Natural Resources Defense Council v. United States Environmental Protection Agency

UNITED STATES DISTRICT COURT SOUTHERN DISTRICT OF NEW YORK -----------------------------------------------------------x NATURAL RESOURCES DEFENSE COUNCIL and ENVIRONMENTAL DEFENSE FUND, Plaintiffs, 18-cv-11227 (PKC) -againstOPINION AND ORDER UNITED STATES ENVIRONMENTAL PROTECTION AGENCY, Defendant. -----------------------------------------------------------x CASTEL, U.S.D.J. Plaintiffs Natural Resources Defense Council (“NRDC”) and Environmental Defense Fund (“EDF”) commenced this Freedom of Information Act (“FOIA”) action against the United States Environmental Protection Agency (“EPA”), seeking the disclosure of a computer program known as the “Optimization Model for Reducing Emissions of Greenhouse Gasses from Automobiles” (“OMEGA”). Currently before the Court are the parties’ crossmotions for summary judgment. The sole issue is whether the EPA may withhold a component of the latest version of OMEGA called the “core model” under FOIA Exemption 5, the “deliberative process privilege.” For the reasons that follow, the Court holds that the EPA properly withheld OMEGA’s latest “core model” pursuant to FOIA Exemption 5. Accordingly, defendant’s cross-motion for summary judgment is granted and plaintiffs’ motion for summary judgment is denied. BACKGROUND The OMEGA Model Pursuant to its duty under the Clean Air Act, the EPA establishes federal standards for greenhouse gas (“GHG”) emissions from new motor vehicles. (Charmley Decl. ¶ 8; Oge Decl. ¶ 5); see 42 U.S.C. § 7521 (a)(1). Vehicle manufacturers can comply with these standards by choosing to incorporate certain emissions-reducing technologies into the design of their vehicle fleets. (See Charmley Decl. ¶ 9). Emissions-reducing technologies can include, for instance, “engine technologies, tires, transmission options, and hybrid and electric vehicle options.” (Lutsey Decl. ¶ 18). For any given emissions standard, there are “an almost infinite number of technology combinations” that a vehicle manufacturer could use to comply with the standard. (Lutsey Decl. ¶ 9; Oge Decl ¶ 7). Different technologies vary in terms of cost and effectiveness. (Id.). The Clean Air Act provides that any emissions standard “shall take effect after such period as . . . necessary to permit the development and application of the requisite technology, giving appropriate consideration to the cost of compliance within such period.” 5 U.S.C. § 7521 (a)(2). The EPA developed OMEGA as a modeling tool to assist itself in establishing emissions standards. (Charmley Decl. ¶¶ 8, 9; Wehrum Decl. ¶ 6; Oge Decl. ¶ 19; Pl. Br., Ex. A). OMEGA operates by evaluating the cost and effectiveness of certain technologies, predicting the various ways that manufacturers could combine technologies to achieve compliance, and estimating the cost of complying with various emissions standards. (Id.). Broadly speaking, OMEGA models consist of five components: the inputs, the pre-processors, the core model (the subject of this FOIA action), the post-processors, and the outputs. (Charmley Decl. ¶ 10; Pl. Br., Ex. B). The “inputs” consist of excel spreadsheets containing quantitative data, such as the specific vehicle models on the market, available emission-reduction technologies and corresponding costs, hypothetical emission targets, and fuel costs. (Charmley Decl. ¶ 10; Lutsey Decl. ¶¶ 11, 13, 20, 22). The “pre-processors” translate the inputs into a form that can be processed by the core model, while the “post-processors” translate the outputs into readable and usable datasets. (Id.). The “core model” is a computer program written in C# (a programming language), which applies a series of algorithms to the input data to yield the output data. (Lutsey Decl. ¶¶ 11, 20; Wehrum Decl. ¶ 5; see, e.g., Pl. Br., Ex. K). The “outputs” consist of excel spreadsheets of quantitative data, including which combinations of technologies a manufacturer could use to meet a given emissions target and the cost to each manufacturer, per vehicle, of implementing those technologies. (Id.). Five versions of OMEGA have been released in the past, the latest being OMEGA version 1.4.56. (Pl. Br., Ex. A). The OMEGA model that is the subject of this action is OMEGA version 1.4.59, which has been updated by EPA program staff in various ways since version 1.4.56 was released in 2016. (Charmley Decl. ¶ 16). Generally, the program is updated monthly and even weekly by EPA staff at the Office of Transportation and Air Quality. (Charmley Decl. ¶ 14; Wehrum Decl. ¶ 11). The Subject EPA Regulatory Action On August 24, 2018, the EPA and the National Highway Traffic Safety Administration (“NHTSA”) jointly proposed the Safer Affordable Fuel-Efficient (“SAFE”) Vehicles Rule for Model Years 2021-2026 Passenger Cars and Light Trucks (the “SAFE Vehicles Rule”). (Charmley Decl. ¶ 17; Wehrum Decl. ¶ 3); see 83 Fed. Reg. 42,986 (proposed Aug. 24, 2018). If finalized, the rule would amend existing Corporate Average Fuel Economy (“CAFE”) and tailpipe carbon dioxide emissions standards for passenger cars and light trucks and establish new standards, covering model years 2021-2026. (Id.). The EPA and NHTSA determined that it was “reasonable and appropriate” to rely on the U.S. Department of Transportation’s CAFE model to analyze regulatory alternatives for the SAFE