Opinion
17-cv-1097 (APM)
09-02-2021
SUSAN B. LONG et al., Plaintiffs, v. IMMIGRATION AND CUSTOMS ENFORCMENT, Defendant.
MEMORANDUM OPINION
AMIT P. MEHTA UNITED STATES DISTRICT COURT JUDGE
I. INTRODUCTION
Plaintiffs Susan B. Long and David Burnham bring this Freedom of Information Act ("FOIA") suit against Defendant U.S. Immigration and Customs Enforcement ("ICE"), seeking the production of 27 fields of data on immigration removals related to ICE's Secure Communities enforcement program. Plaintiffs' request covers fiscal year 2015 through August 2016. ICE previously produced the same fields in response to nearly identical requests from Plaintiffs covering earlier time periods, but now it contends that such production would require the creation of new records and thus is not obligatory under FOIA. Before the court are the parties' cross-motions for summary judgment on that issue. For the reasons that follow, the court grants ICE's cross-motion and denies Plaintiffs' cross-motion.
II. BACKGROUND
A. Factual Background
Plaintiffs are co-directors of the Transactional Records Access Clearinghouse ("TRAC"), "a data gathering, data research, and data distribution organization associated with Syracuse University." See Pis.' Mot. for Summ. J., ECF No. 12 [hereinafter Pis.' First Mot], Decl. of Susan B. Long, ECF No. 12-1 [hereinafter First Long Decl.], ¶ 2. TRAC's primary purpose is to provide "comprehensive information about the staffing, spending, and enforcement activities of the federal government." Id. ¶ 3.
For years, Plaintiffs have submitted monthly FOIA requests to ICE seeking certain data within ICE's Enforcement Integrated Database ("EID"). See Pis.' Second Cross-Mot. for Summ. J., ECF No. 54 [hereinafter Pis.' Mot.], Third Decl. of Susan B. Long, ECF No. 54-1, ¶ 3; First Long Decl. ¶ 9. The EID is an electronic database owned and operated by ICE that "captures and maintains information relating to the investigation, arrest, booking, detention, and removal of persons encountered during immigration and law enforcement investigations and operations conducted by ICE" and other component agencies within the U.S. Department of Homeland Security ("DHS"). Def.'s Mot. for Summ. J., ECF No. 11, Decl. of Maria Jones, ECF No. 11-2 [hereinafter Jones Decl.], ¶ ¶ 6-7; see also Id. ¶ 6 (explaining that "the EID is the common database repository for all records created, updated, and accessed by a number of [DHS] software applications" that "allows ICE officers to manage cases from the time of an alien's arrest, in-processing, or placement into removal proceedings, through the final case disposition"); id. ¶ 8 ("The EID is used as data storage throughout the immigration enforcement lifecycle from arrest to removal or release.").
Plaintiffs' monthly FOIA requests "sought from the EID updated, anonymous case-by-case information about each person whom ICE deported as a result of the Secure Communities Program, an immigration enforcement program administered by ICE, and its temporary successor, the Priority Enforcement Program." Mem. Op. & Order, ECF No. 20 [hereinafter First Mem. Op.], at 2. Each request identified the specific case-by-case information sought by Plaintiffs, including a list of separately numbered sub-requests describing particular fields of information and data elements that Plaintiffs sought from the EID. Id. at 2-3.
Previously, ICE responded to these requests "by providing computer extracts furnished as Excel spreadsheet files derived from the EID." Id. at 3 (internal quotation marks omitted). Although the spreadsheets were derived from the EID, ICE did not query the EID itself in searching for responsive records; instead, it searched the Integrated Decision Support System ("IIDS"), a snapshot of a subset of data from the EID that is updated three days a week. Def's Mot. for Summ. J., ECF No. 53 [hereinafter Def's Mot.], Def's Stmt, of Undisputed Material Facts, ECF No. 53-2, ¶ ¶ 32-33. The spreadsheets "contained fields of information and data elements that corresponded to at least some of the separately numbered requests." First Mem. Op. at 3. But in January 2017, in response to Plaintiffs' FOIA request for data covering fiscal year 2015 through August 2016 ("August 2016 Request"), ICE withheld many of the fields that it had previously provided in response to Plaintiffs' requests covering earlier, overlapping time periods- including a virtually identical request submitted by Plaintiffs several months earlier, which covered fiscal year 2015 through December 2015 ("December 2015 Request"). Id.
