In Bennett v. Spear, 520 U.S. 154 (1997), the Supreme Court held a final biological opinion reviewable because it marked the consummation of the agency process and was an action “from which rights and obligations flow,” even though the opinion was merely delivered to another agency for use in considering its proposed action. Id. at 177-78. What are Bennett v. Spear’s implications for FOIA? Must the Fish & Wildlife Service (FWS) and the National Marine Fisheries Service (NMFS) release impending biological opinions that the Environmental Protection Agency (“EPA”) “moots” by changing the proposal on which FWS and NMFS are to opine?
In Sierra Club, Inc. v. United States Fish and Wildlife Service, — F.3d —, 2018 WL 6713260 (9th Cir. December 21, 2018), a split panel addressed that issue. The panel majority held that even though FWS and NMFS had not transmitted their biological opinions finding that EPA’s proposed action would imperil endangered or threatened species, FWS and NMFS could not invoke the deliberative process privilege to shield their unissued biological opinions from public disclosure. The case exemplifies the difficulty in applying the deliberative process privilege to multi-agency decision-making processes. In the last ten years, scholars have illuminated the variety of ways in which distinct agencies engage in coordinated decision-making. See, e.g., Jody Freeman & Jim Rossi, Agency Coordination in Shared Regulatory Space, 125 HARV. L. REV. 1131 (2012); Bijal Shah, Interagency Transfers Of Adjudication Authority, 34 YALE J. on REG. 279 (2017).
The deliberative process privilege shields pre-decisional, deliberative material from disclosure. See, e.g., Carter v. Dep’t of Commerce, 307 F.3d 1084, 1089 (9th Cir. 2002); Formaldehyde Inst. v. Department of Health and Human Services, 889 F.2d 1118, 1121-22 (D.C. Cir. 1989). Exemption 5 makes the privilege applicable to FOIA requests. NLRB v. Sears-Roebuck & Co., 421 U.S. 132, 148-52 (1975). The privilege “protect[s] the ‘decision making processes of government agencies,’” by shielding from disclosure “documents ‘reflecting advisory opinions, recommendations and deliberations comprising part of a process by which governmental decisions and policies are formulated.’” Id. at 150. The privilege reflects the theory “that the ‘frank discussion of legal or policy matters’ in writing might be inhibited if the discussion were made public; and that the ‘decisions’ and ‘policies formulated’ would be the poorer as a result.” Id.; see generally, United States v. Nixon, 418 U.S. 683, 705 (1974).
The privilege has two secondary purposes. It protects against premature disclosure of proposed policies before their final adoption. Coastal States Gas Corp. v. Department of Energy, 617 F.2d 854, 866 (D.C. Cir. 1980). It also prevents disclosure of reasons the agency did not adopt, thereby reducing public confusion. Id.; see, Renegotiation Board. v. Grumman Aircraft Engineering Corp., 421 U.S. 168 (1975). Of course, for agencies the privilege might serve a more cynical function. The agency may wish to withhold pre-decisional material because such materials might draw into question the soundness of the agency’s final decision. The agency’s failure to address cogent concerns raised during its internal debates could prove embarrassing or even provide a basis for challenging its decision.
The Supreme Court addressed application of the privilege to inter-agency communication in Renegotiation Board. v. Grumman Aircraft Engineering Corp., 421 U.S. 168 (1975). The case involved a complex process for determining whether government contractors had made excess profits. The system involved Regional Boards, which developed the facts and made recommendation for resolution of the issues to the Renegotiation Board. Id. at 170-79. The Court of Appeals had analogized Regional Boards to district courts. The Supreme Court rejected the analogy. District court opinions have “real operative effect independent of ‘review’ by a court of appeals: absent appeal by one of the parties, the decision has the force of law; and, even if an appeal is filed, the court of appeals will be bound, within limits, by certain of the district court’s conclusions.” Id. at 186-87. Regional Board recommendations lacked any such independently operative effect. They could never become final absent Renegotiation Board review, and Renegotiation Board consideration of such recommendations was de novo in every respect. Id. at 187. The Supreme Court concluded that because Regional Board recommendations lacked legal force, they were not final decisions and thus entitled to the protection the deliberative process privilege affords. Id. There was some dispute over whether or not the Regional Boards and the Renegotiation Board were separate agencies. But, the Court said, the same reasoning applied whether the agencies constituted one agency or separate, distinct agencies. Id. at 187-88.
