UNITED STATES DISTRICT COURT FOR THE DISTRICT OF COLUMBIA

OPEN SOCIETY INSTITUTE,

Plaintiff, Civil Action No. 19-3620 (RDM) v.

U.S. CITIZENSHIP & IMMIGRATION SERVICES,

Defendant.

MEMORANDUM OPINION AND ORDER

Open Society brings this action to challenge a determination by the United States

Customs and Immigration Service (“USCIS”) that Open Society did not qualify for an exemption

from the otherwise applicable annual quota on H-1B visas. Dkt. 11. That exemption applies to

“nonprofit research organizations,”

(g)(5)(B), defined under the relevant

regulations as a nonprofit “that is primarily engaged in basic research and/or applied research,”

(h)(19)(iii)(C). Open Society maintains that on over a dozen prior occasions

USCIS found that Open Society satisfied this standard but that in 2020 the agency reversed

course without sufficient explanation or sound reason. Dkt. 11 at 3–4 (Am. Compl. ¶¶ 9–10);

Dkt. 41-7 at 14 (CAR 789).

USCIS’s decision at issue in this case, according to Open Society, violated the

Administrative Procedure Act (“APA”),

et seq., for three reasons: (1) the

decision rests on a misinterpretation of the agency’s own regulations; (2) even accepting that

interpretation, the administrative record does not support the agency’s determination; and (3) the

1 agency failed adequately to explain the change in its position with respect to Open Society’s

status as a “nonprofit research organization.” Dkt. 11 at 4–5 (Am. Compl. ¶¶ 11–15). USCIS

now moves for summary judgment, Dkt. 23, and Open Society responds with its own cross-

motion for summary judgment, Dkt. 29. Open Society also moves for leave to supplement the

administrative record with evidence relating to USCIS’s adjudication of the prior petitions for

cap-exempt H-1B visas, which, Open Society argues, supports its APA claim in this case. Dkt.

30.

As explained below, the Court is unpersuaded by Open Society’s arguments. The Court,

accordingly, will GRANT USCIS’s motion for summary judgment, Dkt. 23, and will DENY

Open Society’s cross-motion for summary judgment, Dkt. 29, and motion to supplement the

administrative record, Dkt. 30.

I. BACKGROUND

A. Statutory and Regulatory Background

The Immigration and Nationality Act (“INA”),

et seq., “govern[s]

temporary work authorization” for noncitizens seeking to work in the United States. Overdevest

Nurseries, L.P. v. Walsh,

(D.C. Cir. 2021). At issue in this case is a specific

kind of temporary work authorization, “the H-1B visa program, which allows American

companies to employ aliens in certain ‘specialty occupations.’” Nat’l Ass’n of Mfrs. v. Dep’t of

Lab.,

(D.C. Cir. 1998) (quoting

(a)(15)(H)(i)(b)); see also

(h)(1)(ii)(B).

The INA “permits employers to petition for ‘H-1B’ nonimmigrant visas on behalf of”

noncitizen beneficiaries. Vision Builders, LLC v. USCIS, No. 19-CV-3159,

,

at *1 (D.D.C. Oct. 5, 2020). “To obtain an H-1B visa, the employer bears the burden of showing

2 [USCIS], which ‘makes the determination under the INA on whether to grant visa petitions,’ that

the proposed role is a ‘specialty occupation’ and that the beneficiary meets the role’s

requirements.”

(first quoting

, then quoting

). Access to

H-1B visas is limited, however, as the INA caps “[t]he total number of aliens who may be issued

visas or otherwise provided nonimmigrant status during any fiscal year” under the H-1B program

at 65,000.

(g)(1)(A).

This annual quota does not apply to those who fall into one of three categories.

§ 1184(g)(5). The first category covers individuals who are “employed (or ha[ve] received an

offer of employment) at an institution of higher education,” while the third includes those who

have “earned a master’s or higher degree from a United States institution of higher education.”

Id. § 1184(g)(5)(A), (C). The second category—the one at issue in this case—provides that the

65,000-person annual quota for H-1B visas, id. § 1184(g)(1)(A), does not apply to those who are

“employed (or ha[ve] received an offer of employment) at a nonprofit research organization or

governmental research organization,” id. § 1184(g)(5)(B).