Plaintiffs dub these withheld fields the "disappearing fields"-i.e., the fields of information and corresponding data elements from the EID/IIDS that ICE provided in response to the December 2015 Request but not the August 2016 Request. See Compl., ECF No. 1, ¶ 18. The court, however, will refer to these fields as the "disputed fields." The following is a list of the disputed fields and the specific numbered requests to which they correspond in the August 2016 Request, see First Long Decl., Ex. F:
“ICE issues detainers using a form called an I-247. Over time, there have been variations of the I-247 form, which have sometimes resulted in different letter designations such as I-247D and I-247N”. First Mot., Pls.' Resp.to Def.'s Stmt. of Facts & Additional Stmt. of Facts, ¶ 84; see Def.'s Mem. Of P.& A. in Opp'n to Pls.' Mot.' Def.'s Resp. to Pls.' Statement. ECF No. 16 ¶ 84.
ICE denied Plaintiffs' administrative appeal of the agency's response to the August 2016 Request. First Mem. Op. at 4. The agency reasoned that because the relevant fields did not exist in the EID, Plaintiffs were not entitled to them under FOIA. Id
B. Procedural Background
On June 8, 2017, Plaintiffs brought this FOIA action to challenge ICE's denial of their administrative appeal and to compel ICE to produce the disputed fields. After the parties filed cross-motions for summary judgment, on September 28, 2018, the court issued a Memorandum Opinion concluding that there remained a genuine dispute of material fact concerning whether the requests at issue require ICE to create new records. See First Mem. Op. The court found that ICE's explanations were "neither request nor data-point specific," and that "to the extent ICE's declarant addresse[d] a particular request or data point in her declarations, she relie[d] upon generic explanations and hyper-technical rhetoric to describe the 'additional analysis' purportedly demanded by Plaintiffs' requests." Id. at 15. The court denied both parties' motions without prejudice. Id. at 16.
In June 2020, the parties submitted renewed cross-motions for summary judgment. See Def's Mot.; Pis.' Mot. The briefing once again centered on whether ICE's production of the disputed fields necessitates the creation of new records. See Def's Reply in Supp. of Mot. for Summ. J. & Opp'n to Pis.' Cross-Mot. for Summ. J., ECF No. 60 [hereinafter Def's Reply], at 4-16; Pis.' Reply in Supp. of Second Cross-Mot. for Summ. J., ECF No. 63 [hereinafter Pis.' Reply], at 2-9. On May 27, 2021, the court held an evidentiary hearing to resolve factual disputes underlying that question. The court heard testimony-both in open court and in a sealed, ex parte session-from Timothy Gibney, Management and Program Analyst and project manager for ITDS. See May 27, 2021 Hr'g Tr., ECF No. 77 [hereinafter Hr'g Tr.], at 13. The court also heard testimony from Plaintiff Long, an Associate Professor of Managerial Statistics at Syracuse University. Id. At 137.
III. LEGAL STANDARD
Although "the vast majority of FOIA cases" are appropriately resolved on motions for summary judgment, the D.C. Circuit has observed that summary judgment is "not always" proper. Brayton v. Office of the U.S. Trade Representative, 641 F.3d 521, 527 (D.C. Cir. 2011); see also Gov't Accountability Project v. FDA, 206 F.Supp.3d 420, 430 (D.D.C. 2016) (noting that while courts "routinely resolve FOIA disputes in the summary judgment context," "summary judgment cannot be granted if dueling affidavits create a genuine dispute over issues of material fact"). In those situations, "[r]esolving issues of material fact. . . has, for example, required ... an evidentiary hearing." Scudder v. CIA, 25 F.Supp.3d 19, 29 (D.D.C. 2014).
Because evidentiary hearings are rare in the FOIA context, the case law outlining the standards that govern those proceedings is unsurprisingly thin. Here, the court applies the same standard it used to resolve a previous FOIA dispute between the same parties that also involved an evidentiary hearing: the agency has the burden to prove by a preponderance of the evidence that the factual dispute should be resolved in its favor. See Long v. ICE, 464 F.Supp.3d 409, 417 (D.D.C. 2020); cf Nat'l Parks & Conservation Ass 'n v. Kleppe, 547 F.2d 673, 679-83 (D.C. Cir. 1976) (reviewing for clear error a trial court's findings-following a two-day evidentiary hearing-that certain factual predicates for an agency's withholding were supported by "[t]he preponderance of the evidence").