With the above background, consider the following three scenarios.
Example A. When an action potentially affecting endangered or threatened species is contemplated, FWS must be consulted. FWS has an obligation to issue a biological opinion, which the agency taking the action must consider in crafting its final action. FWS notifies EPA that its contemplated action will jeopardize one or more endangered or threatened species, but does not officially forward the opinion to EPA. Upon hearing that FWS will issue and adverse biological opinion, EPA advises FWS that it will revise its plans. EPA submits a revised proposal to FWS for a biological opinion. FWS issues a biological opinion concluding that the revised proposal will not put any endangered or threatened species at risk. Is the first biological opinion protected by the deliberative process privilege? (This is a simplified version of the facts in Sierra Club, Inc. v. United States Fish and Wildlife Service.)
Example B. The owner of a saltworks near the San Francisco Bay seeks an Approved Jurisdiction Determination (AJD) regarding the federal government’s jurisdiction over the site under the Clean Water Act (CWA). EPA has authority to make such AJDs but has delegated its authority to the Army Corps of Engineers. However, EPA has retained “special case” authority, allowing it to reclaim from the Corps’ the power to make particular CWA jurisdiction determinations. When the Corps is “on the verge” of issuing an AJD regarding the saltworks, EPA invokes its “special case” authority to reclaim jurisdiction. Is the Army Corps’ AJD protected by the deliberative process privilege? (This is a simplified version of the facts in Hunton & Williams v. EPA. 248 F.Supp.3d 220, 245, 247-48 (D.D.C. March 31, 2017).)
Example C. The Department of Commerce is considering whether to add a question on citizenship to the 2020 census. It asks the Department of Justice whether inclusion of such a question would be useful in its Voting Rights Act litigation. By letter, DOJ responds affirmatively. The Commerce Department decides to add the question. Among the reasons it gives for doing so is the usefulness of the resulting data in voting rights litigation. However, it does not cite specifically the response it received from DOJ. Is the DOJ’s response protected by the deliberative process privilege? (This is a significantly-modified version of the course of events leading to the litigation in New York v. Department of Commerce, 315 F.Supp.3d 766 (S.D.N.Y. 2018)(denying motion to dismiss).)
Addressing the Three Examples
As noted above, the privilege is designed to allow government officials to frankly discuss issues in reaching a final position. The final decision itself must be made public, NLRB v. Sears-Roebuck & Co., 421 U.S. at 153-54, and any pre-decisional documents adopted in the final decision must be disclosed, Id. at 161; National Council of La Raza v. Department of Justice, 411 F.3d 350, 358-60 (2d Cir. 2005). The privilege applies not only to intra-agency communications, but to inter-agency communications as well. The desire to avoid chilling the contributions of those subordinate to the ultimate decision-maker or truly tentative, not-fully-considered thoughts of the ultimate decision-make makes sense. But agencies should have some level of accountability when setting forth fully-developed, final positions, even when those positions are taken in a decision-making process that involves multiple agencies. While individual staffers or non-political officials offering comments on potential agency actions should be shielded public scrutiny, agencies and ultimate decision-makers taking fully-considered, final positions should be accountable for those “positions.” Agencies and ultimate decision-makers should be accountable, even when issuing final decisions whose direct audience is another agency. Such accountability is appropriate all the more when the agency has been given a formal statutory role in constraining another agency’s decision-making, i.e., when its decision has “an operative effect.”
Thus, in example A above, there is no need to shield the final, considered view of the FWS, to be signed by the head of the agency, from public scrutiny, that is, from the right of the people to inform themselves about “what their government is up to,” U.S. Department of Justice v. Reporters Committee for Freedom of the Press, 489 U.S. 749 (1989). The knowledge that such documents will be released will presumably not chill FWS – the documents are intended for release in public form. Moreover, particularly when a document imposes legal constraints on another agency, final opinions that are not released because they succeed in convincing the acting agency to alter its proposal, should be subject to release. It allows the public to fully assess the manner in which FWS is acquitting its statutorily-mandated responsibilities. The situation is distinguishable from the Regional Board opinions at issue in Grumman, which were never intended to have any operative legal effect.