Congress adopted this cap exemption for “nonprofit research organization[s]” in 2000,

see American Competitiveness in the Twenty-First Century Act of 2000,

,

, but left to the implementing agency—then, the Immigration and

Naturalization Service (“INS”)—the task of further defining that phrase. 1 Under the resulting

1 USCIS, the defendant in this action, see Dkt. 11, is now responsible for administering the H- 1B visa program. In 2002, Congress “abolished the Immigration and Naturalization Service and transferred its authority to the Secretary of Homeland Security and two divisions within the Department of Homeland Security: the Bureau of Immigration and Customs Enforcement and the Bureau of Citizenship and Immigration Services.” Fogo De Chao (Holdings) Inc. v. U.S. Dep’t of Homeland Sec.,

, 1131 n.2 (D.C. Cir. 2014). The Bureau of Citizenship and Immigration Services “was later renamed USCIS.” L.M.-M. v. Cuccinelli,

, 30 (D.D.C. 2020). 3 regulations, a “nonprofit research organization is an organization that is primarily engaged in

basic research and/or applied research.”

(h)(19)(iii)(C) (defining that term for

purposes of an exemption from certain H-1B visa application fees); see also

§ 214.2(h)(8)(iii)(F)(3) (incorporating this definition for purposes of the “[c]ap exemptions” for

H-1B visas). An organization qualifies as a “nonprofit,” in turn, if it is “defined as a tax exempt

organization under the Internal Revenue Code” or “has been approved as a tax exempt

organization for research or educational purposes by the Internal Revenue Service.”

(h)(19)(iii)(A)–(B); see also

§ 214.2(h)(8)(iii)(F)(3). “Basic research” and “applied

research” are defined as follows:

Basic research is general research to gain more comprehensive knowledge or understanding of the subject under study, without specific applications in mind. Basic research is also research that advances scientific knowledge, but does not have specific immediate commercial objectives although it may be in fields of present or potential commercial interest. It may include research and investigation in the sciences, social sciences, or humanities. Applied research is research to gain knowledge or understanding to determine the means by which a specific, recognized need may be met. Applied research includes investigations oriented to discovering new scientific knowledge that has specific commercial objectives with respect to products, processes, or services. It may include research and investigation in the sciences, social sciences, or humanities.

Id. § 214.2(h)(19)(iii)(C); see also id. § 214.2(h)(8)(iii)(F)(3).

In defining these terms, the INS “drew on generally accepted definitions of the terms as

well as definitions contained in the regulations of the Internal Revenue Service and the Small

Business Administration.” Petitioning Requirements for the H-1B Nonimmigrant Classification

Under

Fed. Reg. 10678, 10679 (Feb. 29, 2000). The INS “also

consulted with the Department of Labor and nonprofit and academic organizations for assistance

in developing the definitions.” Petitioning Requirements for the H-1B Nonimmigrant

Classification Under

Fed. Reg. 65657, 65658 (Nov. 30, 1998).

4 B. Factual and Procedural Background

Open Society describes itself as “a private operating research and grant-making

foundation that develops and implements a range of programs to foster the development of open

societies around the world.” Dkt. 11 at 7 (Am. Compl. ¶ 25). Open Society “periodically seeks

to employ foreign workers and makes regular use of various immigrant and nonimmigrant visa

types,” including the H-1B visa.

(Am. Compl. ¶¶ 36–37).

According to Open Society, it “has been able to obtain cap-exempt H-1B visas for many

years,” first under the exemption for “nonprofits with ties to institutes of higher education” and,

then, after “notice and comment rulemaking that tightened that exemption,” as a “nonprofit

research organization.” Dkt. 29-1 at 13; see also Dkt. 11 at 10–11 (Am. Compl. ¶¶ 38–46).

After Open Society “began filing H-1B petitions [] under the separate exemption” for “nonprofit

research organization[s],” USCIS granted twelve “petitions throughout 2018 without raising any

questions at all regarding [Open Society’s] claim to that cap exemption.” Dkt. 11 at 10 (Am.

Compl. ¶ 41); see also Dkt. 23-1 at 18 (noting that the agency “acknowledged that previous

petitions by [Open Society] claiming this cap exemption had been granted”). Open Society

argues that the evidence it submitted in support of all but one of those petitions “was much less

robust than the record” it submitted to in support of the petition at issue in this case. Dkt. 29-1 at

13–14. USCIS approved one 2018 petition, however, only after “requesting additional

information to show that Plaintiff was ‘primarily engaged’ in research,” as required by

.

. But when Open Society responded with “substantial additional evidence that

closely tracks the administrative record here,” USCIS approved that petition as well.