IV. DISCUSSION
FOIA requires that federal agencies, "upon any request for records which (i) reasonably describes such records and (ii) is made in accordance with published rules . . ., shall make the records promptly available to any person," 5 U.S.C. § 552(a)(3)(A), provided those records are not exempt from disclosure, see Id. § 552(b); see also Citizens for Resp. & Ethics in Wash. v. U.S. Dep't of Justice, 602 F.Supp.2d 121, 123 (D.D.C. 2009) ("FOIA provides a 'statutory right of public access to documents and records' held by federal government agencies." (quoting Pratt v. Webster, 673 F.2d 408, 413 (D.C. Cir. 1982))). That said, the statute "imposes no duty on the agency to create records." Forsham v. Harris, 445 U.S. 169, 186 (1980) (emphasis added); see NLRB v. Sears, Roebuck & Co., 421 U.S. 132, 162 (1975) (FOIA provides for "disclosure of certain documents which the law requires the agency to prepare or which the agency has decided for its own reasons to create"). FOIA "only requires disclosure of documents that already exist, not the creation of new records not otherwise in the agency's possession." Nat'l Sec. Conns, v. CIA (NSC II), 969 F.3d 406, 409 (D.C. Cir. 2020).
The question this case presents is whether ICE's production of the disputed fields would require the creation of new records under FOIA. Before answering that question, the court provides the following factual findings for context.
A. The Secure Communities Removals Report and the Disputed Fields
To assist ICE's immigration enforcement efforts, when an individual is arrested by local law enforcement, his fingerprints may be taken and run against the DHS immigration fingerprint database and the Federal Bureau of Investigation's fingerprint database. Hr'g Tr. at 14-15. This process is known as an "interoperability query." See Id. at 15. If the query results in a match, the system presents ICE with a notification that the individual is potentially deportable. Id. at 27. ICE can then choose to issue a "detainer," id., which is a notice informing the local law enforcement agency that DHS intends to assume custody of the individual once he is no longer subject to local law enforcement's detention. The information regarding issued detainers is input into the EID. Id. at 24. Critically, however, the EID does not document whether the detainer was issued because of an interoperability match, or due to some other reason. Id. at 31.
See Dep't of Homeland Sec, Immigration Detainer - Notice of Action at 2, https://www.ice.gov/doclib/secure-communities/pdf/immigration-detainer-form.pdf (last visited Sept. 1, 2021).
DHS's Secure Communities Program-which ran from 2008 to 2014 and was revived from 2017 to 2021-required ICE to report which of its detainers and subsequent immigration removals resulted from an interoperability match. Id. at 16-19. To comply with this reporting requirement, ICE engaged in a multistep process. First, to ensure the continued integrity of the EID, ICE relied on an imported copy of the EID's data in the IIDS, its data warehouse system for reporting, analytics, and queries. Id. at 31-33. Next, ICE ran a "correlation analysis" consisting of multiple "complex matching algorithm[s]" that attempted to link the removals and detainers in the IIDS with particular interoperability queries. Id. at 37-38. The algorithms utilized, among other elements, fingerprint identification numbers and various biographical data of the individual subject to the removal/detainer. Id. The linked enforcement events were then aggregated and eventually packaged into the "Secure Communities (SC) Removals Report." See Def.'s Reply, Suppl. Decl. of Patricia J. De Castro, ECF No. 60-2, ¶ 12; Hr'g Tr. at 41-42. Although production of the S.C. Removals Report was not required between 2014 and 2017, see Hr'g Tr. 17-18, ICE continued running the correlation analysis voluntarily as part of its responses to Plaintiffs' periodic FOIA requests for S.C. Program data, see First Mem. Op. at 3 (noting the December 2015 Request).