Sometimes agencies submit formal comments in another agency’s rulemaking proceedings. U.S. v. Nova Scotia Food Products, 568 F.2d 240, 253 (2d Cir. 1977). An agency’s final draft of such a comment that it never files because the agency it seeks to influence agrees to modify its proposal, may likewise qualify as document that must be released. Agencies are not required to submit such comments, and presumably most often use far more informal and confidential means to make their position known. Moreover, unlike the issuance of a biological opinion, issuance of a comment is presumably not reviewable. Nevertheless, given the standards governing judicial review, such comments may well impose upon the rulemaking agency an obligation to explain itself if it chooses to reject those comments. See U.S. v. Nova Scotia Food Products, 568 F.2d at 253. It is not clear, however, that the legally operative effect rule should apply when an agency’s communication has no greater legally operative effect than the comments of a member of the general public.
In example B above, the Army Corps’ is operating under delegated authority. It lacked power to issue any opinion absent a conditional delegation from the EPA. If an agency’s authority is withdrawn, its opinion, even one on the verge of release, should presumably be treated as a nullity. Nevertheless, it is unlikely that the Corps would have written the AJD any differently given the prospect of the AJD’s public release. Indeed, the Corps presumably expected it to be a final decision that would become public.
Release of the Corps’ AJD could theoretically confuse the public as to the ultimate basis of the AJD, implicating a secondary rationale underlying the deliberative process privilege. However, such confusion is unlikely given that the sources of the operative and the “mooted” AJDs are two distinct agencies, EPA and the Corps respectively. The prospect that final AJDs on the verge of being released before special case authority is invoked being subject to public disclosure could potentially discourage communications between the Army Corps and EPA. Hunton & Williams v. EPA. 248 F.Supp.3d at 248. However, such an effect may be unlikely, given the rarity of withdrawal of decisional authority after the Army Corps has completed, but before it has issued, an AJD.
Example B is distinguishable from Grumman. While the Regional Board recommendations at issue in Grumman were never intended to have any operative effect, the Army Corps Clean Water Act AJDs are. But it is not clear that this distinction should lead to a different result, unissued Corps AJDs have no greater operative effect than Regional Board recommendations, regardless of the Corps’ expectations.
Resolution of the issue is difficult – the pull of formality, namely that opinions from agencies that have lost jurisdiction should be considered a nullity, is strong. In Hunton & Williams the district judge tentatively held that even though the Corps’ “final” draft AJD was a “whisker’s breadth” from completion, because it had never been finalized nor adopted the document remained pre-decisional. 248 F.Supp.3d at 246.
Example C is presumably somewhat more typical of inter-agency interactions. The relationship between agencies is not one in which one agency imposes a formal constraint on another agency’s decisionmaking. Nor does one agency have supervisory authority over another. In the ordinary case the communication from the DOJ should be covered by the deliberative process privilege. Stripping deliberative process privilege protection from agency responses to outreach from other government agencies, and subjecting them to disclosure, might well discourage such agencies from responding at all, giving candid responses, or at least giving candid responses in any permanent form (as opposed to a non-permanent form, such via telephone or in-person conversations).
Revealing such communications is admittedly of less concern than disclosure of communications between staffers of various agencies collaborating on a problem or providing their views in response to a question. It is also less of a concern than revealing tentative or preliminary thoughts of agency heads during communications with officials in other government agencies. Unlike in Example A, opinions offered in such a context neither legally constrain agency choice nor even compel an agency to respond on pain of having its decision reversed upon judicial review.
Nevertheless, courts should not discourage inter-agency consultations by requiring disclosure of even agency, as opposed to individual officials’, views lest it discourage the very inter-agency coordination saluted in Sierra Club v. Costle, 657 F.2d 298, 406 (D.C. Cir. 1981). While mandatory consultative arrangements may well not be chilled by disclosure of agencies final, considered positions, voluntary consultative arrangements may be chilled by a restrictive interpretation of the deliberative process privilege.