On February 28, 2019, Open Society filed the petition at issue in this case, seeking an H-

1B visa on behalf of an individual the organization sought to employ as an associate general

5 counsel. Dkt. 41-2 at 60 (CAR 70); see generally Dkt. 41-2 at 57–73 (CAR 67–83). Open

Society filed the petition with USCIS’s California Service Center, Dkt. 41-2 at 74 (CAR 125),

which responded with a request for additional evidence that Open Society was “a nonprofit

research or educational organization” and that it was “primarily engaged in basic and/or applied

research,” Dkt. 41-6 at 120 (CAR 756). Open Society submitted additional evidence on both

questions. See Dkt. 41-7 at 24–166 (CAR 799–940); Dkt. 41-8 at 1–193 (CAR 941–1135); Dkt.

41-9 at 1–180 (CAR 1136–1321).

On June 7, 2019, the California Service Center issued its decision, rejecting Open

Society’s reliance on the H-1B cap exemption for nonprofit research organizations. Dkt. 41-2 at

2–56 (CAR 1–55). The decision concluded that Open Society had failed to carry its burden of

showing that it was a qualifying nonprofit, Dkt. 41-2 at 15–17 (CAR 14–16), and research

organization, Dkt. 41-2 at 5–13 (CAR 4–12). But recognizing that the question was “unusually

complex or novel,”

(a)(1), the California Service Center certified the matter to

USCIS’s Administrative Appeals Office (“AAO”), Dkt. 41-2 at 21 (CAR 20). Rather than

participate in that appeal, however, Open Society filed suit in this Court, alleging that the Service

Center’s decision was final and appealable. Dkt. 1.

On February 27, 2020—before USCIS responded to the complaint in this case—the AAO

issued its decision. Dkt. 41-7 at 2–14 (CAR 777–89). In that decision, it disagreed with the

California Service Center’s conclusion that Open Society had failed to show that it is a qualified

nonprofit. Dkt. 41-7 at 5–6 (CAR 780–81). But, like the Service Center, the AAO concluded

that Open Society had failed to carry its burden of showing that it is “primarily engaged” in

research. Dkt. 41-7 at 6–14 (CAR 781–89). The AAO, accordingly, affirmed the California

Service Center’s decision rejecting Open Society’s petition and finding that its beneficiary was

6 “not eligible for the claimed cap exemption as an employee of a nonprofit research

organization.” Dkt. 41-7 at 2 (CAR 779).

As the AAO explained, the governing regulation limits the nonprofit-research-

organization exemption to organizations that are “primarily engaged in basic research and/or

applied research.” Dkt 41-7 at 6 (CAR 781) (quoting

(h)(19)(iii)(C)). The

phrase “primarily engaged,” however, is left undefined. To fill this gap, the AAO turned for

guidance to the dictionary, Supreme Court precedent, and IRS regulations employing a similar

locution. Dkt 41-7 at 6–8 (CAR 781–83). The goal, according to the AAO, was “to give [these]

words their ordinary meaning absent persuasive reasons to the contrary.” Dkt 41-7 at 6 (CAR

781).

Starting with the dictionary, the AAO observed that “[t]he word ‘primarily’ is defined in

pertinent part as ‘first of all: FUNDAMENTALLY, PRINCIPALLY.’”

(quoting Webster’s

Third New International Dictionary (3d ed. 2002)). The AAO further observed that the Supreme

Court had adopted a similar construction of “the term ‘primarily’” when interpreting the Internal

Revenue Code, noting that the term meant “‘of first importance’ or ‘principally’ in the ordinary

sense.”

(quoting Malat v. Riddell,

(1966)). And the AAO found yet

further support for this construction of the term in IRS regulations, which distinguish between “a

‘main or primary objective’” and “‘incidental or secondary objectives.’”

(quoting

(c)(3)-1(c)(3)(iv)). Drawing on “the common meaning of the word ‘primarily,’ the

Supreme Court’s long-held interpretation, and the intersection between the Department of

Homeland Security . . . and IRS regulations on an entity’s nonprofit status,” the AAO

“conclude[d] that the ordinary meaning of ‘primarily’ as it is used in 8 C.F.R.

7 § 214.2(h)(19)(iii)(C) is ‘principally and as distinguished from incidentally or secondarily.’”

Dkt. 41-7 at 7 (CAR 782).