Vehicles Rule, rather than OMEGA version 1.4.59. (Charmley Decl. ¶ 18; Wehrum Decl. ¶¶ 4, 8); SAFE Vehicles Rule, 83 Fed. Reg. at 43,000. Although the EPA has relied on previous versions of OMEGA in creating proposed rules in the past, the EPA “did not actually rely on the OMEGA model for analysis or otherwise in the rulemaking process” for the SAFE Vehicles Rule. (Charmley Decl. ¶ 19; Wehrum Decl. ¶ 8; Pl. Br., Ex. A). As a result, the EPA did not release an updated version of OMEGA when the SAFE Vehicles Rule was proposed, nor has it done so since then. (Wehrum Decl. ¶ 8). The EPA asserts that it may use OMEGA to inform rulemakings related to vehicle emissions in the future. (Id. ¶ 9). Procedural History In July 2018, NRDC submitted a FOIA request to the EPA, requesting disclosure of, among other things, “[a]ny and all versions of [OMEGA], not previously made public.” (Pl. Br., Ex. F). The request encompassed the latest full version of OMEGA, which includes the OMEGA core model version 1.4.59 (“OMEGA v.1.4.59” or “the core model”). (Pl. Br., Exs. D, F; Doc. 37 at 1). The EPA has already released to plaintiffs the inputs, pre-processors, and postprocessors associated with OMEGA version 1.4.59. (Charmley Decl. ¶ 10). However, in April 2019, the EPA issued a final response to plaintiffs’ FOIA request, in which it decided to withhold the core model pursuant to 5 U.S.C. § 552 (b)(5), the “Deliberative Process Privilege.” (Pl. Br., Ex. C). Plaintiffs commenced this action on December 3, 2018. (Doc. 1). LEGAL STANDARD “FOIA represents Congress’s balance ‘between the right of the public to know and the need of the Government to keep information in confidence.’” New York Times Co. v. U.S. Dep’t of Justice, 872 F. Supp. 2d 309, 314 (S.D.N.Y. 2012) (quoting John Doe Agency v. John Doe Corp., 493 U.S. 146, 152 (1989)). Therefore, while FOIA “strongly favor[s] public disclosure of information in the possession of federal agencies,” Halpern v. F.B.I., 181 F.3d 279, 286 (2d Cir. 1999), the statute recognizes “that public disclosure is not always in the public interest and thus provide[s] that agency records may be withheld from disclosure under any of the nine exemptions defined in [the statute],” C.I.A. v. Sims, 471 U.S. 159 , 166–67 (1985). “These exemptions are explicitly made exclusive . . . and must be narrowly construed.” Milner v. Dep't of Navy, 562 U.S. 562 , 565 (2011) (citations and internal quotation marks omitted). “Summary judgment is the preferred procedural vehicle for resolving FOIA disputes.” Nat'l Immigration Project of Nat'l Lawyers Guild v. U.S. Dep't of Homeland Sec., 868 F.Supp.2d 284, 290 (S.D.N.Y. 2012). “[A] district court must review de novo an agency’s determination to withhold information requested under the FOIA.” Florez v. Cent. Intelligence Agency, 829 F.3d 178, 182 (2d Cir. 2016). “The government bears the burden of demonstrating that [a FOIA] exemption applies to each item of information it seeks to withhold, and all doubts as to the applicability of the exemption must be resolved in favor of disclosure.” Id. A court may grant summary judgment to an agency based on affidavits or declarations that “[1] describe the justifications for nondisclosure with reasonably specific detail, [2] demonstrate that the information withheld logically falls within the claimed exemption, and [3] are not controverted by either contrary evidence in the record nor by evidence of agency bad faith.” Wilner v. Nat’l Sec. Agency, 592 F.3d 60, 73 (2d Cir. 2009) (citation omitted). Declarations that are “conclusory,” “merely recite statutory standards,” or are “overly vague or sweeping” are not sufficient, but the declarations need not contain “factual descriptions that if made public would compromise the secret nature of the information.” N.Y. Times Co., 872 F. Supp. 2d at 314 ; Larson v. Dep’t of State, 565 F.3d 857, 862 (D.C. Cir. 2009). These agency declarations are “accorded a presumption of good faith.” Carney v. U.S. Dep’t of Justice, 19 F.3d 807, 812 (2d Cir. 1994) (citation omitted). “Ultimately, an agency may invoke a FOIA exemption if its justification ‘appears logical or plausible.’” Am. Civil Liberties Union v. Dep't of Justice, 681 F.3d 61, 69 (2d Cir. 2012). “[W]here the agency’s submissions are ‘adequate on their face,’ district courts ‘may forgo discovery and award summary judgment on the basis of affidavits.’” Brennan Ctr. for Justice at New York Univ. Sch. of Law v. Dep’t of Homeland Sec., 331 F. Supp. 3d 74, 83 (S.D.N.Y. 2018). “In order to justify discovery once the agency has satisfied its burden, the plaintiff must make a showing of bad faith on the part of the agency sufficient to impugn the agency’s affidavits or declarations . . . or provide some tangible evidence that an exemption claimed by the agency should not apply or summary judgment is otherwise inappropriate.” Carney, 19 F.3d at 812 (citations omitted). “[S]peculative assertions are insufficient to rebut the presumption accorded to agency affidavits.” Reynolds v. United States, 350 F. App’x 474, 475 (2d Cir. 2009); see Jones-Edwards v. Appeal Bd. of Nat. Sec. Agency, 196 F. App’x 36 , 37–38 (2d Cir. 2006) (“The plaintiff must provide ‘factual support,’ rather than ‘mere speculation.’”). “Conversely, ‘[s]ummary judgment in favor of the FOIA plaintiff is appropriate when an agency seeks to protect material which, even on the agency’s version of the facts, falls outside the proffered exemption.’” Osen LLC v. United States Cent. Command, 375 F. Supp. 3d 409, 418 (S.D.N.Y. 2019). DISCUSSION The EPA’s ground for withholding OMEGA v.1.4.59 is the deliberative process privilege encompassed by FOIA Exemption 5. Under Exemption 5, the federal government may withhold “inter-agency or intra-agency memorandums or letters that would not be available by law to a party other than an agency in litigation with the agency.” 