The FOIA Improvements Act of 2016, Pub. L. No. 114-185, 130 Stat. 538 (2016), will diminish the importance of formalistic distinctions or analysis of whether documents meet the pre-decisional and deliberative metrics established by court. The statute codified the Reno/Holder standard for release of documents. Sen. Rep. No. 114-4, at 3, 7-8 (2015). In particular, even documents that fit within an FOIA exception should be released unless “the agency reasonably foresees that disclosure would harm an interest protected by an exemption . . . or disclosure is prohibited by law”. Id. §2 (adding 5 U.S.C. §552(a)(8)(A)(i)). Under such a standard, resolution of these cases may turn on the extent of the harm, in terms of of chilling speech and undermining agency functioning, likely to flow from disclosure of particular types of deliberative documents. Such considerations will make Example A an easier case, and could potentially tip the scales toward disclosure in Example B. It might even suggest that in some cases resembling Example C agencies will have to disclose such communications.
That said, the courts will have to recognize that the chilling effect of the release of deliberative documents is far more cumulative that the disclosure of documents implicating interests protected by other exemptions. Any one disclosure of deliberative documents will surely have a minimal chilling effect on deliberations. The harm from release of particular documents can ultimately only be judged in relation to the frequency with which deliberative documents are released as a whole. (Contrast this with the release of evidence identifying a government informant — the first release of such information in quite likely to have the adverse effect the exemption was designed to prevent.)
Sierra Club, Inc. v. United States Fish and Wildlife Service
Manufacturing and processing facilities use cooling water intake structures to draw water from natural sources. The Clean Water Act, directs the EPA to regulate such structures to minimize damage to aquatic life. Sierra Club, Inc. v. United States Fish and Wildlife Service, 2018 WL 6713260 at *1. In April 2011, the EPA proposed new cooling water intake structure regulations. On June 18, 2013, EPA initiated the consultations with FWS and NMFS (collectively, the Services) mandated under the Endangered Species Act. On December 3, 2013, the Services informed the EPA that they would issue “jeopardy opinions” on or around December 6, 2013. Id. at *3. After receiving the Services’ joint proposed “reasonable and prudent alternatives” (RFAs) and portions of the jeopardy opinions, EPA chose to revise its proposed regulation. Id. On March 14, 2014, EPA sent the Services a new rule for review. On May 19, 2014, the Services issued a joint final “no jeopardy” biological opinion regarding the March 2014 proposed rule. Id.
The Sierra Club sought 16 documents produced in conjunction with the Services’ review of the cooling water intake structures regulation. The District Court ordered 12 documents released in whole or in part, including NMFS’ 289-page December 6, 2013 draft jeopardy biological opinion and FWS’ December 9, 2013 72-page draft jeopardy biological opinion. Id. at *3-*4. After in camera review, the majority concluded that neither biological opinion could be withheld. Id. at *10.
The majority separately analyzed whether the documents satisfied the privilege’s pre-decisional and deliberative requirements. In concluding that the opinions were not pre-decisional, the majority explained that the opinions stated the Services’ final, considered positions on EPA’s initial proposed regulations, even though never publicly issued. Id. at *7. Because issuance of a biological opinion constitutes final agency action by the issuing agency, id. (citing Bennett v. Spear, 520 U.S. at 178), the opinions’ pre-decisional status from the perspective of the receiving agency’s decisional process, namely EPA’s overall rulemaking, did not matter, id. The December 2013 jeopardy biological opinions were not intended to advise another decision-maker higher up the chain about the position the Services should take. Id. Nor were the December biological opinions pre-decisional in relation to the Services’ subsequent “no jeopardy” opinions, because the December draft opinions and the May opinions addressed different versions of EPA’s rule. Id. at *8.
The December draft opinions were also deliberative. The majority noted that the Supreme Court had cautioned against resolving such questions using a “‘wooden’ facts-versus-opinions dichotomy.” Id.; EPA v. Mink, 410 U.S. 73, 91 (1973). Rather, it applied the Ninth Circuit’s “functional” approach, which considers “whether the contents of the documents ‘reveal the mental processes of the decisionmakers’” and would “expose [the Services’] decision-making process in such a way as to discourage candid discussion.” Sierra Club, at *8 (quoting Assembly of the State of California v. U.S. Department of Commerce, 968 F.2d 916, 920-21 (9th Cir. 1992)). The majority concluded that the biological opinions neither revealed deliberations within FWS or NMSF nor those agencies’ deliberations with EPA. Id.