Next, the AAO construed the word “engaged,” again relying on dictionary definitions to

begin the inquiry. See Dkt. 41-7 at 8 (CAR 783). As the AAO explained, one is “engaged” in an

activity when “OCCUPIED” or “EMPLOYED” in that activity. Id. (quoting Webster’s Third

New International Dictionary (2002)). The AAO also stressed that the regulations define a

“governmental”—as distinct from a “nonprofit”—“research organization” as an entity whose

“primary mission is either ‘the performance or promotion’ of research.” Id.. Because the

regulatory definition of a “nonprofit research organization,” in contrast, does not include a

reference to the “promotion” of research, the AAO reasoned that “engaged” must “denote direct

action.” Id.

The AAO, accordingly, concluded that a nonprofit organization is “primarily engaged” in

research if, and only if, it is “‘directly and principally’ engaged in research.” Id.

The AAO then applied this standard to Open Society’s petition and found that the

organization had failed to meet its evidentiary burden for several reasons. First, on the petition

Form I-129, Open Society “designated itself as an ‘Other Social Advocacy Organization,’”

which describes entities that “‘address issues, such as peace and international understanding;

community action (excluding civic organizations); or advancing social causes, such as firearms

safety, drunk driving prevention, or drug abuse awareness.’” Dkt. 41-7 at 9 (CAR 784). Second,

Open Society’s mission statement—submitted as part of its petition, Dkt. 41-2 at 86–88 (CAR

148–50)—described the organization’s mission as “seek[ing] to strengthen the law; “help[ing] to

shape public policies;” “implement[ing] initiatives;” “build[ing] alliances;” and “protecting and

improving the lives of people in marginalized communities.” Dkt. 41-7 at 9 (CAR 784). These

8 goals, the AAO reasoned, did “not support a finding that [Open Society] directly and principally

engages in research.” Id.

Third, the AAO concluded that three affidavits submitted by Open Society lacked

sufficient detail regarding the relationship between its research activities and the balance of Open

Society’s operations. The first came from Sharmila Mhatre, the Acting Co-Director of the Public

Health Program. See Dkt. 41-4 at 104–43 (CAR 411–50); Dkt. 41-5 at 1–14 (CAR 451–64).

Attached to that affidavit, the AAO wrote, were “six reports produced by the Public Health

Program staff” and published between 2007 and 2017. Dkt. 41-7 at 10 (CAR 785). These

reports were insufficient, according to the AAO, because they “were authored by a total of five

employees,” and “[s]ix published reports in a span of 10 years do not sufficiently establish that

the Public Health Program staff are principally engaged in research.” Id. Moreover, the AAO

found that some of “the reports and publications indicated that [Open Society]’s personnel

collaborated with colleagues from other organizations but the record does not establish whether

[Open Society]’s personnel were the primary writers of these articles or if they assisted other

authors.” Id. The second affidavit came from Erika Dailey, Open Society’s Senior Officer for

Research. Dkt. 41-5 at 17 (CAR 467). Although that affidavit stated that Dailey oversaw “some

55 streams of thematic and geographic applied research,” the AAO concluded that the affidavit

lacked “additional information o[n] the listed streams of research,” Dkt. 41-7 at 10 (CAR 785),

and failed to “provide sufficient context” for these research projects, “such as the personnel

required to work on the project, the budget allotted for the project, or the frequency of such

projects,” Dkt. 41-7 at 11 (CAR 786). Open Society’s third affidavit, submitted by Myka

Carroll, its Head of Research Services, Dkt. 41-5 at 46 (CAR 496), fared no better because, as

9 the AAO explained, it “failed to provide sufficient context” for Open Society’s research, Dkt 41-

7 at 11 (CAR 786).

Fourth, the AAO was unpersuaded by a separate list of 65 ongoing research projects from

Open Society’s Justice Initiative, Dkt. 41-5 at 61–69 (CAR 511–19), which the organization

submitted in response to a USCIS request for evidence, Dkt. 41-7 at 11 (CAR 786). As the AAO

explained, this list did “not establish the amount of time required to perform the listed projects or

the number of staff for each project,” nor did it show “what proportion of [Open Society]’s

organization is engaged in these projects.” Id. The AAO found another list of research

publications, Dkt. 41-5 at 70–73 (CAR 532–35), equally unpersuasive because those

publications—like those submitted from the Public Health Program—indicated that Open

Society’s personnel collaborated “with colleagues from other organizations” but did not

“establish whether [Open Society]’s personnel were the primary writers of these articles or if

they assisted other authors,” Dkt. 41-7 at 11–12 (CAR 786–87). One publication, for example,

“stated . . . that the report was written by an individual not employed by [Open Society]” and that

Open Society had merely “sponsored the research and writing of the report.” Dkt. 41-7 at 12

(CAR 787).