5 U.S.C. § 552 (b)(5). “Courts have interpreted Exemption 5 to encompass traditional common-law privileges against disclosure,” including the “deliberative process privilege” which the EPA now invokes. Brennan Ctr. for Justice, 331 F. Supp. 3d at 93 . “The deliberative process privilege rests on the obvious realization that officials will not communicate candidly among themselves if each remark is a potential item of discovery and front page news, and its object is to enhance ‘the quality of agency decisions[]’ . . . by protecting open and frank discussion among those who make them within the Government.” Dep’t of Interior v. Klamath Water Users Protective Ass’n, 532 U.S. 1 , 8–9 (2001) (citations omitted); Hopkins v. U.S. Dep’t of Hous. & Urban Dev., 929 F.2d 81 , 84 (2d Cir. 1991) (“[E]fficiency of Government would be greatly hampered if, with respect to legal and policy matters, all Government agencies were prematurely forced to ‘operate in a fishbowl.’”). This exemption has two requirements. First, “the communication must be ‘interagency or intra-agency.’” Klamath, 532 U.S. at 9 . Second, “the document must be (i) ‘predecisional, i.e., prepared in order to assist an agency decisionmaker in arriving at his decision,’ and (ii) ‘deliberative, i.e., actually related to the process by which policies are formulated.’” Brennan Ctr. for Justice, 331 F. Supp. 3d at 93 (citing Nat’l Council of La Raza v. Dep't of Justice, 411 F.3d 350 , 356 (2d Cir. 2005)). “In addition, as with all exemptions, withholding under Exemption 5 is permissible only if an ‘agency reasonably foresees that disclosure would harm an interest protected by [the] exemption.’” New York Times Co. v. United States Dep’t of State, No. 19-cv-645 (JSR), 2019 WL 2994288 , at *3 (S.D.N.Y. July 9, 2019) (quoting 5 U.S.C. § 552 (a)(8)(A)(i)(I)). In support of its withholding, the EPA relies on declarations by William Charmley, the Director for the Assessment and Standards Division of the EPA Office of Transportation and Air Quality, and William L. Wehrum, the Assistant Administrator for the EPA Office of Air and Radiation. The Court holds that these declarations are facially adequate to justify the EPA’s withholding of OMEGA v.1.4.59 pursuant to Exemption 5 and thus preclude an award of summary judgment to plaintiffs. The Court also holds that discovery is not warranted because plaintiffs have not presented contradictory evidence or evidence of bad faith sufficient to rebut the EPA’s presumption of good faith. Summary judgment in favor of the EPA is therefore warranted. The Court will discuss in detail the various requirements of FOIA Exemption 5 below. I. Inter-Agency or Intra-Agency Memorandum or Letter Requirement Plaintiffs argue that OMEGA v.1.4.59 does not meet the first requirement of Exemption 5 because it is not a “letter” or “memorandum,” which plaintiffs define as “prose documents used for interpersonal communication.” (Pl. Br. at 14 (citing Webster’s New World Dictionary of the American Language 918 (Coll. Ed. 1966)). Plaintiffs have cited no cases applying Exemption 5 in such a limited manner and the caselaw does not support plaintiffs’ interpretation. In Chilivis v. S. E. C., 673 F.2d 1205 , 1212 n.15 (11th Cir. 1982), for example, the Eleventh Circuit held that a “computer printout” fell within Exemption 5, reasoning that when Congress created Exemption 5, “[it] clearly intended to exempt any document connected with the agency's deliberative process, not just memoranda and letters.” Chilivis, 673 F.2d at 1212 n.15 (“[I]n applying exemption 5, a court must focus on the contents of a document rather than its form.” (emphasis added)); see also Dudman Commc’ns Corp. v. Dep't of Air Force, 815 F.2d 1565 , 1567–68 (D.C. Cir. 1987) (“Congress enacted Exemption 5 to protect the executive’s deliberative processes—not to protect specific materials.”). The legislative history of Exemption 5 confirms this point. See H.R. Rep. No. 89-1497, at 10 (1966) (“[A] Government agency cannot always operate effectively if it is required to disclose documents or information which it has received or generated before it completes the process of awarding a contract or issuing an order, decision or regulation. This clause is intended to exempt from disclosure this and other information and records wherever necessary without, at the same time, permitting indiscriminate administrative secrecy.” (emphasis added)). The Fourth Circuit has similarly declined to interpret the terms “letter” and “memorandum” as plaintiffs insist, noting that “although Exemption 5 addresses itself only to ‘letters and memorandums,’ the privileges Congress sought to preserve would be gutted if FOIA could be used to reach items like draft pleadings, litigation exhibits, and data on government computers.” Hunton & Williams v. U.S. Dep't of Justice, 590 F.3d 272 , 280–81 (4th Cir. 