Ctr. for Investigative Rptg. v. DOJ

982 F.3d 668

• Court: Court of Appeals for the Ninth Circuit
• Decided: December 3, 2020
• Docket: 18-17356
• Status: Published

Opinion

FOR PUBLICATION

UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT

THE CENTER FOR INVESTIGATIVE No. 18-17356 REPORTING, Plaintiff-Appellant, D.C. No. 3:17-cv-06557- v. JSC

UNITED STATES DEPARTMENT OF JUSTICE, OPINION Defendant-Appellee.

Appeal from the United States District Court for the Northern District of California Jacqueline Scott Corley, Magistrate Judge, Presiding

Argued and Submitted March 6, 2020 San Francisco, California

Filed December 3, 2020

Before: Kim McLane Wardlaw, Milan D. Smith, Jr., and Patrick J. Bumatay, Circuit Judges.

Opinion by Judge Wardlaw; Dissent by Judge Bumatay 2 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

SUMMARY *

Freedom of Information Act

The panel reversed the district court’s summary judgment, and remanded for further factual development, in an action brought by the Center for Investigative Reporting (“CIR”) under the Freedom of information Act (“FOIA”), requesting that the Bureau of Alcohol, Tobacco, Firearms, and Explosives (“ATF”) provide records concerning weapon ownership.

CIR sought records depicting the “[t]otal number of weapons traced back to former law enforcement ownership, annually from 2006 to the present.” ATF alleged that Congress had forbidden the release of that information by approving the Tiahrt Rider to the Consolidated Appropriations Acts of 2005, 2006, 2010, and 2012. The district court held that ATF was not required to disclose the requested information under FOIA.

FOIA Exemption 3 relieves an agency of its obligation to disclose material specifically exempted from disclosure by statute if that statute meets certain requirements outlined in 5 U.S.C. § 552(b)(3).

The panel held that the Tiahrt Rider did not exempt the data sought by CIR from disclosure under FOIA. The panel held that the 2012 Tiahrt Rider – which enacted the language of the 2010 Rider without any alteration – was the only

* This summary constitutes no part of the opinion of the court. It has been prepared by court staff for the convenience of the reader. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 3

operative Rider because the 2010 Rider impliedly repealed the 2005 and 2008 Riders in full. Looking to the 2010 Rider, the panel concluded that it was not a statute of exemption for FOIA purposes because even though it was enacted after the OPEN FOIA Act of 2009, it made no reference to 5 U.S.C. § 552(b)(3). Finally, the panel held that the issue of whether the OPEN FOIA Act’s prospective definition of statutes of exemption as those that cite to 5 U.S.C. § 552(b)(3) was an impermissible legislative entrenchment of a later Congress’s ability to create statutes of exemption was clearly waived.

The panel held that the Tiahrt Rider did not deprive ATF of the funding it needed to turn over the data. The panel further held that the use of a query to search for and extract a particular arrangement or subject of existing data from the Firearms Tracing System database did not require the creation of a “new” agency record under FOIA.

The panel held that based on the existing record it could not answer the question whether the Firearms Tracing System database was currently capable of producing the information CIR sought in response to a search query. The panel remanded for further factual development of the record on this issue.

Judge Bumatay dissented because the majority wrongly held that the Tiahrt Amendment of 2012 must conform to an earlier statute – the OPEN FOIA Act of 2009 – to be effective, and because the majority misconstrued federal law as requiring FOIA disclosures that Congress expressly prohibited. 4 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

COUNSEL

D. Victoria Baranetsky (argued), The Center for Investigative Reporting, Emeryville, California; Andrew P. Bridges and Meghan E. Fenzel, Fenwick & West LLP, Mountain View, California; for Plaintiff-Appellant.

Robin M. Wall (argued), Assistant United States Attorney; Sara Winslow, Chief, Civil Division; David L. Anderson, United States Attorney; United States Attorney’s Office, San Francisco, California; for Defendant-Appellee.

Aaron Mackey, Electronic Frontier Foundation, San Francisco, California, for Amicus Curiae Electronic Frontier Foundation.

Mason A. Kortz and Kendra K. Albert, Cyberlaw Clinic, Harvard Law School, Cambridge, Massachusetts, for Amici Curiae Five Media Organizations and Sixteen Data Journalists.

Jack Jordan, Parkville, Missouri, as Amicus Curiae. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 5

OPINION

WARDLAW, Circuit Judge:

When Congress passed the Freedom of Information Act (FOIA), 5 U.S.C. § 552, it sought to “permit access to official information long shielded from public view” and thereby “pierce the veil of administrative secrecy” that clouded the workings of federal agencies. Dep’t of Air Force v. Rose, 425 U.S. 352, 361 (1976) (internal quotation marks and citation omitted). Congress viewed this commitment to government transparency and an “informed citizenry” as “vital to the functioning of a democratic society.” NLRB v. Robbins Tire & Rubber Co., 437 U.S. 214, 242 (1978). After all, “[g]overnment transparency is critical” to ensure “the people have the information needed to check public corruption, hold government leaders accountable, and elect leaders who will carry out their preferred policies.” Hamdan v. U.S. Dep’t of Just., 797 F.3d 759, 769–70 (9th Cir. 2015); accord Robbins Tire, 437 U.S. at 242.

Today, few issues spawn as much political debate as guns and their role in criminal activity and the government’s role in regulating these weapons. Countless individuals and entities participate in this debate, often relying on statistical data as they advocate for their preferred policy outcomes. This debate is unquestionably one of public importance. For its part, the Executive Branch has long recognized the importance of quantitative data in this arena and, to that end, has spent decades systematically investigating, or “tracing,” the origins of firearms linked to criminal activity. As of 2018, the Bureau of Alcohol, Tobacco, Firearms, and Explosives (ATF) had compiled the results of over 6,876,808 of those traces in an electronic database called the Firearms Tracing System (FTS). 6 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

The Center for Investigative Reporting (CIR) participates in the national debate surrounding guns in America. CIR specifically wants to report on the use in crimes of guns that had at one time been owned by law enforcement agencies. To prepare that report, CIR sought hard data from ATF, filing a FOIA request asking ATF for records depicting the “[t]otal number of weapons traced back to former law enforcement ownership, annually from 2006 to the present.” ATF, however, had never before released that information to the public, and it refused to change course in light of CIR’s request. It instead contended that Congress had forbidden the release of that information by approving the Tiahrt Rider to the Consolidated Appropriations Acts of 2005, 2008, 2010, and 2012. ATF also contended that FOIA did not require ATF to run this search in the FTS database because such a query would require it to create a new agency record.

We do not agree. The Tiahrt Rider does not exempt the data sought by CIR from disclosure under FOIA, nor does it deprive ATF of the funding it needs to turn over this data. Moreover, the use of a query to search for and extract a particular arrangement or subset of existing data from the FTS database does not require the creation of a “new” agency record under FOIA. The only question that thus remains is whether the FTS database is currently capable of producing the information CIR seeks in response to a search query. We cannot answer that question on the existing record and accordingly reverse and remand for further factual development consistent with this opinion.

I.

The disputes in this case arise from two federal statutes passed in the 1960’s—FOIA and the Gun Control Act of 1968 (GCA), Pub. L. No. 90-618, 82 Stat. 1213 (1968)—and CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 7

from Congress’s evolving understanding of the circumstances in which ATF should disclose gun-tracing data gathered under the GCA. We therefore recount the history of these statutes, Congress’s recent involvement in this area, and then the case at hand.

A.

In 1966, Congress enacted FOIA to usher in a newfound era of transparency in the Executive Branch. Milner v. Dep’t of Navy, 562 U.S. 562, 565 (2011). FOIA mandated that federal agencies “disclose records on request, unless they fall within one of nine exemptions.” Id. “These exemptions are ‘explicitly made exclusive’” and “must be ‘narrowly construed.’” Id. (quoting Envt’l Prot. Agency v. Mink, 410 U.S. 73, 79 (1973); Fed. Bureau of Investigation v. Abramson, 456 U.S. 615, 630, (1982)). They thus “do not obscure the basic policy that disclosure, not secrecy, is the dominant objective of the Act.” Rose, 425 U.S. at 361. Over the years, Congress has repeatedly updated and strengthened FOIA. See, e.g., OPEN FOIA Act of 2009, Pub L. No. 111- 83, § 564, 123 Stat. 2142, 2184 (2009); Electronic Freedom of Information Act Amendments of 1996, Pub. L. No. 104- 231, 110 Stat. 3048 (1996) (E-FOIA).

Meanwhile, a year after passing FOIA, and in the wake of the assassinations of President John F. Kennedy, Attorney General Robert Kennedy, and Martin Luther King Jr., Congress passed the Gun Control Act of 1968. The GCA sought to reduce the incidence of “crime and violence,” § 101, 82 Stat. at 1213, by, among other things, creating a statutory licensing and recordkeeping scheme for firearms manufacturers, importers, retailers, and dealers, see 18 U.S.C. § 923. The Attorney General or his current designee, ATF, may obtain and inspect the inventory and sales records created under this scheme for certain 8 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

enumerated reasons, including as part of a criminal investigation. Id. § 923(g).

ATF has used this statutory authority to implement “tracing”—“the systematic tracking of a recovered firearm from its manufacturer or importer, through its subsequent introduction into the distribution chain (wholesaler/retailer), to identify an unlicensed purchaser.” As explained in the First Declaration of Charles Houser, Chief of the National Tracing Center Division of ATF, any law enforcement agency in the country can request that ATF trace a firearm. Upon such a request, ATF tracks a firearm from its manufacturer or importer, through the supply chain of licensed dealers and wholesalers, and on to the first retail purchaser of that gun. A trace usually, but not always, stops with the first retail purchaser, because those purchasers are not subject to the GCA’s recordkeeping requirements.

ATF documents each trace it conducts. These tracing records are maintained in the Firearm Tracing System, an electronic database that logs the “trace data” for each individual trace. As of April 2018, the FTS database contained information from over 6.8 million traces. The FTS database retains substantial information about each individual trace, including:

(i) information about the law enforcement agency requesting the trace, such as the agency’s name, address, case number, and investigative notes provided by the agency; (ii) information provided by the requesting agency regarding its recovery of the firearm, such as the date and location where the traced firearm was taken into custody by the requesting agency; (iii) information about purchasers of the traced firearm; CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 9

(iv) information about possessors of the traced firearm and any associates (i.e., persons with the possessor of the firearm when the firearm comes into police custody), such as their names and addresses, driver’s license information and social security numbers, and any related vehicle information; (v) information identifying each [Federal Firearms License] that has sold the traced firearm; and (vi) information about the traced firearm such as the manufacturer, importer, model, weapon type, caliber, and serial number.

This information is situated in the FTS database in “over 75 tables with a combined total of 800 columns/fields, not including subsystems and integrated or associated systems.”