Fifth, the AAO noted that although Open Society’s evidence included tax forms listing

“research functions” as a part of its $186 million annual operating costs, the record did not

include any “research budgets, business plans, or other business metrics” that would enable the

AAO to determine how much of that operating budget Open Society spent on research. Dkt. 41-

7 at 12–13 (CAR 787–88). Noting that Open Society engages in both research and “awarding or

recommending grants to recipients who conduct research,” the AAO stressed that Open Society

had failed to “provide sufficient information . . . regarding the research functions for employees

10 [to show] that it is primarily engaged in research” and failed to indicate whether Open Society

engaged in its own research for the purpose of selecting “grantee[s].” Dkt. 41-7 at 12 (CAR

787).

Viewing the record as a whole, the AAO concluded that Open Society had “not

established by a preponderance of the evidence that it is primarily engaged in research.” Dkt.

41-7 at 13 (CAR 788). The agency wrote:

While the Director requested specific and relevant evidence in the [request for evidence], [Open Society] did not submit sufficient documents in response. Furthermore, [Open Society] did not provide a brief on certification to rebut or overcome the Director’s concerns. The record does not establish that [Open Society] is a nonprofit research organization and, therefore, the Beneficiary does not qualify for cap exemption as an H-1B nonimmigrant employed as a nonprofit research organization. See

(h)(19)(iii)(C).

While we acknowledge that [Open Society] is “focused on research— researching problems in the world, researching possible solutions for those problems, and researching how to implement those solutions,” the regulation at

(h)(19)iii)(C) defines a nonprofit research organization as one that is “primarily engaged” in research, which we interpret to mean directly and principally engaged in research. Based on the totality of evidence in the record, and considering its research activities in proportion to its other activities, we conclude that the record does not demonstrate that [Open Society] is directly and principally engaged in research. The research conducted by [Open Society] is incidental, or, at best, secondary to its principal activities: making grants to promote social, legal and economic reforms.

In a final section of its decision, the AAO considered USCIS’s prior approvals of Open

Society’s cap-exempt status. Dkt. 41-7 at 14 (CAR 789). As the AAO explained, Open Society

“historically applied for an exemption to the cap [based on] its affiliation with institutions of

higher education,” but in 2017, after USCIS narrowed that exemption, the organization switched

course, relying instead on the “exemption for [] nonprofit research organization[s].”

After

that, USCIS continued to approve Open Society’s petitions on this alternative ground.

But

11 the AAO concluded that it was not bound by those prior determinations because (1) Open

Society cited “no legal authority that would support a finding that this petition should be

approved based on prior approvals;” (2) the agency is “not required to approve applications or

petitions where eligibility has not been demonstrated, merely because of prior approvals that may

have been erroneous;” (3) “the AAO is not bound to follow a contradictory decision of a service

center;” and (4) “[m]ore importantly, if the prior approvals were premised on a similar death of

evidence, the approvals would have been issued in error.”

In short, even in the face of

numerous prior approvals, “USCIS does not have authority to confer an immigration benefit

when the petitioner fails to meet its burden of proof in a subsequent petition.”

After receiving the AAO’s decision, Open Society amended its complaint in this matter,

“request[ing] that this Court review the AAO’s adverse decision and reverse and remand to

USCIS with instructions to grant [Open Society’s] H-1B petition.” Dkt. 11 at 6 (Am. Compl.

¶ 17). Now before the Court are the parties’ cross-motions for summary judgment, see Dkt. 23;

Dkt. 29, along with Open Society’s motion to supplement the administrative record, Dkt. 30.

II. LEGAL STANDARD

The APA directs courts to “hold unlawful and set aside agency action” that is “arbitrary,

capricious, an abuse of discretion, or otherwise not in accordance with law,” as well as agency

actions taken “in excess of statutory jurisdiction, authority, or limitations, or short of statutory

right.”

(2). In an APA case, summary judgment “serves as a ‘mechanism for

deciding, as a matter of law, whether the agency action is . . . consistent with the APA standard

of review.’” Fisher v. Pension Benefit Guar. Corp.,

, 18 (D.D.C. 2020)

(quoting Cayuga Nation v. Bernhardt,

(D.D.C. 2019)). “[T]he district judge

sits as an appellate tribunal,” and “the entire case on review is a question of law.” Am.

12 Bioscience, Inc. v. Thompson,

(D.C. Cir. 2001) (quotation marks omitted).