"[I]n the context of an S.C. removal," however, "the [disputed] fields don't exist." Hr'g Tr. at 42. More concretely, to produce those fields, ICE must build upon what is available in the S.C. Removals Report. ICE would need to run a "secondary analysis" of the underlying data in the S.C. Removals Report against the broader IIDS removals dataset. Id. at 52-53. That analysis requires additional complex correlations and calculations to establish connections between various data points across the two sources. Id.; see Id. at 43 (reiterating the same).
B. The Creation of New Records
With that context in mind, the court turns to whether producing the disputed fields would require ICE to create new records. The only D.C. Circuit authority meaningfully referenced by either party is National Security Counselors v. CIA. See Def.'s Reply at 12-14; Pis.' Reply at 7. There, the plaintiff "submitted FOIA requests to the CIA for a listing of all FOIA requesters from fiscal years 2008 to 2010 organized under each of four fee categories contemplated by FOIA." NSC II, 969 F.3d at 408. After the CIA declined to process the requests, the plaintiff sued. Id. On appeal, the Circuit held for the agency and concluded that processing the requests "would require the agency to create new records, not to disclose existing ones." Id. at 409. For support, it cited a declaration from the CIA's Information Review Officer stating that, to produce the requested listings, an analyst "would be required to individually review each FOIA request submitted from 2008 to 2010 and manually sort thousands of requests based on fee category." Id. (internal quotation marks omitted). The Circuit also noted that it had no occasion "to consider whether sorting a database by field would involve the creation of new records" because the CIA's FOIA database "d[id] not contain the relevant field (i.e. fee category) as mandatory information in the first place." Id.
ICE asserts that National Security Counselors "is directly on point" and resolves the matter in its favor. Def.'s Reply at 13. The court disagrees. ICE emphasizes that, like in National Security Counselors, neither the IIDS nor the EID contains the disputed fields as a matter of course. See Id. But unlike that case, the information required to produce the disputed fields does exist in the IIDS. See, e.g., Hr'g Tr. at 60 ("But-so is detainer data in IIDS? Of course, yes."), 61 ("So the underlying detainer data exists. I'm not saying-it's obviously there."); see also Id. at 62, 65. The question remains, therefore, whether producing those fields using the information available necessitates the creation of new records. The court holds that it does.
Courts in this District have recognized that "sorting a pre-existing database of information to make information intelligible does not involve the creation of a new record because . . . computer records found in a database rather than a file cabinet may require the application of codes or some form of programming to retrieve the information." Nat'l Sec. Courts. v. CIA (NSC I), 898 F.Supp.2d 233, 270 (D.D.C. 2012) (cleaned up), aff'd, 969 F.3d 406 (D.C. Cir. 2020); see Long v. CIA, No. 15-cv-1734 (TSC), 2019 WL 4277362, at *4 (D.D.C. Sept. 10, 2019) ("[T]he court is unconvinced by the . . . argument that writing new computer code to locate responsive records in [the agency's] database or compiling the resulting records constitute the creation of a new record."). Here, however, to produce the disputed fields, ICE must do more than simply sort through information using preexisting connections-it must create many of those connections in the first instance. As one of the agency's declarants explains, to fulfill Plaintiffs' requests, ICE analysts must "create new data combinations" and sub-combinations that "do not automatically connect to each other in the ICE database." Def's Mot., Decl. of Patricia J. De Castro, Ph.D, ECF No. 53-3 [hereinafter De Castro Decl.], ¶¶ 40-41; see also Id. ¶ 44 ("Ultimately, the problem is that Plaintiffs seek . . . extensive data connections . . . concerning individuals and all their immigration enforcement experiences, and the ICE database structure simply does not accommodate that request format."). The court addresses two groups of disputed fields that illustrate the point.
First, the four fields that relate to detainers: "Detainer Prepare Date," "Detainer Facility City," "Detainer Facility State," and "Detainer Threat Level." Those fields correspond to the information sought in Request 22 ("Was I-247/I-247D issued for individual before removal yes/no, and date issued"); Request 23 ("Was I-247N issued for individual before removal (yes/no), and date issued"); Request 26 ("City, county and state of jail or facility in which the individual was detained prior where the I-247/I-247D/I-247N was sent"); and Request 27 ("Detainer threat level (or corresponding notice threat level)"). See First Long Decl., Ex. F. For those fields, ICE's testifying witness, Timothy Gibney, noted that the "correlation analysis does attempt to put together a detainer with a removal," but "there's no context between the two." Hr'g Tr. at 54; see De Castro Decl. ¶ 80 ("The database does not connect a particular detainer to a particular removal."). The De Castro Declaration provides additional details. See De Castro Decl. ¶¶ 80-83. To produce the "Detainer Prepare Date" field responsive to Request 22, for example, an analyst would "need to first develop a method to connect millions of individuals (with no universal ID number) with all their detainers." De Castro Decl. ¶ 80. "Then they would need to develop a new way to analyze relative S.C. removals and detainer dates to see if a detainer occurred before a removal." Id. Those steps are made more difficult by the fact that "an individual can have many removals and many detainers." Id.