When it completes a firearms trace, ATF enters a “close- out-code” in the FTS database to signal the status of the completed trace. Firearms traced to a government or law enforcement agency generally receive the close-out code “S5.” Three other codes also reveal that ATF traced a firearm to a law enforcement or government agency: “S6,” “SH,” and “DN.”

ATF prepares various reports and statistical analyses using the FTS database, which it shares with the public on its website and with partnered government and law enforcement agencies. ATF prepares these reports through specialized search queries. After receiving the results of the query, it processes, verifies, and organizes that data through statistical software. Often, ATF creates “visual depictions,” such as graphs or charts from the data, and a “multi-level review process” ensues to verify the accuracy of the data and 10 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

format. “[E]xperienced specialists at the ATF” generally complete this process.

Because the FTS database contains large volumes of quantitative data regarding guns in the United States, ATF has received FOIA requests for permutations of this data. See, e.g., City of Chi. v. U.S. Dep’t of Treas., Bureau of Alcohol, Tobacco & Firearms, 287 F.3d 628, 632 (7th Cir. 2002), vacated by 537 U.S. 1229 (2003). And, prior to 2003, those requests were at times successful in obtaining responsive information. See, e.g., id. at 638.

B.

That status quo began to shift in 2003, when Congress first attached a provision commonly referred to as the “Tiahrt Rider” or “Tiahrt Amendment” to its Consolidated Appropriations Resolution. See Consolidated Appropriations Resolution, 2003, Pub. L. No. 108-7, § 644, 117 Stat. 11, 473–74 (2003) (“2003 Rider”). The 2003 Rider directed that “no funds appropriated” in that Act “or any other Act with respect to any fiscal year shall be available to take any action based upon any provision of 5 U.S.C. § 552 with respect to” firearms tracing records, “except that such records may continue to be disclosed to the extent and in the manner that records so collected, maintained, or obtained have been disclosed under 5 U.S.C. § 552 prior to the date of the enactment of this Act.” § 644, 117 Stat. at 473–74. That is, no appropriated funds could be used to make any new types of FOIA disclosures of firearms tracing records.

Congress has included some version of the Tiahrt Rider in the subsequent Consolidated Appropriations Acts in 2005, CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 11

2008, 2010, and, most recently, 2012. 1 These Tiahrt Riders have become more restrictive—with each Consolidated Appropriations Act since 2005 providing that, beginning in the current fiscal year “and thereafter, no funds appropriated under this or any other Act with respect to any fiscal year may be used to disclose part or all of the contents of the Firearms Trace System database.” 118 Stat. at 2859; 121 Stat. at 1904; 123 Stat. at 3128; 125 Stat. at 609. Each Tiahrt Rider since 2005 has also qualified this general “use of funds prohibition” with varying limitations and exceptions.

The 2005 Rider contained three exceptions to this prohibition. First, it permitted the use of funds to disclose such records to “a Federal, State, or local law enforcement agency or a prosecutor,” but “solely in connection with and for use in a bona fide criminal investigation or prosecution and then only such information as pertains to the geographic jurisdiction of the law enforcement agency requesting the disclosure.” 118 Stat. at 2859. Second, the 2005 Rider permitted the use of trace data in actions commenced by ATF to enforce the GCA or in review of such actions, but it otherwise provided that trace data could not be used, was “immune from legal process,” and could not “be subject to subpoena or other discovery” in any civil action or proceeding. Id. at 2859–60. Third, the 2005 Rider permitted “the disclosure of statistical information concerning total

1 See Consolidated Appropriations Act, 2005, Pub. L. No. 108-447, 118 Stat. 2809, 2859–60 (2004) (2005 Rider); Consolidated Appropriations Act, 2008, Pub. L. No. 110-161, 121 Stat. 1844, 1903– 04 (2007) (2008 Rider); Consolidated Appropriations Act, 2010, Pub. L. No. 111-117, 123 Stat. 3034, 3128–29 (2009) (2010 Rider); Consolidated and Further Continuing Appropriations Act, 2012, Pub. L. No. 112-55, 125 Stat. 552, 609–10 (2011) (2012 Rider). 12 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

production, importation, and exportation by each licensed importer . . . and licensed manufacturer.” Id. at 2860.

The 2008 Rider made six changes that broadened these exceptions. First, it now permitted disclosure to tribal and foreign law enforcement agencies in addition to Federal, State, and local law enforcement agencies or prosecutors. 121 Stat. at 1903–04. Second, it conditioned those disclosures only on their connection to a criminal investigation or prosecution, eliminating the geographic jurisdiction requirement. Id. at 1904. Third, it newly allowed disclosure to “a Federal agency for a national security or intelligence purpose.” Id. Fourth, the 2008 Rider stated that trace data was not only immune from legal process and beyond the reach of subpoena and discovery, but also prohibited from being “used, relied on, or disclosed in any manner” in those proceedings, including through “testimony or other evidence . . . based on the data,” subject to the same exceptions for ATF’s actions enforcing the GCA and in review of such actions. Id.

Fifth and sixth, the 2008 Rider retained the exception for statistical information, now deemed Exception (A), but added two additional lettered exceptions. Exception (B) provided for even greater freedoms to share that information with law enforcement agencies, prosecutors, and national security agencies and officials. Meanwhile, Exception (C)—at issue in this case—allowed for “the publication of annual statistical reports on products regulated by the [ATF], including total production, importation, and exportation by each licensed importer (as so defined) and licensed manufacturer (as so defined), or statistical aggregate data regarding firearms traffickers and trafficking channels, or firearms misuse, felons, and trafficking investigations.” Id. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 13

The 2010 Rider further altered this scheme in three ways. First, though it retained the exception for disclosure to Federal, State, local, tribal, and foreign law enforcement agencies and Federal, State, or local prosecutors, it limited the requirement that such information be “in connection with or for use in a criminal investigation or prosecution” to sharing information with foreign law enforcement agencies. 123 Stat. at 3128. Second, disclosure to any law enforcement agency or prosecutor was prohibited if it “would compromise the identity of any undercover law enforcement officer or confidential informant, or interfere with any case under investigation.” Id. Third, it prohibited any law enforcement officer, agency, or prosecutor that obtained such data from “knowingly and publicly disclosing the data.” Id.

The 2012 Rider is identical to the 2010 Rider, except in the precise language discussing its application for the “current fiscal year and in each fiscal year thereafter.” 125 Stat. at 609. Congress has passed no subsequent Tiahrt Rider, and other than the 2003 Rider, no Rider cites FOIA.

C.

In preparation for a report on gun violence and the links between crime and guns once owned by law enforcement, CIR submitted the FOIA request at issue here. In March 2017, it requested, in relevant part, that ATF provide the “total number of weapons traced back to former law enforcement ownership, annually from 2006 to the present.” 2 CIR maintains that “access to public records

2 The Chief of the National Tracing Center Division of ATF concedes that this request “concerns law enforcement data from the ATF’s Firearms Trace System database.” 14 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

about the involvement of law enforcement weapons in crime is especially important in a functioning civil society” and that this information “is essential for journalists to study so that they may inform the public and ensure government accountability.” When ATF failed to provide a substantive response to its FOIA request, CIR brought this FOIA action to compel disclosure.

The district court found that ATF was not required to disclose the requested information under FOIA and granted summary judgment in favor of the agency. The district court reasoned that the 2005 and 2008 Tiahrt Riders “are still effective prospectively,” were not required to cite FOIA specifically when enacted, and therefore, qualify as withholding statutes under FOIA Exemption 3, 5 U.S.C. § 552(b)(3). To the extent that CIR requested “statistical aggregate data” that fell outside the Tiahrt Rider’s prohibition, the district court held that ATF could not disclose that information without creating a new record, something FOIA does not require the agency to do. The district court entered partial judgment for the Government, and, after the stipulated dismissal of the other claims, this appeal followed.

II.

We have jurisdiction under 28 U.S.C. § 1291. We review a grant of summary judgment in FOIA cases de novo, “employ[ing] the same standard used by the trial court under Federal Rule of Civil Procedure 56(c).” Animal Legal Def. Fund v. U.S. Food & Drug Admin., 836 F.3d 987, 988 (9th Cir. 2016) (en banc). We thus “view the evidence in the light most favorable to the nonmoving party, determine whether there are any genuine issues of material fact, and decide whether the district court correctly applied the relevant substantive law.” Id. at 989. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 15

III.

We consider first whether the Tiahrt Riders render the information CIR seeks exempt from disclosure under FOIA. FOIA Exemption 3 relieves an agency of its obligation to disclose material “specifically exempted from disclosure by statute,” but only if that statute (1) “requires that the matters be withheld from the public in such a manner as to leave no discretion on the issue” or “establishes particular criteria for withholding or refers to particular types of matters to be withheld,” and (2) “if enacted after the date of enactment of the OPEN FOIA Act of 2009, specifically cites to this paragraph.” 5 U.S.C. § 552(b)(3). In applying this exemption, we must ask “whether the statute identified by the agency is a statute of exemption within the meaning of Exemption 3.” Hamdan, 797 F.3d at 776 (citing Cent. Intelligence Agency v. Sims, 471 U.S. 159, 167 (1985)). If so, we assess “whether the withheld records satisfy the criteria of the exemption statute.” Id. (citing Sims, 471 U.S. at 167).

While that inquiry sounds straightforward, we must answer a preliminary question here: which Tiahrt Rider (or Tiahrt Riders) is the asserted “statute of exemption”? After all, Congress passed Tiahrt Riders in five different years, and most of them reflect differing restrictions on ATF’s disclosure of data from the FTS database. Moreover, some of these Riders were passed before the enactment of the OPEN FOIA Act of 2009, while others were enacted after that Act. That timing matters because Riders passed before the OPEN FOIA Act could serve as statutes of exemption without citing to 5 U.S.C. § 552(b)(3), but those passed afterwards must expressly cite to that subsection to constitute statutes of exemption. Given all this, we must 16 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

determine which Tiahrt Rider or Riders are currently operative law.

We conclude that the 2012 Rider—which enacted the language of the 2010 Rider without any alteration—is the only operative Rider because the 2010 Rider impliedly repealed the 2005 and 2008 Riders in full. 3 Having reached that conclusion and upon looking to the 2010 Rider, we conclude that it is not a statute of exemption for the simple reason that, though enacted after the OPEN FOIA Act of 2009, it makes no reference to 5 U.S.C. § 552(b)(3).

1.