When conducting this review, “it is black-letter administrative law” that the Court “should have

before it neither more nor less information than did the agency when it made its decision.” Hill

Dermaceuticals, Inc. v. Food & Drug Admin.,

(D.C. Cir. 2013) (per curiam)

(quotation marks omitted). The exceptions to this rule “are quite narrow and rarely invoked.”

CTS Corp. v. EPA,

(D.C. Cir. 2014).

III. ANALYSIS

A. Motion to Supplement the Administrative Record

The Court must first consider Open Society’s motion for leave to supplement the

administrative record. Open Society argues that the Court should consider “a handful of

additional documents” relating to Open Society’s other petitions for cap-exempt visas because

“these additional documents are necessary for the Court’s judicial review of this [m]atter.” Dkt.

30. at 3; see also Dkt. 30-1 (index of proposed supplemental administrative record). USCIS

objects that “[g]oing beyond the administrative record . . . is done only in exceptional cases” and

that this is not such a case. Dkt. 34 at 2 (alteration in original) (quoting Cape Hatteras Access

Pres. All. v. U.S. Dep’t of Interior,

(D.D.C. 2009)). Open Society “is

seeking to supplement the administrative record to establish a pattern of petitions being granted,”

USCIS continues, but here “the pattern of approval has already been established and admitted by

[USCIS]” and, thus, “supplementation is not needed.”

at 2–3. As explained below, USCIS

has the better of the arguments.

All agree that the default rule is that “[i]n cases brought under the APA, the Court’s

review is confined to the administrative record.” Ad Hoc Metals Coal. v. Whitman,

(D.D.C. 2002). This principle flows from the APA itself, which directs courts to

13 review agency actions based on “the whole record or those parts of it cited by a party,”

, meaning “those documents that were before the administrative decisionmaker,” Cape

Hatteras Access Pres. All. v. U.S. Dep’t of Interior,

(D.D.C. 2009).

The agency is charged with collecting those materials “that were before the agency at the time

the decision was made,” James Madison Ltd. v. Ludwig,

(D.C. Cir. 1996)

(quotation marks omitted), and its “submission and certification of [the] administrative record as

filed is entitled to a strong presumption of regularity,” Sara Lee Corp. v. Am. Bakers Ass’n,

(D.D.C. 2008). This Court’s review is limited to the administrative record as

compiled by the agency, save for “certain narrow exceptions.” Amfac Resorts, L.L.C. v. U.S.

Dep’t of the Interior,

(D.D.C. 2001).

Those “narrow exceptions,”

encompass two circumstances, a distinction the parties

elide in their briefing but that dictates the required analysis. First, a party may “request to ‘add

to the administrative record documents the agency considered’ when it made its decision, that is,

to complete the administrative record.” Ctr. for Biological Diversity v. U.S. Army Corps of

Engineers, No. 20-CV-103,

, at *8 (D.D.C. Sept. 22, 2020) (alterations

omitted) (quoting Pac. Shores Subdivision v. Army Corps of Engineers,

(D.D.C. 2006)). The question in those circumstances is whether “the materials in question

actually were before the agency at the time of its decision.” Bazzi v. Gacki, No. 19-CV-484,

, at *3 (D.D.C. Sept. 23, 2020). A motion premised on the ground that the

record fails to include all material that was before the agency requires the movant to “rebut[] the

presumption of administrative regularity and show[] that the documents to be included were,” in

fact, “before the administrative decisionmaker.” Ctr. for Biological Diversity,

, at *8 (alterations omitted) (quoting Pac. Shores Subdivision,

).

14 Second, a party may ask the Court to consider or “review . . . extra-record evidence”—

that is, “evidence outside of or in addition to the administrative record that was not necessarily

considered by the agency.” Pac. Shores Subdivision,

. A motion premised

on this ground “runs counter to the principle that judicial review of an administrative action is

just that—a review of the administrative action the agency took based on what was before it at

the time it acted”—and so “courts are leery of requests to consider extra-record evidence in APA

cases.” Ctr. for Biological Diversity,

, at *8. Where a party requests that the

Court consider extra-record evidence, “a more stringent standard applies,” Univ. of Colo. Health

at Mem’l Hosp. v. Burwell,

(D.D.C. 2015), which requires the party to

“demonstrate unusual circumstances justifying a departure from the general rule,” Oceana, Inc.