2010). The Second Circuit has applied Exemption 5 to “tabular or graphic summaries” of factual information. Lead Indus. Ass'n, Inc. v. Occupational Safety & Health Admin., 610 F.2d 70 , 84–85 (2d Cir. 1979). Cases from the D.C Circuit1 have applied the exemption to a myriad 1 The Court discusses a number of cases arising out of the D.C. Circuit throughout this Opinion because it finds the reasoning therein to be persuasive and applicable to the facts of this case. See Roman v. Nat’l Sec. Agency, No. 07cv-4502 (JFB) (WDW), 2009 WL 303686 , at *5 (E.D.N.Y. Feb. 9, 2009), aff’d sub nom. Roman v. NSA, 354 F. App’x 591 (2d Cir. 2009) (“[T]he Second Circuit has evidenced a willingness to look to the law of other circuits— particularly the D.C. Circuit—in the area of FOIA, even when it has not specifically adopted other circuits’ law. This is especially the case when the Second Circuit defines the contours of the FOIA exemptions.” (collecting cases)). of items that do not fall within plaintiffs’ strict definition, such as “cost estimates,” Quarles v. Dep’t of Navy, 893 F.2d 390 , 392–93 (D.C. Cir. 1990), draft historical manuscripts, Russell v. Dep’t of the Air Force, 682 F.2d 1045, 1049 (D.C. Cir. 1982), draft autopsy reports, Charles v. Office of the Armed Forces Med. Exam’r, 935 F. Supp. 2d 86, 95 (D.D.C. 2013), both the inputs used in a “groundwater flow model” and a draft of the model itself, Goodrich Corp. v. U.S. E.P.A., 593 F. Supp. 2d 184 , 189–90 (D.D.C. 2009), and a “computer software program,” Cleary, Gottlieb, Steen & Hamilton v. Dep’t of Health & Human Servs., 844 F. Supp. 770 , 782– 83 (D.D.C. 1993). Based on these counterexamples and the policy considerations underlying Exemption 5, the Court rejects plaintiffs’ argument that OMEGA v.1.4.59 does not fall within the exemption because it is not a “letter” or “memorandum.” II. Predecisional and Deliberative Requirement “A document is predecisional when it is ‘prepared in order to assist an agency decisionmaker in arriving at [a] decision.’” Grand Cent. P’ship, Inc. v. Cuomo, 166 F.3d 473, 482 (2d Cir. 1999). “To find that a document is predecisional, [a] court must be able ‘to pinpoint an agency decision or policy to which the document contributed,’ or was intended to contribute.” Brennan Ctr. for Justice, 331 F. Supp. 3d at 93 . However, “the agency need not ‘point to a specific decision made by [the agency] in reliance on [the document] . . . [as long as the document] was prepared to assist . . . decisionmaking on a specific issue.” Amnesty Int'l USA v. C.I.A., 728 F. Supp. 2d 479, 516, 518 (S.D.N.Y. 2010) (citing Tigue v. U.S. Dep't of Justice, 312 F.3d 70, 80 (2d Cir. 2002)) (“An ‘agency need not pinpoint a particular final decision to which the material contributed.’”); Color of Change v. United States Dep’t of Homeland Sec., 325 F. Supp. 3d 447, 454 (S.D.N.Y. 2018) (“[A]gencies are perennially ‘engaged in a continuing process of examining their policies,’ which requires ‘generat[ing] memoranda containing recommendations which do not ripen into agency decisions; and the lower courts should be wary of interfering with this process.’” (citing N.L.R.B. v. Sears, Roebuck & Co., 421 U.S. 132, 153, n.18 (1975)). “A document is deliberative when it is actually . . . related to the process by which policies are formulated,” i.e. when it “bear[s] on the formulation or exercise of agency policyoriented judgment.” Cuomo, 166 F.3d at 482 (internal quotation marks omitted); Brennan Ctr. for Justice, 331 F. Supp. 3d at 93–94. “[T]he document must ‘reflect[ ] the give-and-take of the consultative process,’ . . . and have been ‘generated as part of a definable decision-making process.’” Id. (citations omitted). Relevant factors to consider include “whether the document (i) formed an essential link in a specified consultative process, (ii) reflect[s] the personal opinions of the writer rather than the policy of the agency, and (iii) if released, would inaccurately reflect or prematurely disclose the views of the agency.” Cuomo, 166 F.3d at 482 (internal quotation marks omitted). The deliberative process privilege encompasses, for example, “draft documents, proposals, suggestions, and other subjective documents which reflect the personal opinions of the writer rather than the policy of the agency” as well as “advisory opinions, recommendations and deliberations comprising part of a process by which governmental decisions and policies are formulated.” Am. Civil Liberties Union v. U.S. Dep't of Justice, 90 F. Supp. 3d 201, 212 (S.D.N.Y. 2015) (citing Cuomo, 166 F.3d at 482 ). “It is well-settled that draft documents, by their very nature, are typically predecisional and deliberative [because] [t]hey reflect only the tentative view of their authors; views that might be altered or rejected upon further deliberation by their authors or by their superiors.” Color of Change, 325 F. Supp. 3d at 453 . A. OMEGA v.1.4.59 Is Predecisional Plaintiffs argue that OMEGA v.1.4.59 is not predecisional because, by defendant’s admission, it was not used to develop the proposed SAFE Vehicles Rule. It is inconsequential that the EPA did not ultimately rely on OMEGA v.1.4.59 in creating the SAFE Vehicles Rule. Amnesty Int'l, 728 F. Supp. 2d at 516 (“[T]he agency need not ‘point to a specific decision made by [the agency] in reliance on [the document] . . . [as long as the