While “[r]epeals by implication are not favored,” Radzanower v. Touche Ross & Co., 426 U.S. 148, 154

3 The parties identify no binding or other appellate case law that answers this question, and the district courts that have directly addressed the implied repeal issue are split. Compare Everytown for Gun Safety Support Fund v. Bureau of Alcohol, Tobacco, Firearms & Explosives, 403 F. Supp. 3d 343, 353 (S.D.N.Y. 2019) (“Congress intended each Rider to comprehensively replace its predecessor.”), appeal docketed, No. 19-3438 (2d Cir. Oct. 21, 2019), with Abdeljabbar v. Bureau of Alcohol, Tobacco & Firearms, 74 F. Supp. 3d 158, 175 (D.D.C. 2014) (“Congress’s decision to incorporate similar language into appropriations bills after 2009 demonstrates its intent to continue the disclosure prohibition.”). The other district court cases cited by the Government either blindly follow Abdeljabbar or otherwise gloss over the OPEN FOIA Act and the implied repeal issue. See Reep v. U.S. Dep’t of Just., 302 F. Supp. 3d 174, 183 (D.D.C. 2018); Fowlkes v. Bureau of Alcohol, Tobacco, Firearms & Explosives, 139 F. Supp. 3d 287, 292 (D.D.C. 2015); Higgins v. U.S. Dep’t of Just., 919 F. Supp. 2d 131, 145 (D.D.C. 2013); see also P.W. Arms, Inc. v. United States, No. C15-1990- JCC, 2017 WL 319250, at *4 (W.D. Wash. Jan. 23, 2017). Moreover, in many of these cases, including Abdeljabbar, the plaintiff proceeded pro se. Abdeljabbar, 74 F. Supp. 3d at 164; Reep, 302 F. Supp. 3d at 179; Fowlkes, 139 F. Supp. 3d at 288; Higgins, 919 F. Supp. 2d at 137. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 17

(1976) (quoting United States v. United Cont’l Tuna Corp., 425 U.S. 164, 168 (1996)), the Supreme Court has recognized “two well-settled categories of repeals by implication[:] (1) where provisions in the two acts are in irreconcilable conflict . . . . ; and (2) [where] the later act covers the whole subject of the earlier one and is clearly intended as a substitute,” id. (quoting Posadas v. Nat’l City Bank of N.Y., 296 U.S. 497, 503 (1936)); In re Glacier Bay, 944 F.2d 577, 581 (9th Cir. 1991). “In either case, the intention of the legislature to repeal must be clear and manifest,” Posadas, 296 U.S. at 503, based usually on “the language or operation of [the] statute,” Kremer v. Chem. Constr. Corp., 456 U.S. 461, 470 (1982). Otherwise, “the later act is to be construed as a continuation of, and not a substitute for, the first act,” and the later act “will continue to speak, so far as the two acts are the same, from the time of the first enactment.” Posadas, 296 U.S. at 503.

CIR argues that each of the subsequent Riders was a substitute for the prior one. We thus examine whether the 2010 Rider, repeated in the 2012 Rider in full, “cover[s] the whole subject covered by an earlier act, embraces new provisions, and plainly shows that it was intended . . . to prescribe the only rules with respect thereto.” United States v. Lovely, 319 F.2d 673, 679–80 (4th Cir. 1963) (finding such an implied repeal); accord United States v. Tynen, 78 U.S. 88, 92 (1870) (same); see also 1A Norman J. Singer & Shambie Singer, Sutherland Statutory Construction § 23:13 (7th ed. 2013) (“Legislation which operates to revise the entire subject to which it relates gives strong implication of a legislative intent to repeal former statutory law and also to supersede the common law relating to the same subject.”). While not necessarily an exhaustive list, this analysis involves examining at least: how Congress described its own actions in the subsequent act, Posadas, 296 U.S. at 502; how 18 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

many aspects of the statutory scheme differ and in what depth, id.; and whether the subject matter addressed in both acts is identical in scope, Tynen, 78 U.S. at 92; Lovely, 319 F.2d at 679–80.

2.

Applying these principals here, we conclude that Congress impliedly repealed the 2005 and 2008 Tiahrt Riders through its later passage of the 2010 and 2012 Riders. 4 Because the 2008 Rider was the last version passed before the OPEN FOIA Act, which provided that Exemption 3 would apply to only those statutes enacted thereafter that “specifically cited” to it, and the 2010 Rider was the first version passed after the passage of the OPEN FOIA Act, our analysis focuses there.

The 2008 Rider provides that:

[B]eginning in fiscal year 2008 and thereafter, no funds appropriated under this or any other Act may be used to disclose part or all of the contents of the Firearms Trace System database . . . except to

(1) a Federal, State, local, tribal, or foreign law enforcement agency, or a Federal, State, or local prosecutor, solely in connection with

4 We note also that there is no meaningful distinction between the 2012 Rider and the 2010 Rider. The 2012 Rider applies in the “current fiscal year and in each fiscal year thereafter,” a slightly different wording from the concept expressed in the 2010 Rider, “beginning in fiscal year 2010 and thereafter.” CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 19

and for use in a criminal investigation or prosecution; or

(2) a Federal agency for a national security or intelligence purpose;

and all such data shall be immune from legal process, shall not be subject to subpoena or other discovery, shall be inadmissible in evidence, and shall not be used, relied on, or disclosed in any manner, nor shall testimony or other evidence be permitted based on the data, in a civil action in any State . . . or Federal court or in an administrative proceeding other than a proceeding commenced by the [ATF to enforce the Gun Control Act] . . . or a review of such an action or proceeding . . . .

2008 Rider (emphases added to identify portions affected by the 2010 Rider). In addition, the 2008 Rider includes the three lettered exceptions to the use of funds prohibition, including “that this proviso shall not be construed to prevent . . . (C) the publication of annual statistical reports . . . or statistical aggregate data regarding firearms traffickers and trafficking channels, or firearms misuse, felons, and trafficking investigations.” Id.

The 2010 Rider made substantive modifications to the use of funds prohibition, providing that:

[B]eginning in fiscal year 2010 and thereafter, no funds appropriated under this or any other Act may be used to disclose part 20 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

or all of the contents of the Firearms Trace System database . . . except to:

(1) a Federal, State, local, or tribal law enforcement agency, or a Federal, State, or local prosecutor; or

(2) a foreign law enforcement agency solely in connection with or for use in a criminal investigation or prosecution; or

(3) a Federal agency for a national security or intelligence purpose;

unless such disclosure of such data to any of the entities described in (1), (2) or (3) of this proviso would compromise the identity of any undercover law enforcement officer or confidential informant, or interfere with any case under investigation; and no person or entity described in (1), (2) or (3) shall knowingly and publicly disclose such data;

and all such data shall be immune from legal process, shall not be subject to subpoena or other discovery, shall be inadmissible in evidence, and shall not be used, relied on, or disclosed in any manner, nor shall testimony or other evidence be permitted based on the data, in a civil action in any State . . . or Federal court or in an administrative proceeding other than a proceeding commenced by the [ATF to enforce the Gun Control Act] . . . or a review of such an action or proceeding . . . . CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 21

2010 Rider (emphases added to reflect modifications from the 2008 Rider). The 2010 Rider maintained the same three lettered exceptions as the 2008 Rider.

We have no doubt that the 2010 Rider “cover[s] the whole subject of the” matters discussed in the 2005 and 2008 Riders. See Lovely, 319 F.2d at 679. Like those earlier Riders, it lays out a prohibition on the use of appropriated funds to disclose trace data and spells out a series of specific exceptions to that general prohibition. It likewise generally purports to insulate this data from administrative or civil discovery and bars the use of this data in adjudicatory proceedings. Finally, it also lists scenarios where ATF may both turn over certain types of information from the FTS database and use generally appropriated funds for that purpose.

Moreover, the 2010 Rider is, like its predecessors, clearly intended to prescribe the only rules for the release of data from the FTS. It presumptively bans the disclosure of all such data, making exceptions only as provided in that 2010 Rider. It does not cross reference other statutes or regulations that discuss the FTS data, and neither we nor the parties have located another federal law outside the Tiahrt Rider that speaks directly to the data contained in the FTS. Congress has also effectively endorsed this view, given that it anticipated the 2010 Rider to apply “beginning in fiscal year 2010 and thereafter.” See also Everytown, 403 F. Supp. 3d at 353 (“The use of express repetition of language of futurity in every [Rider] indicates that Congress understood each Rider to provide specific, ongoing rules for Firearms Trace System database disclosure that did not necessitate examining prior enactments on the subject.”).

Furthermore, the 2010 Rider and the two “new provisions” it “embrace[d]” redefined what FTS disclosures 22 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

are even possible. See Lovely, 319 F.2d at 679–80 (noting two operative differences). Consider that the 2010 Rider, in one respect, broadened the scope of the disclosures allowed by the 2008 Rider, because it allowed disclosures to law enforcement without any “connection to a criminal investigation or prosecution.” Yet the 2010 Rider also simultaneously cut back on the previously allowed FTS disclosures because it prohibited disclosures that might compromise the identity of undercover agents or informants. Given these asymmetric changes, a permissible disclosure under the 2008 Rider may or may not be permissible under the 2010 Rider. Or it may still be impermissible, but for a new reason entirely—e.g., a request by law enforcement, unrelated to a criminal investigation or prosecution, may nevertheless pose a risk of revealing an undercover agent’s identity. In short, it makes no sense to look back to the 2005 or 2008 Riders, as the government suggests, because they do not reflect current law. Cf. Gallenstein v. United States, 975 F.2d 286, 292 (6th Cir. 1992) (“The second category of implied repeal is where the later statute fills the entire area of law such that the prior statute has no effect.”).

The above analysis convinces us that Congress intended the 2010 Rider to repeal the previous 2008 Rider. See Tynen, 78 U.S. at 92; Lovely, 319 F.2d at 679–80. Indeed, this case presents a situation far different from Posadas, in which an amendment added only one provision to a much larger and largely independent statutory scheme and in which the Supreme Court concluded that the Federal Reserve Act of 1916 did not impliedly repeal the Federal Reserve Act of 1913, but instead only amended six of the earlier Act’s thirty sections. See Posadas, 296 U.S. at 502.

That this case is quite distinct from Posadas and its progeny is further evidenced by the fact that the 2010 Rider CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 23

does not reference previous Riders or explicitly purport to “amend” previous Riders. 296 U.S. at 502. Indeed, Congress’s refusal to use the word “amend” in the 2010 Rider is notable, as it used the phrase “is amended” 49 times throughout the entire 2010 Consolidated Appropriations Act. See, e.g., § 326, 123 Stat. at 3106 (“The matter under the heading ‘Community Development Fund,’ . . . is amended by striking ‘: Provided further, That none of the funds provided under this heading may be used by a State or locality as a matching requirement, share, or contribution for any other Federal program.”); § 176, 123 Stat. at 3068 (“Section 51314 of title 46, United States Code, is amended in subsection (b) by inserting at the end ‘Such fees shall be credited to the Maritime Administration’s Operations and Training appropriation, . . . .’”). This language suggests that Congress viewed the 2010 Rider as something other than an “amendment.” See Russello v. United States, 464 U.S. 16, 23, (1983) (“[W]here Congress includes particular language in one section of a statute but omits it in another section of the same Act, it is generally presumed that Congress acts intentionally and purposefully in the disparate inclusion or exclusion.”).