v. Ross,

(D.D.C. 2018) (alterations omitted) (quoting City of Dania Beach

v. FAA,

(D.C. Cir. 2010)). The D.C. Circuit has identified “at least three

instances,” Am. Wildlands v. Kempthorne,

(D.C. Cir. 2008), in which

consideration of extra-record evidence is appropriate: (1) “the agency deliberately or negligently

excluded documents that may have been adverse to its decision;” (2) “the district court need[s] to

supplement the record with background information in order to determine whether the agency

considered all of the relevant factors;” or (3) “the agency failed to explain administrative action

so as to frustrate effective judicial review,” James Madison,

(quotation marks,

citations, and alterations omitted). 2

2 This Court has sometimes articulated this test slightly differently. See Ctr. for Biological Diversity,

, at *9 n.6. Under the alternative articulation, consideration of extra-record evidence is appropriate where “the agency (1) acted in bad faith in reaching its decision, (2) engaged in improper behavior in reaching its decision, (3) failed to examine all relevant factors, or (4) failed to adequately explain its grounds for decision.” Oceana, Inc. v. Pritzker,

(D.D.C. 2015) (quoting Oceana, Inc. v. Locke,

(D.D.C. 2009)). But the primary difference is that this four-part test replaces the first 15 Open Society is not clear whether it seeks to complete the actual administrative record or

to introduce extra-record evidence. This is not unusual, as “[t]he question whether to admit

extra-record evidence is often confused with the question whether the administrative record is

complete, perhaps because the same phrase—‘supplementing the administrative record’—is

often used to refer to both.” Ctr. for Biological Diversity,

, at *8 (quotation

marks and citation omitted) (quoting Oceana, Inc. v. Locke,

(D.D.C.

2009)). But the clear implication of Open Society’s argument that “additional documents are

necessary for the Court’s judicial review,” Dkt. 30 at 3 (emphasis added), is that the evidence it

seeks to introduce lies “outside of or in addition to the administrative record,” Pac. Shores

Subdivision,

. Indeed, neither party argues that the “additional documents”

at issue were “before the administrative decisionmaker” when the agency considered the petition

underlying this litigation, Pac. Shores Subdivision,

. The Court will,

accordingly, construe Open Society’s motion as a request to introduce extra-record evidence.

Understood in this light, the motion fails because Open Society has failed to

“demonstrate unusual circumstances justifying a departure from the general rule against

considering extra-record evidence,” Ross,

(quotation marks and alteration

omitted). Open Society makes no effort to argue that USCIS “deliberately or negligently

excluded documents that may have been adverse to its decision” or that this Court “need[s] to

supplement the record with background information in order to determine whether the agency

considered all of the relevant factors,” James Madison Ltd.,

. Although Open

“unusual circumstance[],”Ross,

, described above—deliberate or negligent exclusion of adverse documents—with separate inquiries into the agency’s “bad faith” or “improper behavior,” Pritzker,

. And because Open Society does not accuse USCIS of bad faith or improper behavior, this difference is immaterial to the case at hand.

16 Society does, in fact, seek to introduce a kind of “background information,” it does not argue that

USCIS failed to “consider[] all of the relevant factors,” id.; rather, the argument Open Society

seeks to buttress with extra-record evidence is that USCIS applied those factors in a manner at

odds with its prior treatment of Open Society’s petitions, see, e.g., Dkt. 29-1 at 22.

Open Society maintains that the additional evidence illuminates the USCIS’s “pattern of

decisionmaking” and that it is “necessary to understand the context surrounding [USCIS’s]

action.” Dkt. 30 at 3. That argument most closely resembles the third category of

circumstances in which extra-record evidence is appropriate, where an agency “failed to explain

administrative action so as to frustrate effective judicial review,” James Madison Ltd.,

(quotation marks and alterations omitted). And Open Society is correct that the D.C.

Circuit has recognized that “it may be that a pattern of visa grants of sufficient magnitude could

obligate the agency to provide a ‘reasoned explanation for . . . treating similar situations

differently,’” but only if the claimant establishes that “the prior petitions were factually

equivalent.” Fogo De Chao,

(quoting ANR Pipeline Co. v. FERC,

(D.C. Cir. 1995)). So, Open Society argues, the prior petitions are needed to establish the

“pattern of visa grants” necessary to trigger USCIS’s obligation to offer a “reasoned

explanation” for its change of course,

with respect to Open Society’s status as a nonprofit

research organization, see Dkt. 30 at 2 (“Plaintiff challenges the denial decision in the instant

[m]atter” in part because “USCIS had previously approved its numerous other petitions, such

that Defendant’s denial would run counter to those prior approvals.”).