The agency offers a similar explanation for the ten fields that relate to an individual's criminal history. Those fields correspond to the information sought in Request 43 ("Charged with a crime (yes/no) (any charge, not restricted to convictions)"); Request 54 ("Information on every conviction not just the most serious (date of the charge, date of the conviction, NCIC code for charge, level of offense (felony, misdemeanor, citation, etc.), sentence received)"); Request 55 ("Information on every charge not just the most serious for which a conviction has not occurred (date of charge, current status, NCIC code for charge, level of offense (felony, misdemeanor, citation, etc.))"); and Request 57 ("Aggravated felon yes/no"). See First Long Decl., Ex. F. When asked what the agency would need to do to produce those fields, Gibney testified that after pulling the full criminal history, ICE would have to create[] a unique identifier that would link ... the removal case in the Secure Communities dataset to the criminal history." Hr'g Tr. at 56 (emphasis added). To produce the "Charged with Crime" field responsive to Request 43, for instance, an analyst would need to identify the relevant Secure Communities matches and "would then [need to] create a new connection between all those individuals and the criminal data . . . [to] determine a 'yes' or 'no' based on whether they were charged with a crime." De Castro Decl. ¶ 84. That process "is complicated by the fact that an individual may have numerous removals and numerous [potential] criminal charges," the latter of which would require an analyst to "look[] at each matching record for an individual to determine if even one met the standard of 'charged with a crime.'" Id.
As these examples demonstrate, the production of the disputed fields requires ICE to manufacture complex and often imprecise connections between otherwise facially unrelated data in the IIDS. See, e.g., Hr'g Tr. at 54 (stating that "the matching algorithm is a little inconsistent" and certain detainers may not ultimately be related to the matched removal case). Because constructing those connections requires the agency to go beyond merely "sorting . . . information to make [it] intelligible," NSC I, 898 F.Supp.2d at 270, or "writing new computer code to locate responsive records," Long, 2019 WL 4277362, at *4, production of the disputed fields is tantamount to the creation of new records.
Plaintiffs analogize the production of the disputed fields to a hypothetical in which "DOJ had a set of files on all cities that received a DOJ law enforcement grant and a separate list of cities that had a DOJ investigation of a police shooting." Pis.' Reply at 6. In Plaintiffs' view, if DOJ received a request for the file of every city that (1) received a DOJ grant and (2) was also the subject of a DOJ investigation, "it would have to use the list[s] to identify and gather the responsive files but doing so would not involve the creation of new records." Id. But Plaintiffs' hypothetical is distinguishable from the instant case. The former requires simple sorting based on a preexisting identifier (i.e., the city) that naturally connects the two data sources. Producing the disputed fields, by contrast, requires ICE to use analysis and calculations to create predicate links between the relevant data.
Plaintiffs advance two related objections to the agency's characterization of what it takes to produce the disputed fields. Neither moves the needle. First, Plaintiffs highlight the information ICE already produced in response to the August 2016 Request, noting that those responsive records included some S.C. Removals-specific data. Pis.' Reply at 4-5 (citing Jones Decl. ¶ 42). For example, ICE provided Plaintiffs with the "state and county that originated the fingerprint submission resulting in [the S.C. match]." Jones Decl. ¶ 42. Plaintiffs assert that if such information can be produced "straight from IIDS without need for the 'analysis' or 'calculations' that ICE contends constitutes the creation of new records," then the same should be true for the disputed fields "because all records responsive to the FOIA request relate to individuals identified as S.C. matches." Pis.' Reply at 4.