The D.C. district court’s analysis in Abdeljabbar, on which the district court in this case heavily relied, does not persuade us otherwise. That decision does not contemplate the issue of repeal by comprehensive replacement. Rather, the Abdeljabbar court rested its holding that the 2005 and 2008 Riders remained in effect despite the passage of the 2010 and 2012 Riders solely on its conclusion that the statutes were not in “irreconcilable conflict.” Citing United States v. Fausto, 484 U.S. 429, 453 (1988), for the proposition that “a later statute will not be held to have implicitly repealed an earlier one unless there is a clear repugnancy between the two,” the district court found that at 24 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

the abstract level of “disclosure prohibitions” the 2008 and 2010 Riders were consistent. Abdeljabbar, 74 F. Supp. 3d at 175 (quoting Fausto, 484 U.S. at 453). Its strong disbelief “that Congress intended to repeal by implication a disclosure prohibition, at least with respect to FOIA, by reiterating that very prohibition in” the 2010 Rider is thus unjustified. Id. Depending on the statute under consideration and its context, the recitation of a previous prohibition with modifications can be evidence of either a mere amendment or of a decision to repeal and replace. Compare Posadas, 296 U.S. at 505 (describing this as a “well-approved form” of “amendment”) with Singer & Singer, supra, § 23.13 (noting that even comprehensive legislative overhauls may “restate, or at least [] include, some provisions of a former law”).

Moreover, Abdeljabbar’s inference that Congress meant to confirm “the plethora of decisions . . . holding that the appropriations language in question” satisfies Exemption 3 by its “uninterrupted use of [the general prohibition] in appropriations bills after 2009,” Abdeljabbar, 74 F. Supp. 3d at 175, is flawed.

First, just seven weeks before passing the 2010 Rider, Congress enacted the OPEN FOIA Act. The OPEN FOIA Act was a direct response to “exemption creep,” whereby “an alarming number of FOIA (b)(3) exemptions” were snuck into legislation “to the detriment of the American public’s right to know.” 155 Cong. Rec. S3175 (daily ed. Mar. 17, 2009) (statement of Sen. Leahy). To address this problem, the OPEN FOIA Act directed courts and agencies to consider future legislation as exempting documents from disclosure only if Congress “cites to [5 U.S.C. § 552(b)(3)].” 5 U.S.C. § 552(b)(3)(B). This measure guaranteed “an open and deliberative process in Congress” before any future statute exempted documents from disclosure under FOIA CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 25

and promised “to reinvigorate and strengthen FOIA.” 155 Cong. Rec. S3175 (daily ed. Mar. 17, 2009) (statement of Sen. Leahy).

If anything, the enactment of the OPEN FOIA Act represented a clear break from Congress’s past habit of creating statutes of exemption in a legislative dead of night. That the 2010 Rider may have sufficed to exempt FTS data from disclosure before the Open FOIA Act is thus irrelevant. “[I]nsofar as Congress wished to enact statutes that would exempt Firearms Trace Database data from disclosure following the enactment of the OPEN FOIA Act, it gave itself explicit instructions for how to do so.” Everytown, 403 F. Supp. 3d at 354; see, e.g., Nat’l Def. Authorization Act for Fiscal Year 2012, Pub. L. No. 112-81, 125 Stat 1298, 1600–01, 1604 (2011) (demonstrating Congress’s compliance with the OPEN FOIA Act). 5

5 We note also that whether the 2005 and 2008 Riders qualified as withholding statutes was by no means settled when the OPEN FOIA Act was enacted in 2009. Other than a handful of district court decisions, it appears that only the Seventh Circuit had held that those earlier Riders qualified as withholding statutes. See City of Chi. v. U.S. Dep’t of Treas., Bureau of Alcohol, Tobacco & Firearms, 423 F.3d 777, 780 (7th Cir. 2005); Cornish F. Hitchcock, Guidebook to the Freedom of Information and Privacy Acts § 8:9 (2020 ed.) (collecting cases). The parties do not cite, and we have not found, any additional precedent on this issue.

Moreover, arguably the 2005 and 2008 Riders are not in fact withholding statutes. We have held that “only explicit nondisclosure statutes . . . will be sufficient to qualify under . . . [E]xemption [3],” whether or not the statute was enacted prior to the OPEN FOIA Act. Cal- Almond, Inc. v. U.S. Dep’t of Agric., 960 F.2d 105, 108 (9th Cir. 1992) (quoting Church of Scientology v. U.S. Postal Serv., 633 F.2d 1327, 1329 (9th Cir. 1980)). The 2005 and 2008 Riders do not explicitly prohibit the disclosure of trace data itself. They merely prohibit the use of funds to make disclosures and make this data “immune from legal process.” 26 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

In sum, the 2010 Rider impliedly repealed the 2005 and 2008 Riders. The 2012 Rider simply reenacted the 2010 Rider. Neither qualifies as a withholding statute because they were both enacted after the effective date of the OPEN FOIA Act and do not cite to 5 U.S.C. § 552(b)(3). Therefore, the documents requested by CIR are not exempted from disclosure under FOIA, 5 U.S.C § 552(b)(3). 6

At best, one might have argued before the OPEN FOIA Act that this “achieve[d] an Exemption 3 effect in an indirect fashion,” Hitchcock, supra, § 8.9, but it was by no means an explicit prohibition on disclosure itself.

But “if Congress intended to prohibit the release of [information] under FOIA—as opposed to the expenditure of funds in releasing the [information]—it could easily have said so.” Cal-Almond, 960 F.2d at 108. In fact, Congress did say so elsewhere in the 2008 Consolidated Appropriations Act. See Br. for Jack Jordan as Amicus Curiae in Supp. of Appellant and Reversal at 24. The 2008 Act stated, “[n]otwithstanding section 552 of title 5, United States Code, . . . the Secretary may not disclose to any person any information obtained” under 6 U.S.C. § 488a, which concerned the maintaining of records of the sale or transfer of ammonium nitrate, Consolidated Appropriations Act, 2008, Pub. L. No. 110-161, § 899B, 121 Stat. 2084–85 (2007) (quoting 6 U.S.C. § 488a(h)). Thus, Abdeljabbar’s reasoning that Congress intended to confirm a “long-standing” interpretation of the Tiahrt Rider as a withholding statute is unpersuasive. See 74 F. Supp. 3d at 176–77. 6 We acknowledge that, in a prior unpublished memorandum disposition of our court, we held that “ATF correctly relied on the Appropriations Act of 2010 as a withholding statute explicitly barring disclosure [of trace data] under FOIA Exemption 3.” Caruso v. U.S. Bureau of Alcohol, Tobacco & Firearms, 495 F. App’x 776, 778 (9th Cir. 2012). Unpublished dispositions are not precedential, however. Nor is Caruso of any persuasive value, as the panel failed to address the doctrine of implied repeal or the impact of the OPEN FOIA Act, possibly CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 27

The dissent does not object to our discussion regarding the implicit repeal of previous Tiahrt Riders. Instead, it criticizes the OPEN FOIA Act’s prospective definition of statutes of exemption as those that cite to 5 U.S.C. § 552(b)(3) as an impermissible legislative entrenchment on a later Congress’s ability to create statutes of exemption. For our part, we have found no federal precedent addressing this weighty issue, and the dissent cites none. We thus view this question as one of first impression. 7

But this is not the case to address that question, for the issue is clearly waived. Neither party raised this point before the district court, see Conn. Gen. Life Ins. Co. v. New Images of Beverly Hills, 321 F.3d 878, 882 (9th Cir. 2003), and no party or amici discussed this issue in briefing this appeal, see United States v. McEnry, 659 F.3d 893, 902 (9th Cir. 2011). Indeed, the parties did not “frame[] this case as [] whether the Tiahrt Amendment or the OPEN FOIA Act governs.” Dissenting Op. at 51. They have instead asked us to resolve whether the 2005 and 2008 Tiahrt Riders remain in effect, even after the enactment of the 2010 and 2012 Tiahrt Riders. See, e.g., Br. of Appellee at 19 (arguing and quoting

because the parties there failed to raise these issues below. See Caruso v. U.S. Bureau of Alcohol, Tobacco, & Firearms, No. Civ. 10-6026-HO, 2011 WL 669132, at *3 n.1 (D. Or. Feb. 16, 2011). 7 We note that, because the 2010 and 2012 Tiahrt Riders do not explicitly purport to repeal the OPEN FOIA Act, this issue at least implicates the following: (1) whether there exists an “irreconcilable conflict” between the Rider and the OPEN FOIA Act, see, e.g., In re Glacier Bay, 944 F.2d 577, and (2) the Supreme Court’s “especially strong” aversion to implied repeals of standalone laws through appropriation riders. Me. Cmty. Health Options v. United States, 140 S. Ct. 1308, 1323 (2020) (internal quotation marks and citation omitted). 28 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

Abdeljabbar to assert there is no need to address the OPEN FOIA Act’s effect, because the 2005 and 2008 Riders “provide[] a permanent prohibition against disclosure”); Reply Brief of Appellant at 4 (noting correctly that the government did not “provide any statutory argument regarding” the application of the OPEN FOIA Act and instead relied on Abdeljabbar’s reasoning).

Indeed, the first—and only—mention of legislative entrenchment came at oral argument, when our dissenting colleague asked the parties about this principle. The government admitted it had not made this argument and that it chose to focus on whether the 2010 and 2012 Tiahrt Riders implicitly repealed prior Tiahrt Riders. Recording of March 6, 2020 Oral Argument at 17:35–18:00. 8 And even after oral argument, the government has not sought to explore this issue through further briefing. CIR, meanwhile, has never offered any briefing on this issue, as it reasonably understood the government had not pursued this argument.

The dissent is, of course, correct that “we have discretion to affirm on any ground supported by the law.” Upper Skagit Indian Tribe v. Lundgren, 138 S. Ct. 1649, 1654 (2018). But “in this case we think restraint is the best use of discretion,” id., as we lack the benefit of analysis from the concerned parties and the district court on this weighty issue of first impression. That course seems doubly wise because, based on oral argument, it appears that the government knew of this potential argument, but may have deliberately chosen not to raise it. See generally United States v. Sineneng- Smith, 140 S. Ct. 1575, 1579 (2020) (“As a general rule, our system ‘is designed around the premise that parties represented by competent counsel know what is best for

8 https://tinyurl.com/y5orpf3l (last visited Nov. 25, 2020) CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 29

them, and are responsible for advancing the facts and argument entitling them to relief.’” (quoting Castro v. United States, 540 U.S. 375, 386 (2003) (Scalia, J., concurring in part and concurring in judgment))).