document] was prepared to assist . . . decisionmaking on a specific issue.”). The parties do not dispute that the EPA developed OMEGA v.1.4.59 to assist EPA decisionmakers in establishing standards for GHG emissions from new automobiles pursuant to the EPA’s duty under the Clean Air Act. (Pl. Br. at 21; Def. Br. at 9). Though the EPA ultimately relied on the Department of Transportation’s CAFE model instead of OMEGA v.1.4.59 in developing the proposed Safe Vehicles Rule, OMEGA v.1.4.59 nonetheless qualifies as “predecisional” because it was “intended to contribute” to EPA policy decisions regarding GHG emissions standards for new vehicles. See Brennan Ctr. for Justice, 331 F. Supp. 3d at 93 . The fact that, historically, the EPA has used prior iterations of OMEGA to generate data to inform its decisions on GHG emissions standards for vehicles corroborates the EPA’s claim that this iteration was developed to inform future standards. B. OMEGA v.1.4.59 Is Deliberative The EPA’s declarations adequately explain how OMEGA v.1.4.59 “bear[s] on the formulation or exercise of agency policy-oriented judgment.” See Cuomo, 166 F.3d at 482 . In Wehrum’s declaration, for instance, he explains the EPA’s decisionmaking process and why OMEGA’s role in that process renders disclosure of the core model susceptible to revealing EPA deliberations. Specifically, he states that “[t]he regulatory development process and the process of making upgrades to the OMEGA model have traditionally proceeded in parallel” such that “the policy choices made throughout the regulatory development process are inextricably tied to the analytical choices internal to the OMEGA model itself.” (Wehrum Decl. ¶ 12). He continues, “[a]s a regulation develops, EPA’s high level policymakers may realize that they need a different or more substantial type of analysis in a certain area to determine the available policy options that are supported by a robust technical record.” 2 (Wehrum Decl. ¶ 12). He explains that EPA did not release OMEGA v.1.4.59 alongside its proposed SAFE Vehicles Rule because the EPA has not yet “formally relied upon [OMEGA v.1.4.59] in its analysis of a regulatory action such as a proposed or final rule” and that “public release of interim forms of [the OMEGA core model] would divulge information only reflecting the initial opinions of staff and, as such, would reveal the agency’s deliberations.” (Wehrum Decl. ¶¶ 7, 13). Specifically, he asserts that “[t]he inclusion or exclusion of analytical tools, including changes to the algorithms themselves, track the analytical and policy framework of draft versions of or discussions about potential accompanying regulations.” (Wehrum Decl. ¶ 15). In contrast, plaintiffs argue that OMEGA v.1.4.59 is not deliberative because it is an “accounting program” that reads the inputs and performs a pre-set series of mathematical computations and thus constitutes “factual, investigative material.” (Pl. Br. at 16-17). However, 2 Plaintiffs dispute Wehrum’s assertions on this point with declarations from former EPA employee Margo Oge, who served as the Director of the Office of Transportation and Air Quality of the EPA from 1994 until 2012. She declares, among other things, that she was not aware of any finalization process that OMEGA underwent before being released, that she did not know of any policymakers who reviewed OMEGA before release, and that “EPA staff were empowered to share information about the model with stakeholders.” (Oge Supp. Decl. ¶¶ 2-4, 7, 8, 10, 11, 16, 17). These declarations do not overcome the EPA’s presumption of good faith because it is possible that EPA policy regarding the development of OMEGA has changed since Oge left the EPA or that Oge was not personally involved in making substantive changes to OMEGA, while others EPA employees were. With respect to the information that “EPA staff were empowered to share,” the information she describes appears to relate to the generation of inputs, rather than the OMEGA core model. (Oge. Supp. Decl. ¶¶ 17, 16). the deliberative process privilege “was intended to protect not simply deliberative material, but also the deliberative process of agencies.” Mapother v. Dep't of Justice, 3 F.3d 1533, 1538 (D.C. Cir. 1993). Thus, though “[t]he [deliberative process] privilege does not . . . as a general matter, cover ‘purely factual,’ material,” Cuomo, 166 F.3d at 482 , “use of the factual matter/deliberative matter distinction produce[s] incorrect outcomes in a small number of cases,” Dudman, 815 F.2d at 1567–68 (“[R]elease of materials plausibly labeled ‘factual’ will occasionally reveal much about that process.”). These incorrect outcomes often arise in cases involving numerical data or quantitative tools. See, e.g., Quarles, 893 F.2d at 392–93 (holding that “cost estimates” fell within the deliberative process privilege); Cleary, 844 F. Supp. at 782–83 (holding same for a “computer software program”); Goodrich, 593 F. Supp. 2d at 189–90 (holding same for a “groundwater flow model”). The D.C. Circuit explained this issue in Quarles, where it held that “cost estimates prepared by [the Navy] in the course of . . . selecting homeports for ships” were deliberative, though they could plausibly be labeled “factual material.” Quarles, 893 F.2d at 391–93. These cost estimates included estimates for “costs of land, ship berthing, dredging, buildings