using methods which can be reasonably

expected to produce the information requested.” Watkins L. & Advoc., PLLC v. U.S. Dep’t of Just.,

78 F.4th 436, 442 (D.C. Cir. 2023) (internal quotation marks and citation omitted). The search

need only be “reasonably calculated to uncover all relevant documents”; the “agency need not

search every record system or demonstrate that all responsive documents were found and that no

other relevant documents could possibly exist.” Id. (internal quotation marks and citations

omitted). The court may determine the sufficiency of the agency’s efforts from a “reasonably

detailed affidavit, setting forth the search terms and the type of search performed, and averring

that all files likely to contain responsive materials (if such records exist) were searched.” Valencia-

Lucena v. U.S. Coast Guard, 180 F.3d 321, 326 (D.C. Cir. 1999) (internal quotation marks and

citation omitted).

IV.

The parties’ primary dispute is whether the agency’s searching and then providing the data

dictionaries from Matrix—rather than the databases themselves—renders the search inadequate.

See Pls.’ Mot. at 9; Def.’s Reply at 1–3. The court concludes it does not.

Through declarations, Defendant has demonstrated that its search of the Matrix repository

was “reasonably calculated to uncover all relevant documents.” Watkins, 78 F.4th at 442 (internal

quotation marks omitted). Matrix is “the current authoritative place for technical documents, such

as data dictionaries, associated with the EID and IIDS.” Second Lewis Decl. ¶ 7; see also Lewis

Decl. ¶ 7. That is because the data dictionaries as they exist in Matrix do not exist in the databases

themselves. See Second Lewis Decl. ¶ 6. Although the Matrix dictionaries are initially populated

from each database’s source code, Lewis Decl. ¶ 9, that direct query only “yield[s] table names,

4 column names, and column format information without providing other necessary information

such as definitions or the plain-English names by which non-technical staff refer to each column

and table,” Second Lewis Decl. ¶ 5. Defendant then matches this information with plain-English

descriptions of the codes, reformats the data, and loads the resulting tables into the Matrix

repository. Id. ¶¶ 5–6. And the tables are “stored only in the Matrix repository.” Id. ¶ 6. As a

result, “ICE produces all of its data dictionaries from the Matrix repository” when responding to

FOIA requests. Id. ¶¶ 6, 9.

Contrary to Plaintiffs’ protestations, Defendant cannot simply query the databases

themselves for the data dictionaries as Plaintiffs requested them. Second Lewis Decl. ¶ 9; see Am.

Oversight v. U.S. Dep’t of Health & Hum. Servs., 101 F.4th 909, 923 (D.C. Cir. 2024) (noting that

“reasonably detailed, nonconclusory affidavits” are “accorded a presumption of good faith” unless

“rebutted by positive indications of overlooked materials” (internal quotation marks and citations

omitted)). At the least, any production directly from the EID or IIDS would not have contained

“descriptive information” like the “description of each table and field of information along with

the definition of all codes used.” Compl. ¶ 8; see also Second Lewis Decl. ¶ 6 (noting that the

information as it exists in the databases “on its own does not constitute a complete data dictionary

for either database”). Defendant’s decision to search Matrix—the “authoritative place” for the

requested records—in response to Plaintiffs’ request was therefore reasonable.

However, Plaintiffs respond that, “[e]ven within Matrix, ICE’s search was inadequate.”

Pls.’ Reply in Supp. of Pls.’ Mot., ECF No. 45 [hereinafter Pls.’ Reply], at 5. They point to

multiple categories of records that were missing from Matrix at the time of Defendant’s searches,

which they reason shows that the search was not likely to uncover all responsive records. See id.

at 5–7; Pls.’ Mot. at 7–11. The court disagrees. To be sure, as Defendant concedes, the Matrix

5 dictionaries are not always perfect reflections of the data as it exists in their respective databases.

“Minor inaccuracies are inevitable due to the numerous manual hands-on steps required to produce

them.” Lewis Decl. ¶ 8. Moreover, the Matrix dictionaries may not fully capture the contents of

the databases at a given time because there may be new additions that have yet to be reflected in

an annual Matrix update. Id. ¶ 11c.iii. But “a search need not be perfect, only adequate, and

adequacy is measured by the reasonableness of the effort in light of the specific request.” Machado

Amadis v. Dep’t of Just., 388 F. Supp. 3d 1, 14 (D.D.C. 2019) (quoting Meeropol v. Meese,

790 F.2d 942, 956 (D.C. Cir. 1986)). The agency “searched the computer database that most likely

would identify responsive records”—here, the only place that contained the data dictionaries as

requested. Id. (internal quotation marks omitted). “Nothing more is required.” Id. (internal

quotation marks omitted). That the resulting production did not capture some information because

of how the databases are maintained and the information extracted does not render the search

inadequate. See Saldana v. Fed. Bureau of Prisons, 715 F. Supp. 2d 10, 20 (D.D.C. 2010) (“[T]he

adequacy of a search is not determined by its results, but by the method of the search itself.” (citing

Weisberg v. U.S. Dep’t of Just., 745 F.2d 1476, 1485 (D.C. Cir. 1984))).1

The court is also largely unpersuaded by Plaintiffs’ arguments regarding “discrepancies

between the materials produced in the 2014 case and the code tables and data dictionaries produced

in this case.”2 Pls.’ Mot. at 10. The court agrees with Defendant that, because the databases are

ever-changing, comparisons to prior productions are not an optimal way to determine adequacy.