IV.

Although neither the 2010 nor 2012 Tiahrt Riders exempts the records CIR seeks, they nevertheless generally preclude the expenditure of funds to disclose any of the FTS database’s contents. 9 However, the Riders contain three exceptions to the stated funding prohibitions, and CIR contends that one of those exceptions applies. Specifically, it maintains that its request for “the total number of weapons traced back to former law enforcement ownership, annually from 2006 to the present” fits within Exception (C) of the Tiahrt Rider, which has been included in each Tiahrt Rider since 2008. Exception (C) provides that the Tiahrt Rider:

shall not be construed to prevent: . . . (C) the publication of annual statistical reports on products regulated by the [ATF], including total production, importation, and exportation by each licensed importer (as so defined) and licensed manufacturer (as so defined), or statistical aggregate data regarding firearms traffickers and trafficking channels, or firearms misuse, felons, and trafficking investigations.

9 In other words, under FOIA, the agency has a legal obligation to disclose the materials that CIR seeks, but it cannot fulfill that obligation using congressionally appropriated funds. See, e.g., Me. Cmty. Health Options, 140 S. Ct. at 1324. 30 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

123 Stat. at 3129. We agree that this exception applies in this case.

CIR’s request seeks statistical aggregate data. “Statistical” is defined as “of, relating to, based on, or employing the principles of statistics.” Merriam-Webster Online Dictionary. 10 “Statistics” in turn is defined as “a branch of mathematics dealing with the collection, analysis, interpretation, and presentation of masses of numerical data.” Id. 11 “Aggregate” refers to a summary form of information “formed by the collection of units or particles into a body, mass, or amount.” Id. 12 Given these definitions, ATF concedes that the number of firearms traced to each state annually, the numbers of each type of firearm recovered annually, and the top source states for firearms, are each examples of “statistical aggregate data” within the meaning of Exception (C). Similarly then, “[t]he total number of weapons traced back to former law enforcement ownership, annually from 2006 to the present” likewise reflects an aggregated statistic derived from an underlying set of data.

Moreover, ATF’s production of these documents to CIR will result in the “publication” of this data. Because the Tiahrt Rider does not define that term, we begin with that word’s plain meaning. The plain meaning of “publication” signifies “disclosure to the public, rather than the disclosure of information to another individual or corporation within the context of a business or professional relationship.” Integrated Genomics, Inc. v. Gerngross, 636 F.3d 853, 861

10 https://tinyurl.com/y2ydrvak (last visited Nov. 25, 2020)

11 https://tinyurl.com/yxgbvuny (last visited Nov. 25, 2020)

12 https://tinyurl.com/y4b4zfc8 (last visited Nov. 25, 2020) CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 31

(7th Cir. 2011) (collecting dictionary definitions); see also Oxford English Dictionary Online (“The action of making something publicly known.”). 13 In the legal context, the phrase generally suggests “the act of declaring or announcing to the public,” Black’s Law Dictionary (11th ed. 2019), or “[n]otification or communication to a third party or to a limited number of people regarded as representative of the public,” Oxford English Dictionary Online.

Turning over data regarding firearms in the United States to “a reporter” or “a representative of the news-media” like CIR, which reports on the topic of guns in the United States, will make that data “generally known” to the public. Indeed, the record reveals that the requested data will play a role in CIR’s upcoming “project” on gun violence. That ATF’s compliance with its FOIA obligation will thus result in the publication of the data CIR seeks is not surprising: news media organizations are precisely the sort of “representative[s] of the public” through which individuals and entities commonly distribute information to the public. Cf. Courthouse News Serv. v. Planet, 750 F.3d 776, 786 (9th Cir. 2014) (“We have observed that the news media, when asserting the right of access, are surrogates for the public.” (internal quotation marks and citation omitted)). Given that complying with its legal obligation under FOIA will make generally known the statistical aggregate data that CIR seeks here, the Tiahrt Rider authorizes ATF’s expenditure of funds to complete this request. 14

13 https://tinyurl.com/yxhe3p85 (last visited Nov. 25, 2020) 14 We do not share the dissent’s slippery-slope concerns about defining “who counts as a ‘representative of the news-media.’” Dissenting Op. at 55 n.8. After all, courts must already grapple with this 32 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

The government and dissent object, maintaining that “publication” refers only to the formalized distribution of prepared, formal information—though they reach that conclusion for different reasons. We turn first to the dissent, which notes that Congress used both the words “disclosure” and “publication” in the 2012 Rider and fears that our definition of publication subsumes the word “disclosure.”

“Disclosure” suggests “revealing new or secret information” or “the action of making something openly known.” Oxford English Dictionary Online 15; see also Black’s Law Dictionary (11th ed. 2019) (defining “disclosure” as “[t]he act or process of making known something that was previously unknown; a revelation of facts”). Thus, unlike publication, “disclosure” does not necessarily connote revealing information to the public at large. An agency can disclose information to a limited number of people or under conditions such that information is unlikely to spread amongst the public.

Of course, some disclosures may also effectively constitute publication. For example, if ATF discloses information to the entire public, whether in printed form or not, that information becomes generally known. Similarly, if ATF turns over information to a “representative of the public,” such as a reporter, it in effect makes that information generally known to the public. But such an equivalence is by no means a sure thing. If ATF sought to provide records to a limited number of people, who did not fairly represent

issue in the FOIA context. See 5 U.S.C. § 552(a)(4)(A)(ii); Cause of Action v. Fed. Trade Comm’n, 799 F.3d 1108, 1118–1125 (D.C. Cir. 2015).

15 https://tinyurl.com/y2wmpacy (last visited Nov. 25, 2020) CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 33

the public, the Tiahrt Rider’s “publication” exception would not permit the use of congressionally appropriated funds.

In this way, both “disclosure” and “publication” retain contextually distinct meanings in the Tiahrt Rider. As relevant here, the Rider first forbids the use of funds for “disclos[ing] part or all of the contents of the [FTS],” 125 Stat. at 609 (emphasis added), meaning that ATF cannot turn this material over to even a single person. Later, the Rider explains that this funding bar does not apply to “(A) the disclosure of statistical information concerning” the production, exportation, or importation of guns. Id. at 610 (emphasis added). In other words, ATF may use appropriated funds to reveal this information to whomever and however many people it likes, and appropriated funds are available regardless of how ATF exercises that discretion. Finally, the Rider permits the use of appropriated funds for “the publication of” certain “annual statistical reports . . . or statistical aggregate data.” Id. (emphasis added). ATF can thus use these funds to release these materials only if doing so would make these reports or data generally known to the public. 16

The plain meanings of these words also comport with FOIA’s use of those terms. FOIA does not explicitly define the terms “disclosure” or “publication.” See 5 U.S.C. § 551. But its use of the term “disclosure” clearly covers some situations where it envisions the wide dissemination of information to the public, see, e.g., id. § 552(a)(4)(A)(iii)

16 None of this means that ATF, specifically, is “publish[ing] information.” Dissenting Op. at 55. For though the Tiahrt Rider permits the use of funds to enable “publication” it never states that such “publication” must come at the direct hand of ATF. 125 Stat. at 610. The statute is agnostic in this regard. 34 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

(asking whether “disclosure” will “contribute significantly to public understanding”), and some situations when it does not, see, e.g., id. § 552(a)(8)(A)(i)(I). “Publication,” meanwhile, is always used in a manner that suggests widespread dissemination. See id. § 552(a)(2)(E); 552(a)(4)(A)(ii). Tellingly, FOIA also uses the term “publication” to connote more than “the act or process of publishing printed matter or an issue of printed material offered for distribution or sale.” Dissenting Op. at 53; compare 5 U.S.C. § 552(a)(2)(E) (discussing “publication” in the Federal Register) with id. § 552(a)(4)(A)(ii) (explaining “[a] freelance journalist shall be regarded as working for a news-media entity if the journalist can demonstrate a solid basis for expecting publication through that entity,” and that such entities include “television or radio stations broadcasting to the public at large” (emphasis added)). 17 These understandings of “disclosure” and “publication” thus accord with both the Tiahrt Rider’s context and FOIA’s use of those words.

Meanwhile, reading the word “publication” to reference only the “formalized, prepared release of information” because the Tiahrt Rider contemplates the “publication” of “reports,” Dissenting Op. at 53–54, is itself a misadventure in contextual analysis. Such a reading ignores that the Tiahrt Rider also permits publication of “statistical aggregate data.”

17 Though we therefore conclude that the use of the terms “disclosure” and “publication” in FOIA correspond with the meanings we discussed earlier, we view that fact as confirmatory, not determinative. The 2012 Rider is, after all, not a part of FOIA’s organic statute. And while it affects the funding available to ATF to comply with its FOIA obligations, it also speaks both to ATF’s funding for “disclosing” FTS data to law enforcement agencies and those entities’ abilities to disclose this information. See 125 Stat. at 609–10. It thus addresses a broader array of issues than FOIA. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 35

125 Stat. at 610. The word “data” does not carry an inference of formal dissemination. See Oxford English Dictionary (defining “data” as “information considered collectively, typically obtained by scientific work used for reference, analysis, or calculation”). 18 So even if the word “reports” standing alone could support reading “publication” in the way the dissent suggests, the full context of the Tiahrt Rider simply doesn’t lend itself to the dissent’s preferred meaning of “publication.” See S.D. Warren Co. v. Me. Bd. of Envt’l Prot., 547 U.S. 370, 379–80 (2006) (explaining that the contextual canon of “noscitur a sociis is no help absent some sort of gathering [of words] with a common feature to extrapolate”).

The government’s reading of the legislative history commits the same error. It notes that in discussing Exemption C, Congress stated:

[t]he Committee is concerned that the previous year’s language has been interpreted to prevent publication of a long-running series of statistical reports on products regulated by ATF. This was never the intention of the Committee, and the fiscal year 2008 language makes clear that those reports may continue to be published in their usual form as they pose none of the concerns associated with law enforcement sensitive information.

H.R. Rep. No. 110-240, at 63 (2007). But again, while this statement provides context for understanding Congress’s intention as to “statistical reports,” it casts no light on

18 https://tinyurl.com/yy7qdgjm (last visited Nov. 25, 2020) 36 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

Congress’s intention with regard to the release of “aggregate data.” 125 Stat. at 610. And the meaning of “publication” must account for both of the items that Congress listed in Exemption C.