and facilities, and utilities.” Id. The court there noted that “[n]umbers have a surface precision that may lead the unsophisticated to think of them as fixed,” but sometimes they can “derive from a complex set of judgments—projecting needs, studying prior endeavors and assessing possible suppliers, [and thus] partake of just that elasticity that has persuaded courts to provide shelter for opinions generally.” Id. (“[C]ost estimates . . . [are] far from fixed, as anyone knows who has . . . compared budget estimates with final costs of a government project.”). As a modeling program, OMEGA v.1.4.59 has a similar “surface precision” that may create the perception that it is “fixed.” However, OMEGA v.1.4.59 also partakes of the same “elasticity” as the cost estimates in Quarles—Indeed, one of OMEGA v.1.4.59’s functions is to estimate costs. The difference is that the core model does not “derive from a complex set of judgments” like the cost estimates in Quarles. Rather, it is a manifestation of that “complex set of judgments” itself. In this way, the OMEGA core model is analogous to the “groundwater flow model” in Goodrich and the “computer software program” in Cleary, both of which were held to be deliberative by district courts in the District of Columbia. In Goodrich, the plaintiffs requested the disclosure of a draft “groundwater flow model,” that the EPA developed to analyze groundwater contamination in Rialto, California. Goodrich, 593 F. Supp. 2d at 187 , 189–90. The model “simulate[d] the movement of groundwater at the [Rialto] Site under varying conditions.” Id. The court in Goodrich ruled that the model was deliberative because “evolving iterations of the Model’s inputs and calibration reflect the opinions of the staff currently developing the Model, which may not represent EPA’s ultimate opinions relating to these matters.” Id. Similarly, in Cleary, the Court held that a computer software program that was modified over time and “designed to manipulate a set of [epidemiological] data in a certain way” to analyze the causal relationship between a certain dietary supplement and eosinophilia-myalgia syndrome was deliberative because “computer programs reflect their creator’s mental processes.”3 Cleary, 844 F. Supp. at 774, 783 . 3 In their papers, plaintiffs attempt to distinguish Cleary by pointing out that Cleary addressed the “bespoke computer programs of a single researcher,” which “made it possible to trace her personal ‘mental processes.’” (Pl. Reply at 13). They argue that “Cleary is best understood as protecting the record of the modeling performed by a single, identifiable individual from disclosure,” while in this case, “the public would [not] have any way to identify which agency official(s) directed any changes to the OMEGA core model after its latest release.” (Id. at 13, 13 n.3). This argument is unpersuasive because even if disclosure of OMEGA v.1.4.59 would not allow the public to attribute changes in the OMEGA core model to an individual author, it would still reveal the preliminary thinking of the staff or segment of the staff of the EPA. See Dudman, 815 F.2d at 1569 (“We cannot deny that a person comparing [the draft in issue] with the final manuscript would confront a wealth of changes and would have no way to ascribe a given change to a particular person or editorial stage. But we think this point of little significance . . . The danger of ‘chilling’ arises from disclosure that the [agency] as an institution made changes in a draft at some point—not from disclosure that particular [agency] employees at particular stages in the editorial process made such changes.”). Like the modeling programs in Goodrich and Cleary, OMEGA has evolved over time and its “calibration” reflects the mental processes of OMEGA’s authors. Specifically, in developing the core model, OMEGA’s authors had to make choices regarding which algorithms to use and which “analytical tools” to include—these choices constitute judgments about how to manipulate the data in OMEGA’s input files in a meaningful way. See Ctr. for Biological Diversity v. U.S. Envtl. Prot. Agency, 369 F. Supp. 3d 1, 20 (D.D.C. 2019) (“[T]he deliberative process privilege operates to shield from disclosure agency decision-making reflecting the . . . assessment of factual information or scientific data.” (emphasis added)). As Wehrum put it, “[t]he inclusion or exclusion of analytical tools, including changes to the algorithms themselves, track the analytical and policy framework of draft versions of or discussions about potential accompanying regulations.” (Wehrum Decl. ¶ 15). Because the EPA has not yet relied on OMEGA v.1.4.59 in developing a specific policy, disclosing OMEGA v.1.4.59 may prematurely reveal the EPA’s opinions regarding “how to analyze the data in [OMEGA’s] input files” and “which analytical tools [to] employ[]” when creating a factual basis for GHG emissions standards. (Def. Reply at 3-4; Wehrum Decl. ¶ 15, 16). The risk that the public may be able to discern these preliminary opinions is especially salient because previous versions of OMEGA are available to the public for side-by-side comparison. In his declaration, Wehrum provided an example of an analytical tool. He states that “for at least seven years,” the EPA has considered “the policy question of whether to add an economic simulation or consumer