Plaintiffs' argument misapprehends the agency's reason for refusing to produce the disputed fields. Gibney testified that, as part of the agency's "legacy standard reporting process," ICE ran the necessary correlative analysis to build the S.C. Removals Report. Hr'g Tr. at 68. And "if what was requested [in Plaintiffs' FOIA request] . . . was in [the S.C. Removals Report], [the agency] provided it [to Plaintiffs]." Id; see Id. at 69 (agreeing that "if there are Secure Communities fields that are in the database that are responsive, the agency has produced that data"). That explains why some of the fields produced in response to the August 2016 Request incorporated S.C. Removals data. But the disputed fields, Gibney noted, are not native to the S.C. Removals Report. See Id. at 68 ("Other things that they're asking for don't exist and they require analysis and calculation to produce those records."). Recall that to produce those fields, ICE would need to run a "secondary analysis" of the underlying data in the S.C. Removals Report against the broader IIDS removals dataset. Id. at 52-53. It is that required extra analysis that (1) differentiates the disputed fields from other, already produced S.C. Removals-related fields; and (2) forms the basis for the agency's justified refusal to produce the disputed fields.
Second, Plaintiffs assert that ICE's representations about the disputed fields are "belied by the IIDS database schema admitted in evidence in Long v. ICE, [464 F.Supp.3d 409 (D.D.C. 2020)]." Pis.' Reply at 5. The schema's introduction states that the IIDS includes "Secure Communities (SC)" data and has among its sources the "SC Interoperability Controlling Agency (CRI) Master List, and Automated Biometric Identification System (IDENT)." Pis.' Mot., Ex. 1 of Third Decl. of Susan B. Long, ECF No. 54-1, at 1. Various figures and dimensional models further indicate that the IIDS contains multiple fields related to the Secure Communities program. See, e.g., Id. at 9-14. This evidence, Plaintiffs submit, contradicts "ICE's assertion that IIDS 'do[es] not label events as related to SC" Pis.' Reply at 6 (first alteration in original) (quoting Def's Reply, Suppl. Decl. of Patricia J. De Castro, ECF No. 60-2, ¶ 23).
Plaintiffs are correct about the presence of SC-related fields in the IIDS. Hr'g Tr. at 81. But that fact is entirely consistent with Gibney's testimony that the "[disputed] data fields ... are not reflected in the[] schema" in the first place. See Id. at 80-81. The SC-related fields Plaintiffs identify in the schema were created via the correlative analysis process that resulted in the S.C. Removals Report. See Id. at 82 ("[I]t is another extract, transform, load process that creates SC-related data within IIDS, which we then query in the form of our standard Secure Communities removal populations."). And as Gibney testified, ICE already provided Plaintiffs with all the records contained within the S.C. Removals Report that are responsive to the August 2016 Request. See Id. at 68. The disputed fields were not produced precisely because they "simply do[n't] exist in the S.C. removals report." Id. at 42.
As a general matter, Plaintiffs fail to call Gibney's testimony into question. Their cross-examination gave the court no concrete reason to discount the accuracy of Gibney's characterization of the steps required to produce the disputed fields. See Id. at 92-131. Nor did Plaintiff Long's own testimony. See id at 137-64. She testified, for instance, that the work involved in responding to Plaintiffs' requests "certainly isn't creating records" because "[t]here are . . . built-in cross-links between the [data] tables [within the IIDS]." Id. at 140; see Id. ("You just link them and the database management system pulls them together."). She could assert as much because, to her knowledge, ICE is using a "publicly available" commercial database management system for the IIDS. Id. at 163. Yet, on cross-examination, she admitted that she had never directly worked with the IIDS or seen a query being run from within the database. Id. at 162. And, in any event, the "cross-links" that Long posited must exist in the IIDS in fact do not independently appear in the database; rather, they must be created through the correlative analysis process that Gibney described. Accordingly, the court resolves the factual dispute over how the disputed fields are produced in favor of ICE.
In sum, because the court credits ICE's explanation of how the disputed fields are produced and further concludes that such production requires the creation of new records, it grants
Defendant's cross-motion for summary judgment and denies Plaintiffs' cross-motion.
V. CONCLUSION
For the reasons set forth above, the court grants Defendant's cross-motion for summary judgment and denies Plaintiffs' cross-motion. A separate final order accompanies this Memorandum Opinion