ATF itself appears to have recognized in past cases that Exemption C’s use of the term “publication” encompasses more than formal statistical reports. For example, it has used appropriated funds to release aggregate trace data during litigation. See, e.g., Def.’s Br. in Opp’n to Pls.’ Mot. to Suppl. the Administrative Record at 5 n.2, Ron Peterson Firearms, LLC v. Jones, Civil No. 11-CV-678 JC/LFG, 2013 WL 12091518 (D.N.M. Mar. 27, 2013). After all, aggregate data does not “jeopardiz[e] criminal investigations and officer safety” or impinge upon “the privacy of innocent citizens.” H.R. Rep. No. 107-575, at 20 (2002); cf. H.R. Rep. No. 110-240, at 63 (noting that the disclosure of statistical reports “pose[s] none of the concerns associated with law enforcement sensitive information”). 19

V.

According to ATF, the court nevertheless cannot compel disclosure for a separate reason: ATF is not required to disclose under FOIA the “[t]otal number of weapons traced 19 If Exemption C did not apply, there could still be other ways to address this funding issue besides the use of congressionally appropriated funds. To begin, the record does not reveal the extent to which fulfilling CIR’s request would require the use of government funds. Moreover, we note that FOIA provides for the charging of fees “applicable to the processing of requests,” but those fees are limited to “reasonable standard charges for document duplication when records are not sought for commercial use and the request is made by . . . a representative for the news media.” 5 U.S.C. § 552(a)(4)(A). Finally, in some cases the requesting party has offered to pay for the disclosure of the requested records. See, e.g., Cal-Almond, 960 F.2d at 108. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 37

back to former law enforcement ownership, annually from 2006 to the present,” because FOIA establishes a right of access to existing agency records only, and searching its trace database would require the creation of a new record.

As CIR and amici recognize, whether a search query of an existing database entails the creation of a “new record” is a question of great importance in the digital age. “[D]atabase journalism is now fundamental to modern newsrooms,” and “exactly how journalists can request and use information from [government] databases . . . has the potential to make or break efforts to hold the government accountable using its own data.” Br. of Amici Curiae Five Media Organizations & Sixteen Data Journalists in Supp. of Appellant & Reversal (Media Orgs. Br.) at 10. Amici also explain that the number of government databases is ever expanding, as agencies continue to collect massive amounts of data about American society, which they store in electronic databases. Br. of Amicus Curiae Elec. Frontier Found. in Supp. of Pl.-Appellant at 1–3. Moreover, as in this case, “[r]eleasing statistical aggregate data from government databases” may sometimes prove the “only[] way to comply with FOIA’s mandate while properly balancing the public’s and the government’s interests in safeguarding sensitive information.” Id. at 2. Thus, if running a search across these databases necessarily amounts to the creation of a new record, much government information will become forever inaccessible under FOIA, a result plainly contrary to Congress’s purpose in enacting FOIA.

FOIA establishes a right of access to existing agency records only. See NLRB v. Sears, Roebuck & Co., 421 U.S. 132, 161–62 (1975). Although FOIA requires federal agencies to make “reasonable efforts to search for” the 38 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

records requested, 5 U.S.C. § 552(a)(3)(C), it does not require agencies to create new records, Kissinger v. Reps. Comm. for Freedom of the Press, 445 U.S. 136, 151–52 (1980); see also Inst. for Just. v. Internal Revenue Serv., 941 F.3d 567, 569 (D.C. Cir. 2019) (“FOIA imposes no duty on agencies to create new records in response to FOIA requests.”); Yagman v. Pompeo, 868 F.3d 1075, 1080–81 (9th Cir. 2017) (explaining that an agency is under no duty to simply answer questions under FOIA). However, “the burden is on the agency to demonstrate, not the requester to disprove, that the materials sought are not [currently existing] agency records.” U.S. Dep’t of Just. v. Tax Analysts, 492 U.S. 136, 142 n.3 (1989) (internal quotation marks omitted).

Against this backdrop, in 1996, Congress enacted the Electronic Freedom of Information Act Amendments of 1996 (E-FOIA) to update FOIA. Congress recognized that “FOIA face[d] a new challenge” as the federal government began storing and analyzing massive amounts of information on electronic networks and in electronic databases. See H.R. Rep. No. 104-795, at 11 (1996). So that “FOIA [may] stay abreast of these developments,” id. at 12, Congress amended the term “record” to include “any information that would be an agency record subject to the requirements of this section when maintained by an agency in any format, including an electronic format,” 5 U.S.C. § 552(f)(2). Thus, “computer database records are agency records subject to the FOIA.” H.R. Rep. No. 104-795, at 19 (1996); accord Inst. for Just., 941 F.3d at 571. And recognizing the malleability of digital data, E-FOIA also required that the agency “provide the record in any form or format requested by the person if the record is readily reproducible by the agency in that form or format.” § 5, 110 Stat. at 3050. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 39

E-FOIA also amended the definition of “search” to mean “to review, manually or by automated means.” Id. Congress acknowledged that “[c]omputer 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,” but emphasized that “the review of computerized records would not amount to the creation of records.” H.R. Rep. No. 104–795, at 22 (1996). Thus, E-FOIA codified a principle already established by the courts of appeal: “Although accessing information from computers may involve a somewhat different process than locating and retrieving manually-stored records, these differences may not be used to circumvent the full disclosure policies of the FOIA.” Inst. for Just., 941 F.3d at 571 (quoting Yeager v. Drug Enf’t Admin., 678 F.2d 315, 321 (D.C. Cir. 1982)).

Applying E-FOIA, courts have consistently held that database searches do not involve the creation of new records. See id. at 569. Moreover, district courts have held that sorting, extracting, and compiling pre-existing information from a database does not amount to the creation of a new record. See Long v. Immigr. & Customs Enf’t, No. 17-cv- 01097 (APM), 2018 WL 4680278, at *4 (D.D.C. Sept. 28, 2018) (“[N]either sorting a pre-existing database of information to make information intelligible, nor extracting and compiling data . . . as to any discrete pieces of information that [an] agency does possess in its databases, amounts to the creation of a new agency record.” (internal quotation marks omitted)); Nat’l Sec. Couns. v. Cent. Intelligence Agency, 898 F. Supp. 2d 233, 270 (D.D.C. 2012) (Nat’l Sec. Couns. I) (“[S]orting a pre-existing database of information to make information intelligible does not involve the creation of a new record.”); Schladetsch v. Dep’t of Hous. & Urb. Dev., 2000 WL 33372125, at *3 (D.D.C. Apr. 4, 2000) (“Because HUD has conceded that it possesses 40 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

in its databases the discrete pieces of information which [plaintiff] seeks, extracting and compiling that data does not amount to the creation of a new record.”).

We agree that using a query to search for and extract a particular arrangement or subset of data already maintained in an agency’s database does not amount to the creation of a new record. 20 In some ways, typing a query into a database is the modern day equivalent of physically searching through and locating data within documents in a filing cabinet. The subset of data selected is akin to a stack of redacted paper records. It makes no difference if the query produces a set of documents, a list, a spreadsheet, or some other form of results that the agency has not previously viewed. For one thing, “[a] request is not flawed simply because the agency has not anticipated it and preassembled the desired information.” Ferri v. Bell, 645 F.2d 1213, 1220 n.9 (3d Cir. 1981). Further, “[t]he fact that [the agency] may have to search numerous records to comply with the request and that the net result of complying with the request will be a document the agency did not previously possess is not unusual in FOIA cases nor does this preclude the applicability of the Act.” Disabled Off.’s Ass’n v. Rumsfeld, 428 F. Supp. 454, 456 (D.D.C. 1977), aff’d, 574 F.2d 636 (D.C. Cir. 1978); Schladetsch, 2000 WL 33372125, at *3 (applying this principal to electronic databases). So long as the relevant information and data fields already exist in the

20 We use the term “query” as defined by amici Five Media Organizations and Sixteen Data Journalists: “A query is an instruction that tells a database management system to select a specific subset of information from a database and return it in a particular arrangement.” Media Orgs. Br. at 15. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 41

database maintained by the agency, the result produced by a query is an existing record, regardless of the form it takes. 21

The nature of electronic databases firmly grounds this principal in common sense. Unlike paper documents, which present information in a largely fixed form, “databases store information in a highly structured format that is easily divided and recombined into a variety of arrangements.” Media Orgs. Br. at 24; see also id. at 12–13. Thus, as amici argue, an agency that stores information in a database creates “a multitude of different arrangements [of the data] . . . , each of which is in the agency’s possession or control.” Id. at 22 (internal quotation marks omitted). The agency has access to these different arrangements of data, and under E- FOIA, the public presumably has the same rights of access.

Were we to agree with ATF that the results of a search query run across a database necessarily constituted the creation of a new record, we may well render FOIA a nullity in the digital age. The federal government has increasingly recognized the importance of aggregate data, and, as amici again point out, uses this information in significant ways, implicating profound issues of public importance.

21 We reject the bright-line distinction some courts have made between producing “particular points of data” and producing a “listing or index” of a database. Nat’l Sec. Couns. I, 898 F. Supp. 2d at 271. It cannot be that some arrangements of data available through a query of a database are “records” created and obtained by an agency, while others are not. See Nat’l Sec. Couns. v. Cent. Intelligence Agency, 960 F. Supp. 2d 101, 160 n.28 (D.D.C. 2013) (Nat’l Sec. Couns. II) (calling the content-index distinction “legal hair-splitting” and “fraught with tension”). 42 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

ATF insists that CIR did not request trace records themselves, but statistical information about those records that does not already exist in the FTS database. The district court agreed, reasoning that ATF had not yet prepared and published a formal, annualized report on the number of firearms traced back to former law enforcement ownership. The absence of an annual statistical report does not end the inquiry, however, because CIR’s request does not ask for nor necessarily require the production of such a formal report. 22

CIR argues that ATF can obtain the information requested with a simple query using preexisting close-out codes to sort the FTS database. ATF concedes that the FTS database “includes close-out codes for each trace, including those related to law enforcement and government agencies” and that it could search the FTS database to identify the trace records involving traces back to former law enforcement ownership. Yet ATF admits it has not conducted a search of the FTS database in response to CIR’s request for statistical aggregate data.

ATF can theoretically respond to CIR’s request in at least two ways. First, it could search the FTS database for records tagged with the relevant close-out codes and produce the resulting traces or list of traces, with any necessary redactions, for CIR to tabulate. Although the 2012 Tiahrt Rider prohibits ATF from using appropriated funds to do so, the Rider is not a withholding statute for purposes of FOIA, 22 We reject ATF’s argument that complying with disclosure would require it to engage in a further review process after running a search. The fact that ATF voluntarily engages in a multi-step review process when preparing and publishing its own statistical reports does not require it to do the same in response to an otherwise proper FOIA request. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 43

and ATF does not contend that any of FOIA’s other limited exemptions apply. Second, ATF could produce the precise statistical aggregate data that CIR seeks, with no further counting or analysis required, if, for example, a query or queries for the relevant close-out codes produces a “hit count” reflecting the number of records involving a firearm traced to law enforcement, the number of matching records is contained in FTS metadata, or if the database produces an otherwise responsive result separate from the trace data itself. Because the Tiahrt Rider permits the disclosure of such statistical aggregate data, this second option would avoid any unauthorized use of funds.