choice sub-model as an analytical tool to the OMEGA model.” (Wehrum Decl. ¶ 18). He asserts that when the EPA released version 1.4.1 of the OMEGA model in 2012, the EPA stated in version 1.4.1’s “Model Documentation” that it had “begun development of an economic simulation or consumer choice component to OMEGA,” and in OMEGA version 1.4.56, the EPA stated “OMEGA may be expanded in the future” to include such a tool. (Id.). Wehrum declares that releasing OMEGA v.1.4.59 could, for example, reveal “whether or not policy consideration was given to including such an analytical tool in the current version of [OMEGA],” “the outlines and parameters of any such hypothetical tool,” and thus “[the] EPA’s pre-decisional thinking about the role of consumer choice in the regulatory development process.” (Id. ¶ 19, 20). Conversely, if no such tool were included, the public might infer that the EPA does not view consumer choice as a relevant factor. (Id. ¶ 19). To address this example, plaintiffs present the declaration of Dr. David Cooke, who declares that a consumer choice tool was “encoded” in OMEGA version 1.4.56, but that it was “turned off.” (Cooke Decl. ¶¶ 10-15). He also opines that, based on his review of OMEGA version 1.4.56 and the input files for version 1.4.59, it is “highly unlikely that EPA has altered the v.1.4.59 core model by changing the consumer choice model.” (Cooke Decl. ¶¶ 10-15). First, the Court notes that the EPA’s reference (and this Court’s reference) to a consumer choice analytical tool serves merely as an illustrative example of a programming choice that would be revealed if OMEGA v.1.4.59 were released, not an exhaustive list of possible choices. In any case, even if a consumer choice tool were encoded in an earlier version of OMEGA, that fact along with Dr. Cooke’s opinion that it is “highly unlikely” that the tool has changed is not inconsistent with the EPA’s claim that disclosing OMEGA v.1.4.59 will reveal the EPA’s current position about the role of consumer choice in developing emissions standards. For instance, disclosure of OMEGA v.1.4.59 would reveal as matter of fact (rather than likelihood) whether the consumer choice tool still exists, whether it has been updated, and whether it has been expanded. The Court concludes that the EPA’s declarations demonstrate that OMEGA v.1.4.59 is deliberative because its disclosure “would inaccurately reflect or prematurely disclose the views of the agency” regarding the how to analyze input data and the role of certain analytical tools (such as the consumer choice tool) in determining GHG emissions standards. See Cuomo, 166 F.3d at 482 (internal quotation marks omitted). III. Foreseeable Harm Requirement As amended in 2016, FOIA requires that “[a]n agency shall withhold information under this section only if the agency reasonably foresees that disclosure would harm an interest protected by an exemption described in subsection (b).” 5 U.S.C. § 552 (a)(8)(A)(i). “An agency may not ‘perfunctorily state that disclosure of all the withheld information—regardless of category or substance—would jeopardize the free exchange of information’ among or between government officials.” Nat. Res. Def. Council v. U.S. Envtl. Prot. Agency, No. 17-cv-5928 (JMF), 2019 WL 3338266 , at *1 (S.D.N.Y. July 25, 2019). “To satisfy the ‘foreseeable harm’ standard,” when invoking the deliberative process privilege, “an agency must explain how a particular Exemption 5 withholding would harm the agency’s deliberative process.” Id. (citations and internal quotation marks omitted). Thus when withholding material based on Exemption 5, the agency must explain “the Exemption 5-related interests that would be harmed by disclosure of the documents at issue.” Id. In doing do, however, “an agency ‘is not required to provide so much detail that [any] exempt material effectively would be disclosed.” Id. The EPA describes in reasonable detail the foreseeable Exemption-5 related interests that would be harmed upon disclosure of OMEGA v.1.4.59. Specifically, Wehrum declares that the release of OMEGA v.1.4.59 would be harmful to the EPA because “it would chill free and open discussions of EPA staff regarding their opinions on the appropriate analytical tools to be included in the model . . . [such that] they would be less likely to test or experiment with new calibrations or tools that could help create a more effective and robust version of the OMEGA model.” (Wehrum Decl. ¶ 21). He states further that this “chilling effect would impact EPA’s decisionmaking processes and ability to have internal discussions and consultations while designing and updating complex models like OMEGA and may harm the agency’s decisionmaking capabilities in the future regulatory development process.” (Id.). He also asserts that the release of OMEGA would “cause public confusion” because the current version “was not relied on in developing the SAFE [Vehicles Rule],” “does not represent the final form that the model would take if it were tied to a regulatory action,” and “does [not] reflect [the EPA’s] final decisions about how the model should be calibrated or run, or which analytical tools it should contain.” (Id. ¶ 22). These assertions adequately “explain how [disclosure] would harm the agency’s deliberative process.” Nat. Res. Def. Council, 2019 WL 3338266 , at *1; see Color of Change, 325 