1 Defendant also acknowledges that some tables may not meet the criteria “for adding schemas to [ICE’s] data dictionaries” and, for that reason, be left out of Matrix altogether. Lewis Decl. ¶ 11; Second Lewis Decl. ¶ 10. Plaintiffs seize on this detail as further evidence of an inadequate search. See Pls.’ Mot at 8. But if these tables are not included in the data dictionaries to begin with, whether ICE searches Matrix or, as Plaintiffs want, the databases themselves is immaterial. “[A] data dictionary entry for [such a] table does not exist in Matrix or in any of ICE’s databases.” Second Lewis Decl. ¶ 10. 2 See infra Section V for the one category of records for which changes to the databases over time did not alone justify omissions from Defendant’s productions.

6 Defendant has made “countless changes” to the databases, including that “the IIDS schema has

been entirely restructured.” Lewis Decl. ¶ 11a; Second Lewis Decl. ¶ 13. This renders direct

comparisons of little value. See Lewis Decl. ¶ 11a; Second Lewis Decl. ¶ 13. Contents Plaintiffs

identify as “missing” may simply no longer exist. See Second Lewis Decl. ¶ 13. And, importantly,

Defendant avers that it has “provided all responsive information that was in the data dictionaries

located in the Matrix at the time that the searches were conducted.” Id.

The court therefore concludes that Defendant’s decision to search Matrix for the data

dictionaries made for an adequate search.

V.

There are, however, several questions Defendant does not answer.

Plaintiffs point out, and Defendant admits, that Defendant did not produce “the

‘ENFORCE’ table subset included within the complete EID database.” Def.’s Resp. Stmt. ¶ 48.

Yet Defendant offers no explanation for this omission in either its briefs or declarations, despite

Plaintiffs having pointed it out in their cross-motion. See id.; Long Decl. ¶ 9.

Defendant also does not adequately explain its failure to produce plain-English translations

of codes. Once again, Defendant admits to this omission. Def.’s Resp. Stmt. ¶ 41. Defendant

invokes only the changing nature of the databases—and the resulting futility of comparisons

between the 2014 productions and the present ones—to explain the “alleged discrepancies.”

Second Lewis Decl. ¶ 13. But Plaintiffs sum up the shortcoming of this explanation well: “Even

if the plain-English translations of the database field names have changed with the passage of time,

such fields still exist, are responsive, and should have been produced.” Pls.’ Reply at 7; see also

Second Lewis Decl. ¶ 6 (discussing the process of creating “plain-English names” for the Matrix

7 data dictionaries). ICE has not represented that it produced the plain-English translations that exist

in the current versions of the data dictionaries or, if it omitted them, explained why.

Next, Defendant does not explain its withholding of technical database information.

Plaintiffs state that Defendant did not produce fields related to the databases’ functioning, “such

as fields containing field format [or] field width.” Def.’s Resp. Stmt. ¶ 40. Defendant disputes

this accusation. Id. Defendant avers that it produced a column entitled “Technical Data type,”

which includes information about field “length, type, and format.” Second Lewis Decl. ¶ 11. But

according to Plaintiffs, ICE redacted the column’s contents in full, “even though much of that

information was produced in the [related] case.” Pls.’ Reply at 8. In this case, Plaintiffs challenge

only the adequacy of Defendant’s search, not its withholdings. But Plaintiffs did so “in reliance

on ICE’s representation that ‘the redactions are based on the exemptions at issue in [the related

case], and not on any additional grounds.’” Id. (quoting Joint Status Report, ECF No. 31, ¶ 3).

Because it was not clear the propriety of a potentially new redaction was at issue until Defendant’s

representations with its reply brief, the merits of this issue have not yet been fully presented to the

court. The court therefore cannot rule on the propriety of the withholding at this juncture,

particularly without a justification from Defendant for redacting this information.

Lastly, Plaintiffs fault Defendant for using the Matrix dictionaries as a starting point for its

search for code lookup tables, arguing the inadequacies in the dictionaries caused Defendant to

exclude some tables it should have produced. See Pls.’ Reply at 8–9; Pls.’ Mot. at 11. Defendant

disputes that it relied on the Matrix data dictionaries to inform this aspect of its search, but not

without contradictions. Compare Def.’s Resp. Stmt. ¶ 50 (admitting that it “used the data

dictionaries it pulled from Matrix to identify the code lookup tables in each database and then

extracted the code lookup tables by directly querying the IIDS and EID”), with Second Lewis Decl.

8 ¶ 14 (stating that it “did not refer to the Matrix data dictionaries to identify responsive lookup

tables; [it] directly queried the databases and identified the responsive tables by the suffix in the

name”). For the court to evaluate Plaintiffs’ claim, Defendant must first clarify whether it used

the Matrix dictionaries to inform its search for code lookup tables and, if it did, whether and how

that affects the comprehensiveness of the search results.

Given these shortcomings, before the court can determine whether Defendant performed

an adequate search, Defendant must either (1) provide the court with a supplemental declaration(s)

supporting the adequacy of its search with respect to these issues, or (2) conduct a search for these

additional records and produce them to Plaintiffs. See McKinley v. FDIC, 756 F. Supp. 2d 105,

112–13 (D.D.C. 2010). Defendant also must justify its withholding of the information in the

“Technical Data type” column, unless it produces that information unredacted.

VI.

For the foregoing reasons, the parties’ cross-motions for summary judgment, ECF Nos. 34,

38, are denied without prejudice.

Defendant shall file either a supplemental declaration supporting the adequacy of its search

or a Joint Status Report updating the court on its progress on an expanded search by April 29,

2026. Either must specifically address the issues noted above. See supra Section V.

Dated: March 30, 2026 Amit P. Mehta United States District Judge