Without a further understanding of the specifics of the FTS database, however, these are only theoretical possibilities. We have an insufficiently developed record from which to determine with any certainty whether the information CIR seeks could be produced by a reasonable search of the FTS database or would require more significant human analysis. The record evidence only generally describes the FTS database and does not describe its search functions or the form that the results of a query or search of the database will take. As a result, CIR can only speculate based on data that ATF produced in other proceedings that “the FTS database already appears to contain the responsive count” or that it otherwise contains responsive data. Likewise, amici can only surmise “to the best of its knowledge” that the FTS database “is built in Oracle, a relational database management system” and that it can analyze its capabilities based on “a typical relational database,” not evidence specific to how the FTS database itself is organized and functions. Media Orgs. Br. at 13–14. Because ATF bears the burden of justifying that records were properly withheld, Tax Analysts, 492 U.S. at 142 n.3, we remand to the district court to provide ATF the 44 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

opportunity to better explain the nature of the FTS database, and determine whether CIR’s search query will yield the responsive information it seeks.

VI.

For the reasons stated herein, we REVERSE the district court’s grant of summary judgment and REMAND for further proceedings. 23

BUMATAY, Circuit Judge, dissenting:

The Constitution provides that once legislation is approved by both houses of Congress and signed by the President, it becomes law. With today’s decision, the majority approves another requirement: that an act must also conform to “magical passwords” dictated by previous congresses. The majority also misconstrues federal law as requiring FOIA disclosures that Congress has expressly prohibited. For these reasons, I respectfully dissent.

Against the weight of precedent, the majority holds that the Tiahrt Amendment of 2012 must conform to an earlier statute—the OPEN FOIA Act of 2009—to be effective. As I explain below, this offends our constitutional scheme.

23 CIR’s motion for judicial notice is DENIED AS MOOT. Likewise, amicus Jack Jordan’s three motions for miscellaneous relief are DENIED AS MOOT. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 45

For a bill to become law, the Constitution’s sole requirements are bicameralism and presentment. See U.S. Const. art. I, § 7, cl. 2; see also I.N.S. v. Chadha, 462 U.S. 919, 951 (1983) (“It emerges clearly that the prescription for legislative action in Art. I, §§ 1, 7 represents the Framers’ decision that the legislative power of the Federal government be exercised in accord with a single, finely wrought and exhaustively considered, procedure.”). The Constitution imposes no requirement that new statutes must comply with past statutes. In other words, when passing laws, Congress is not bound by previous congresses. Chief Justice Marshall articulated this early on: “one legislature cannot abridge the powers of a succeeding legislature.” Fletcher v. Peck, 10 U.S. 87, 135 (1810); see also United States v. Winstar Corp., 518 U.S. 839, 873 (1996) (“[W]e have recognized that a general law . . . may be repealed, amended or disregarded by the legislature which enacted it, and is not binding upon any subsequent legislature[.]”) (simplified). Congressional enactments that attempt to bind subsequent congressional action are known as entrenchments. See John C. Roberts & Erwin Chemerinsky, Entrenchment of Ordinary Legislation: A Reply to Professors Posner and Vermeule, 91 Cal. L. Rev. 1773, 1777–78 (2003). Such legislative entrenchments cannot bind future congresses. As long as we are dealing with “general law enacted by the legislature”—and not “a constitutional provision”—the law “may be repealed, amended, or disregarded by the legislature which enacted it.” Manigault v. Springs, 199 U.S. 473, 487 (1905).

The prohibition on legislative entrenchment has ancient roots and stems from the fundamental nature of legislative power itself. See, e.g., 1 W. Blackstone, Commentaries on 46 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

the Laws of England 90 (1765) (“Acts of parliament derogatory from the power of subsequent parliaments bind not.”). As a result, members of the founding generation took the revocability of ordinary (non-constitutional) legislation as self-evident. The Virginia Statute for Religious Freedom of 1779—introduced by Madison and drafted by Jefferson— stated that “we well know that this Assembly, elected by the people for the ordinary purposes of legislation only, ha[s] no power to restrain the acts of succeeding Assemblies, constituted with powers equal to our own, and that therefore to declare this act irrevocable would be of no effect in law.” A Bill for Establishing Religious Freedom, 18 June 1779; 1 see also John O. McGinnis & Michael B. Rappaport, Symmetric Entrenchment: A Constitutional and Normative Theory, 89 Va. L. Rev. 385, 405 (2003) (Evidence of Madison’s public embrace of the “antientrenchment principle” “strongly suggests . . . [it] was widely accepted among the Framers’ generation.”). 2

Entrenchment also runs counter to the principles of our representative democracy. “Frequent elections are unquestionably the only policy by which” the legislature’s accountability to the People can be achieved. The Federalist No. 52, at 251 (James Madison) (David Wootton ed., 2003).

1 https://founders.archives.gov/documents/Jefferson/01-02-02-0132- 0004-0082 2 Some commentators suggest that entrenchment is also in tension with Article I’s Rulemaking Clause, which says that “[e]ach House may determine the Rules of its Proceedings.” U.S. Const. art. I, § 5, cl. 2. They argue that this clause, properly understood, represents a “powerful constitutional principle that effectively walls off the entire process of enacting legislation from outside scrutiny or control”—including the control of former congresses. Roberts & Chemerinsky, supra, at 1789– 95. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 47

Accordingly, each “election furnishes the electorate with an opportunity to provide new direction for its representatives.” Julian N. Eule, Temporal Limits on the Legislative Mandate: Entrenchment and Retroactivity, 1987 Am. B. Found. Res. J. 379, 404–05. Yet, this “process would be reduced to an exercise in futility were the newly elected representatives bound by the policy choice of a prior generation of voters.” Id.

Express-statement laws are no exception to this rule. See Marcello v. Bonds, 349 U.S. 302, 310 (1955); Dorsey v. United States, 567 U.S. 260, 274 (2012). Express-statement laws are a form of entrenchment: they require a later enacted law to expressly reference a prior law if it is to actually supersede that law. But express-statement laws cannot impose some sort of “recitation requirement” on future congresses. As Justice Scalia observed, “[w]hen the plain import of a later statute directly conflicts with an earlier statute, the later enactment governs, regardless of its compliance with any earlier-enacted requirement of an express reference or other ‘magical password.’” Lockhart v. United States, 546 U.S. 142, 149 (2005) (Scalia, J., concurring).

In Marcello, the Court held that the Administrative Procedure Act’s hearing provisions did not apply to deportation proceedings, notwithstanding the immigration statute’s failure to include an express statement of exemption as required by the APA. 349 U.S. at 310. The Court reasoned that “[u]nless we are to require the Congress to employ magical passwords in order to effectuate an exemption from the Administrative Procedure Act, we must hold that the present statute expressly supersedes the hearing provisions of that Act.” Id. 48 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

Similarly, in Dorsey, the Court found that a more recently enacted sentencing law impliedly repealed an earlier one, despite the later statute’s failure to comply with the express-statement requirement in the prior statute. 567 U.S. at 273–74. The Court concluded that “statutes enacted by one Congress cannot bind a later Congress, which remains free to repeal the earlier statute, to exempt the current statute from the earlier statute, to modify the earlier statute, or to apply the earlier statute but as modified.” Id. at 274. For the Court, the express-statement requirement was merely a “background principle of interpretation,” not a binding rule. Id.

Altogether, the weight of constitutional history and precedent show that where two statutes conflict, the later statute controls, regardless of attempts by past congresses to hobble the current legislature. As Hamilton stated, as “between the interfering acts of an equal authority, that which was the last indication of its will should have the preference.” The Federalist No. 78, at 468 (Alexander Hamilton) (Clinton Rossiter ed., 1961). Simply put, Congress is not bound by the dead hand of the past—at least not when it comes to statutory law.

Against that backdrop, this case is a straightforward one. The OPEN FOIA Act contains a legislative entrenchment: it says that, to be effective, any exemption from FOIA disclosure must “specifically cite[] to this paragraph [5 U.S.C. § 552(b)(3)(B)]” if enacted after the 2009 law. 5 U.S.C. § 552(b)(3)(B). The Act, thus, purports to prevent future congresses from passing FOIA exemptions without an express citation to “5 U.S.C. § 552(b)(3)”—in other words, without using the “magical password.” CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 49

The 2012 Tiahrt Amendment doesn’t contain any passwords, but still seeks to exempt certain information from disclosure. Subject to a few exceptions, it explicitly prevents any funds appropriated by Congress from being used to “disclose part or all of the contents of the Firearms Trace System [FTS] database” maintained by the ATF. Pub. L. No. 112-55, 125 Stat. 552, 609–10 (2011). As an appropriations rider, whatever release of information FOIA mandates, the Amendment blocks funding when it comes to the FTS database. 3

As is clear from their texts, the two laws conflict. Under the OPEN FOIA Act, the Tiahrt Amendment would not be a lawful exemption to FOIA’s disclosure rules since it doesn’t reference § 552(b)(3)(B). Conversely, the Tiahrt Amendment withdraws funding from any attempt to disclose FTS data under FOIA or any other law. Functionally, then, the OPEN FOIA Act says “disclose,” and the Tiahrt Amendment says “do not disclose.” The majority sides with the earlier OPEN FOIA Act rather than the Tiahrt Amendment, reasoning that Congress gave itself “explicit instructions” for how “to enact statutes that would exempt [FTS] data from disclosure.” Maj. Op. at 25 (quoting Everytown for Gun Safety Support Fund v. Bureau of Alcohol, Tobacco, Firearms & Explosives, 403 F. Supp. 3d

3 An appropriations rider is no little matter. Under the Appropriations Clause, “[n]o Money shall be drawn from the Treasury, but in Consequence of Appropriations made by Law.” U.S. Const. art. I, § 9, cl. 7. “This straightforward and explicit command means simply that no money can be paid out of the Treasury unless it has been appropriated by an act of Congress.” United States v. McIntosh, 833 F.3d 1163, 1174 (9th Cir. 2016) (simplified). Congress has given this clause criminal bite through the Antideficiency Act, which penalizes unauthorized government expenditures with hefty fines and imprisonment. See 31 U.S.C. §§ 1341(a)(1), 1350. 50 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

343, 354 (S.D.N.Y. 2019)). Congress’s instructions to its future self are not controlling, though. The only binding limitations on how a particular Congress can exercise its legislative power are those outlined in the Constitution. 4

Because these two statutes are in conflict, I would construe the OPEN FOIA Act’s express-statement rule as merely a “background principle of interpretation,” Dorsey, 567 U.S. at 274, and hold that the later-enacted Tiahrt Amendment controls.