F. Supp. 3d at 454–55 (“[T]he very risk that the deliberative process privilege is meant to protect against [is] public confusion over agency official policy, which may chill frank and candid discussion among government employees.”). IV. Segregability Plaintiffs argue that even if “some portions” of OMEGA v.1.4.59 fall within the deliberative process privilege, other portions are segregable and should be disclosed pursuant to 5 U.S.C. § 552 (b). Specifically, plaintiffs distinguish between OMEGA v.1.4.59’s “uncompiled source code” and its “executable package.” OMEGA v.1.4.59 consists of source code, written in C#, and an executable package consisting of “compiled code.” (Charmley Supp. Decl. ¶ 1). EPA staff draft the source code using C#, while the compiling process translates that source code into a language that can only be processed by a computer (“machine language”). (Id.; Cooke Decl. ¶ 7). To actually run the OMEGA program, a user can open a file within the executable package which will open “a simple user interface analogous to using a phone app or computer program.” (Cooke Decl. ¶ 8). Plaintiffs argue that the executable package should be segregated because “the underlying computer code of the executable package is unintelligible to humans, even programmers who wrote the source code.” (Pl. Br. at 22). FOIA provides that “[a]ny reasonably segregable portion of a record shall be provided to any person requesting such record after deletion of the portions which are exempt under this subsection.” 5 U.S.C. § 552 (b). “Non–exempt information is not reasonably segregable when it is ‘inextricably intertwined’ with the exempt information in a document ‘such that disclosure would compromise the confidentiality of [exempt] information that is entitled to protection.’” Am. Civil Liberties Union v. United States Dep't of Justice, 252 F. Supp. 3d 217, 227 (S.D.N.Y. 2017) (citing Hopkins, 929 F.2d at 86). Though “the agency must provide a detailed justification for its decision that non-exempt material is not segregable,” the agency is also “entitled to a presumption that it complied with its obligation to disclose reasonably segregable material.” Bishop v. U.S. Dep't of Homeland Sec., 45 F. Supp. 3d 380 , 393–94 (S.D.N.Y. 2014). “The quantum of evidence required to overcome that presumption is not clear.” Sussman v. U.S. Marshals Serv., 494 F.3d 1106, 1117 (D.C. Cir. 2007) (discussing two standards—one requiring “clear evidence to the contrary” and the other requiring “evidence that would warrant a belief by a reasonable person” (citing United States v. Chem. Found., 272 U.S. 1, 14-15 (1926) and Nat’l Archives & Records Admin. v. Favish, 541 U.S. 157, 174 (2004))). The EPA has provided the requisite “detailed justification,” but plaintiffs have not presented evidence to rebut the “presumption that [the EPA] complied with its obligation to disclose reasonably segregable material.” See Bishop, 45 F. Supp. 3d at 393–94. Assuming that the executable file on its own is not exempt from disclosure under Exemption 5,4 Charmley explains in his declaration that the executable file is nonetheless not segregable from the source code because there are numerous free “decompiler” programs available online that can convert the compiled code back into the original source code. (Id. ¶ 2). He declares that his staff ran such a decompiler program on OMEGA v.1.4.59, which yielded a “functionally identical version of the OMEGA version 1.4.59 source code.” (Id.). Based on the Charmley declaration, the Court agrees that the executable file is not reasonably segregable from the source code because it is “inextricably intertwined” with the source code “such that disclosure would compromise the confidentiality of [the source code].” See Am. Civil Liberties Union, 252 F. Supp. 3d at 227 . CONCLUSION For the reasons discussed above, the Court holds that the EPA’s declarations are facially adequate to justify withholding OMEGA v.1.4.59 pursuant to Exemption 5. Thus, summary judgment in favor of plaintiffs is not warranted. See Osen, 375 F. Supp. 3d at 418 . Moreover, discovery is not warranted because plaintiffs have not made a showing of bad faith or presented tangible evidence contrary to the EPA’s declarations to overcome the EPA’s presumption of good faith. See Carney, 19 F.3d at 812 (“In order to justify discovery once the agency has satisfied its burden, the plaintiff must make a showing of bad faith . . . or provide some tangible evidence that an exemption claimed by the agency should not apply or summary judgment is otherwise inappropriate.”); Wilner, 592 F.3d at 73 (requiring that agency declarations “are not controverted by either contrary evidence in the record nor by evidence of agency bad faith.”). The Court accordingly holds that the EPA properly withheld OMEGA 4 In any case, it is not clear to the Court why the executable file, on its own, would not independently fall within Exemption 5 because a user who runs OMEGA using the executable file, like a viewer of the source code, would be able to observe which analytical tools the EPA chose to employ. v.1.4.59 under FOIA Exemption 5. Defendant’s cross-motion for summary judgment (Doc. 46) is GRANTED and plaintiffs’ motion for summary judgment (Doc. 39) is DENIED. SO ORDERED. P. Kevin Castel United States District Judge Dated: New York, New York August 22, 2019 — 22 —