The Services argued that plaintiff sought the December opinions to “uncover any discrepancies [in] the findings, projection and recommendations” between those opinions and the final May 2014 biological opinions. Citing National Wildlife Federation v. U.S. Forest Service, 861 F.2d 1114 (9th Cir. 1988), the Services argued that the deliberative process privilege was designed to preclude requesters from making such comparisons. Sierra Club, at *8.
The majority noted that the draft forest plans at issue in National Wildlife Federation were a collection of “tentative opinions and recommendations of Forest Service employees.” Id. at *9. And the draft environmental impact statements at issue compared alternative forest plan proposals recommended by Forest Service employees, thereby revealing the agency’s deliberations in choosing a final plan. In the majority’s view, the deliberative process privilege was designed to protect “the opinions of individuals or groups of employees,” not “the position of an entire agency.” Id. The FWS and NMSF December 2013 biological opinions were neither prepared by low-level officials nor contained merely tentative findings. They contained no line edits, marginal comments, nor other written material that exposed any internal agency discussion. Id. at *10. And, both opinions both state they were prepared on behalf of the entire agency and represent that agency’s opinion. Id.
Nor would releasing these opinions and accompanying documents allow a reader to reconstruct the “mental processes” that led to the production of the May 2014 no jeopardy opinion by allowing comparison of an earlier draft to the final opinion. There was no later version of an opinion regarding the EPA’s initial proposal. Id. at *11.
Judge Wallace, in dissent, concluded that the biological opinions were pre-decisional “because they do not reflect the Services’ final jeopardy [assessment]” of EPA’s initial proposed regulation. Id. at *14. In his view, draft FWS and NMFS biological decisions become final only upon issuance. That the agencies never issued a final assessment because EPA modified the initial proposal was irrelevant — a draft that “die[s] on the vine . . . is still a draft and thus still pre-decisional and deliberative.” Id. at *13 (quoting National Security Archive v. CIA, 752 F.3d 460, 463 (D.C. Cir. 2014)(Kavanaugh, J.). The December opinions were also deliberative, because they were “part of the deliberative process” by which the Services and the EPA consult on the biological consequences of proposed regulatory actions and potential alternative measures. Id. at *14. The privilege does not distinguish inter-agency and intra-agency documents, but rather accords equivalent protection to both. Id. (citing Grumman, 421 U.S. at 188). Accordingly, the protection the privilege affords employees’ contribution to agency deliberation is not exclusive, and does not preclude conferral of equivalent protection upon “the position of an entire agency.” Id.
# # # # # # #
 Exemption 5 allows agencies to 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, provided that the deliberative process privilege shall not apply to records created 25 years or more before the date on which the records were requested.” 5 U.S.C. §552(b)(5). The italicized clause was added by the FOIA Improvements Act of 2016, Pub. L. No. 114-185, 130 Stat. 538 (2016).
 For a broad criticism of the privilege’s breadth and major premise, see, Edward J. Imwinkelried, The Government’s Increasing Reliance On—And Abuse Of—The Deliberative Process Evidentiary Privilege: “[T]He Last Will Be First,” 83 MISS. L.J. 509 (2014).
 Though this is often inappropriate, given the importance of public participation in many of agencies’ rulemaking and adjudicatory proceedings. See, Wolfe v. Department of Health & Human Services, 815 F.2d 1527 (D.C. Cir. 1987), rev’d, 839 F.2d 768 (D.C. Cir. 1988) (en banc). In Wolfe, requester sought a log revealing the dates on which concurrences were sought and received from other agencies. The agency invoked the deliberative process privilege because the log’s release would “permit and encourage increased and more focused lobbying.” In the panel decision, the majority observed that “[n]o case holds that fear of increased lobbying alone is sufficient to bring information within the scope of the privilege and the exemption,” id at 1532, and that the privilege is not meant to isolate agency decisionmakers from public opinion or to silence public voices, id. (emphasis in original). However, the dissenter prevailed at the en banc stage.
The expansion of public participation in rulemaking and adjudication proceedings postdate FOIA’s enactment. Richard B. Stewart, The Reformation of American Administrative Law, 88 HARV. L. REV. 1669, 1713-16, 1748-55 (1975).