The majority doesn’t meaningfully contest any of the foregoing analysis and mainly contends that the parties did not analyze the law as I have. But this criticism ignores our longstanding precedent that “we can affirm a ruling on any ground supported by the record, even if that ground is not asserted by the appellee.” Angle v. United States, 709 F.2d 570, 573 (9th Cir. 1983).

While it is true that we generally rely on the arguments advanced by the parties, see Maj. Op. at 28–29 (quoting United States v. Sineneng-Smith, 140 S. Ct. 1575, 1579 (2020)), we never abdicate our independent role in interpreting the law. If the parties don’t offer the correct reading of a particular statute, we are not bound to blindly follow their lead. Instead, as judges, our duty is to get the

4 If the Act instead premised future FOIA exemptions on lawmakers’ performance of the Cha Cha Slide on the Senate floor, surely the majority wouldn’t uphold such an “explicit instruction” as binding. What about a rule requiring a “supermajority” for an FOIA exemption? Or an explicit instruction that the OPEN FOIA Act can’t be repealed? Would the majority hold them as binding on future congresses? Surely not, but who can tell from their ruling today? CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 51

law right. See Kamen v. Kemper Fin. Servs., Inc., 500 U.S. 90, 99 (1991) (“When an issue or claim is properly before the court, the court is not limited to the particular legal theories advanced by the parties, but rather retains the independent power to identify and apply the proper construction of governing law.”). As Justice Ginsburg aptly stated, “a court is not hidebound by the precise arguments of counsel.” Sineneng-Smith, 140 S. Ct. at 1581. This principle applies even if the matter involves a “weighty issue of first impression.” Maj. Op. at 28. After all, judges are not like lemmings, following the parties off the jurisprudential cliff.

Here, the parties framed this case as to whether the Tiahrt Amendment or the OPEN FOIA Act governs. I believe the doctrine against legislative entrenchment answers that question. Ironically, so does the majority. For all the pages spent dissecting why the Tiahrt Amendment is not a FOIA withholding statute, the majority ends up at exactly the same place I do—the Tiahrt Amendment governs nonetheless. So, the majority’s holding on this score is in no conflict with my own; even if a statute is not a recognized exemption under the OPEN FOIA Act, a later-enacted law prevails.

Because the Tiahrt Amendment controls, the next question is whether it prohibits ATF from disclosing the information requested by the Center for Investigative Reporting (“CIR”). The Amendment prevents ATF from disclosing the contents of the FTS database save a few, specific exceptions. Pub. L. No. 112-55, 125 Stat. 552, 609– 10 (2011). One of those exceptions is “the publication of . . . [1] annual statistical reports on [the importation and manufacturing of] products regulated by [ATF] . . . or [2] statistical aggregate data regarding firearms traffickers 52 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

and trafficking channels, or firearms misuse, felons, and trafficking investigations.” Id.

CIR requests FTS data showing the total number of weapons traced back to former law enforcement ownership from 2006 to the present. It contends that ATF may disclose this information under the Tiahrt Amendment’s exception for “statistical aggregate data.” But that exception allows for the “publication of . . . statistical aggregate data,” not the FOIA disclosure of such data. Pub. L. No. 112-55, 125 Stat. 552, 609–10 (2011) (emphasis added). To the contrary, the disclosure of that data is explicitly prohibited by the main provision of the Amendment. See id. (“[N]o funds appropriated under this or any other Act may be used to disclose part or all of the contents of the Firearms Trace System database[.]”). That Congress used both “disclos[ure]” and “publication” in the Tiahrt Amendment indicates that the two terms mean different things. As our court has explained, “[i]t is a well-established canon of statutory interpretation that the use of different words or terms within a statute demonstrates that Congress intended to convey a different meaning for those words.” SEC v. McCarthy, 322 F.3d 650, 656 (9th Cir. 2003). Thus, distinguishing between “publication” and “disclos[ure]” is essential to this case.

Without statutory definitions, we look to the common, contemporary meaning of the words when enacted. See Food Mktg. Inst. v. Argus Leader Media, 139 S. Ct. 2356, 2362 (2019). Dictionaries define “disclose” as “expos[ing] to view . . . mak[ing] known” or “mak[ing] secret or new information known.” Pocket Oxford American Dictionary (2d. ed., 2008); Merriam-Webster’s Collegiate Dictionary (2d. ed., 2008). In contrast, “publication” references the release of prepared information usually in print or electronic CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 53

form. See Pocket Oxford American Dictionary, supra (defining “publish” as “to prepare and issue a book, newspaper, piece of music for public sale” or to “print something in a book, newspaper, or journal so as to make it generally known”). Thus, in common usage, “publication” means “the act or process of publishing printed matter” or “an issue of printed material offered for distribution or sale.” Merriam-Webster’s Collegiate Dictionary, supra; see also American Heritage Dictionary (5th ed., 2020) (defining “publication” as “[c]ommunication of information to the public, [as in] the publication of the latest unemployment figures”).

The context of the Amendment supports this plain- meaning interpretation. See Food & Drug Admin. v. Brown & Williamson Tobacco Corp., 529 U.S. 120, 133 (2000) (“It is a fundamental canon of statutory construction that the words of a statute must be read in their context and with a view to their place in the overall statutory scheme.”) (simplified). The Amendment permits “publication” of two specific materials: (1) “annual statistical reports” and (2) “statistical aggregate data.” 125 Stat. at 610. While “data” could be disseminated in formal and informal ways, the word “reports” commonly refers to a formal—i.e., published—distribution of prepared information. 5 Consequently, the most natural reading of “publication” in

5 See Report, Oxford English Dictionary Online, https://www.oed.com/view/Entry/162917? (“An evaluative account or summary of the results of an investigation, or of any matter on which information is required (typically in the form of an official or formal document), given or prepared by a person or body appointed or required to do so.”). 54 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

the statute refers to the formalized, prepared release of information by the ATF. 6

How Congress uses “disclosure” and “publication” in the FOIA context also supports this plain meaning construction. In FOIA itself, Congress repeatedly used “disclosure” to describe an agency’s direct release of information under the Act to a requester. See, e.g., 5 U.S.C. § 552(a)(8), (b)(3), (b)(6), (c)(1). FOIA’s uses of “publication” or “publish,” by contrast, unambiguously refer to the formal release of information to the public at large by the agency. For example, FOIA requires agencies to “publish” its general rules and procedures “in the Federal Register for the guidance of the public.” Id. § 552(a)(1). It similarly provides that an agency must provide certain information in electronic format “unless the materials are promptly published and copies offered for sale.” Id. § 552(a)(2). Thus, FOIA itself uses “publication” differently from “disclos[ure].” 7

6 To be sure, the word “publication” does have a broader meaning. For example, some dictionaries also define “publication” to mean the “[c]ommunication of information to the public.” Publication, Merriam- Webster’s Collegiate Dictionary, supra. Nevertheless, as discussed above, this broader meaning doesn’t fit into Congress’s specific use of the term in the Tiahrt Amendment. 7 The majority discounts these examples as non-determinative because, as it says, the Tiahrt Amendment is not a “part of FOIA’s organic statute.” Maj. Op. at 34 n.17. But, “courts generally interpret similar language in different statutes in a like manner when the two statutes address a similar subject matter.” United States v. Novak, 476 F.3d 1041, 1051 (9th Cir. 2007); see also Brown & Williamson Tobacco Corp., 529 U.S. at 133 (“[T]he meaning of one statute may be affected by other Acts.”). Accordingly, FOIA’s use of the same terms as the Tiahrt Amendment can inform the latter’s meaning. CENTER FOR INVESTIGATIVE REPORTING V. USDOJ 55

Although the majority acknowledges that “publication” means widespread dissemination to the public while “disclosure” means production only to another individual, it finds no problem in conflating the two. See Maj. Op. 30–31. That’s because, says the majority, the ATF’s “disclosure” of the data here will count as a “publication” since CIR intends to make that data public. Thus, according to the majority, ATF publishes information if it discloses such information to someone else who happens to be a “representative of the news-media,” who will then communicate it to the masses. Maj. Op. at 31. 8 This novel interpretation of “publication” apparently turns, not on ATF’s actions, but on the actions of the requesters who receive the information from ATF. This reasoning improperly shoehorns “disclosure” into the definition of “publication” and eviscerates the prohibition on funding in the Tiahrt Amendment. Every disclosure request for data is now a publication request so long as the requester claims an intention to disseminate the information widely. The majority thus permits a narrow, textually limited exception to circumvent the prohibition on disclosure itself. 9

8 What’s more, the majority doesn’t define who counts as a “representative of the news-media” or what amount of attenuation, if any, is too much for the majority’s definition of “publication.” For example, is a citizen journalist with a Twitter account a “representative of the news-media”? What if ATF gives the information to someone who then promises to give it to someone else who publishes it? Does that count? The majority’s analysis opens up a can of worms ripe for endless litigation. And it does so by missing the law’s simple command: it is ATF’s “publication” of the data, not the requester’s, that is permitted by the Tiahrt Amendment.

9 The majority believes FOIA justifies its analysis because it permits the “disclosure” of certain information without charge if it will “contribute significantly to public understanding.” See Maj. Op. at 33– 34 (quoting 5 U.S.C. § 552(a)(4)(A)(iii)). This subparagraph just 56 CENTER FOR INVESTIGATIVE REPORTING V. USDOJ

But the accurate interpretation of “publication of . . . statistical aggregate data” dooms CIR’s case. This exception refers to ATF’s publication of prepared, formal documents of aggregated statistics—not ad hoc responses to FOIA requests. Because the Tiahrt Amendment prohibits the type of disclosure sought by CIR, and no exceptions apply, the district court’s grant of summary judgment in favor of ATF should be affirmed.

As the majority observes, the discourse over guns, crime, and firearms regulation ignites passions across our country. CIR’s wish to further that public debate with the evidence from ATF may be laudable. CIR’s FOIA request may very well, as the majority surmises, advance an issue of public importance. But that a party comes before this court for pure-hearted reasons does not empower us to rewrite the law. Our duty always remains the same—to say what the law is. And here, Congress has spoken: the law prohibits disclosure of the information CIR seeks. Because the majority holds otherwise, I respectfully dissent.

acknowledges the obvious point that the “disclosure” of information to a requester can lead to its widespread dissemination. But this doesn’t turn the word “disclosure” into “publication.” Tellingly, this subparagraph of FOIA doesn’t use the word “publication” at all. Instead, it describes exactly what is happening here—the production of information to a news media entity that will distribute it to an audience— yet explicitly refers to that course of conduct as a “disclosure.” Accordingly, despite the majority’s reinvention of terms, the Tiahrt Amendment only permits “publication” of certain FTS data by the ATF and prohibits the release of any of the FTS information sought by CIR.