Pre-decision secrecy can be quite important with regard to some agency decision-making. See, e.g., Federal Open Market Committee v. Merrill, 443 U.S. 340, 363 (1979)(if the FOMC’s Domestic Policy Directives contain sensitive information their immediate release would significantly harm the Government’s monetary functions, then a slight delay in the publication of the Directives would be permitted under Exemption 5).
 Jodi Freeman and Jim Rossi categorize inter-agency consultative arrangements as either: (a) discretionary consultation, (b) mandatory consultation, (c) public response requirements, (d) default position requirements, and (e) concurrence requirements. Agency Coordination in Shared Regulatory Space, 125 HARV. L. REV. at 1157-60.
 For a discussion of delegations of adjudicatory authority, see Bijal Shah, Interagency Transfers of Adjudication Authority, 34 YALE J. on REG. 279 (2017).
 For an extended discussion of an early ruling in the case see Bernard W. Bell, Citizenship and the Census: State of New York v. U.S. Department of Commerce (Round One), Part I , Part II, Part III, and Part IV. Other challenges to the Department of Commerce’s plans are being litigated in other federal districts.
 It is not clear how often this occurs.
 There is a ready analogy to judicial opinions. Under United States v. Munsingwear, Inc., 340 U.S. 36 (1950), when the parties settle a case before appellate review, lower court opinions in the matter must ordinarily be vacated. See generally, Pattie Millet, Practice Pointer: Mootness and Munsingwear Vacatur, SCOTUSBLOG (June 10, 2008); 13C CHARLES R. WRIGHT & ARTHUR R. MILLER, ET AL., FEDERAL PRACTICE & PROCEDURE: JURISDICTION § 3533.10 (3d ed.).
 “Our form of government simply could not function effectively or rationally if key executive policymakers [a]re isolated from each other and from the Chief Executive. Single mission agencies do not always have the answers to complex regulatory problems. An overworked administrator exposed on a 24-hour basis to a dedicated but zealous staff needs to know the arguments and ideas of policymakers in other agencies as well as in the White House.”
 There is a countervailing consideration for agencies. Often agencies will want to rely on the expertise of or the concerns expressed by other agencies in defending their own actions. To do so, they need documentation that can be included in an administrative record. Thus, far from hiding DOJ’s involvement in its consideration of adding a citizenship question to the 2020 census, the Commerce Department expressly invoked DOJ’s position that the citizenship data produced by the added question would be helpful in its voting rights litigation. The Commerce Department did so even though the impetus for DOJ’s opinion was the Department of Commerce itself. It had not only suggested the argument to DOJ, but had also insistently urged DOJ to adopt that position.
 As of yet, the standard has not been frequently construed. See Rosenberg v. U.S. Department of Defense, — F.Supp.3d —, 2018 WL 4637363 (D.D.C. September 27, 2018). The Senate Committee discussed the issue at length in its report. Sen. Rep. No. 114-4, at 9-10. While the discussion included cautionary notes regarding application of the standard to a number of exemptions, it contained no such cautionary statements regarding invocation of exemption 5.
 See, e.g., Moye, O’Brien, Hogan & Pickert v. Nat’l R.R. Passenger Corp., 376 F.3d 1270, 1279 (11th Cir. 2004); Grand Cent. Partnership, Inc. v. Cuomo, 166 F.3d 473, 483 (2d Cir. 1999); Providence Journal Co. v. U.S. Dep’t of Army, 981 F.2d 552, 560-61 (1st Cir. 1992).
 National Security Archive v. CIA is one of the strongest affirmations of the deliberative process privilege, and suggests that as a Justice on the Supreme Court will be one of the strongest defenders of the privilege. Indeed, in two respects the FOIA Improvements Act of 2016 rejected Judge Kavanaugh’s approach in National Security Archive. Most directly, the enactment imposed a time limit on the applicability of the deliberative process privilege, an approach Judge Kavanaugh argued would undermine the vitality of the privilege. Id. at 464-65. Moreover, Judge Kavanaugh rejected the argument that the draft agency history at question in the case should be released because the CIA could show no concrete harm from its release, explaining that courts were not free second-guess, on a case-by-case basis, Congress’ conclusion that release of deliberative materials was harmful. Id. at 464. The FOIA Improvements Act of 2016’s enshrinement of the Reno/Holder policy serves as a partial rebuttal to this aspect of Judge Kavanaugh’s reasoning.