Open Society’s reliance on Fogo de Chao is misplaced for several reasons. Indeed, if

anything, Fogo De Chao weighs against considering extra-record evidence in this case. In Fogo

De Chao, the plaintiff, which operated Brazilian steakhouse restaurants, challenged the agency’s

17 determination that a particular chef did not possess “specialized knowledge” for purposes of a

visa application. 769 F.3d at 1129–30. Although the plaintiff argued that the agency had

approved prior petitions for other chefs holding the same position and “asserted that the prior

petitions were factually equivalent,” the plaintiff “never introduced any evidence corroborating

that assertion.” Id. at 1144. And because “nothing in the administrative record reveal[ed]

whether even a sampling of those cases involved factually and legally similar contexts,” the D.C.

Circuit could not “conclude that the [agency] in fact treated ‘similar situations differently.’” Id.

(quoting ANR Pipeline,

). The key, in other words, was that the plaintiff could not

put the agency to the task of providing a “reasoned explanation for . . . treating similar situations

differently,”

without raising the issue in the administrative process and proffering evidence

showing that the situations were, in fact, similar.

After Fogo De Chao was decided, those seeking visas for prospective employees were on

notice that, if they want to rely on a “a pattern of visa grants of sufficient magnitude” to argue

that USCIS is bound to follow that precedent or to offer a reasoned explanation for departing,

they need to “introduce . . . evidence corroborating that assertion” in the administrative process.

Here, Open Society had every opportunity to do so. It responded to a request for evidence

but did not introduce the evidence that it now asks the Court to consider; it failed to respond in

any way to USCIS’s invitation to participate in the AAO proceeding; and it never sought

reconsideration before the agency or asked for permission to supplement the administrative

record. In short, Open Society could have, but did not, provide USCIS with the opportunity to

consider the evidence and arguments upon which it now seeks to rely. That is not how judicial

review of the administrative process works.

18 But even beyond that difficulty, Open Society never explains how the extra-record

evidence that it seeks to introduce matters. Unlike in Fogo De Chao, the dispute in this case

does not turn on the specific facts of specific precedents; rather, the AAO acknowledged that

USCIS had previously approved petitions in which Open Society had argued that it qualified for

exemption from the H-1B visa cap based on its status “as a nonprofit research organization,”

Dkt. 41-7 at 14 (CAR 789); see also Dkt. 34 at 3, obviating any need to look outside the record

to “corroborat[e] [Open Society’s] assertions” that USCIS acted in a manner inconsistent with

prior practice, Fogo De Chao,

; see also Dkt. 29-1 at 44 (noting that the

agency’s “opinion explicitly acknowledged [this] track record” (citing Dkt. 41-7 at 14

(CAR789)). In Fogo De Chao, the factual question—whether a given chef possessed

“specialized knowledge”—could vary from one petition to the next, so that inconsistent results

alone were insufficient to prove “an actual shift in how factually similar petitions are disposed

of.”

. In this case, in contrast, both parties agree that the sole issue is whether

the same organization (Open Society) qualifies under the same standard (the “nonprofit research

organization” requirement). To be sure, Open Society is correct that USCIS has stopped short of

“admit[ting] in th[is] litigation that its prior approvals were based on nearly identical records.”

Dkt. 37 at 4–5. But what matters is what the agency said in the administrative decision that is

subject to judicial review, and, here, the AAO both recognized the existence of precedents and

did not attempt to distinguish them on their facts. Dkt. 41-7 at 14 (CAR 789); see also Dkt. 23-1

at 19 (“[T]he AAO relied on a thorough review of the evidence submitted by [Open Society] and

determined that [Open Society] failed to satisfy its burden.”). The Court can decide whether the

AAO’s reasons for departing from agency precedent withstand scrutiny without considering

extra-record evidence.

19 In sum, this is not one of the “rare occasions,” Ctr. for Biological Diversity,

, at *8, in which review of extra-record evidence is needed. The Court will,

accordingly, adhere to the “general rule against considering extra-record evidence.” Ross,

(quotation marks omitted). But as explained below, even if the Court were to

consider the extra-record evidence submitted by Open Society, it would make no difference to

the Court’s resolution of the pending cross-motions.

B. Cross-Motions for Summary Judgment

The Court turns next to the parties’ cross-motions for summary judgment. USCIS, for its

part, presses three arguments: (1) USCIS’s interpretation of the regulatory requirement that

“nonprofit research organizations” be “primarily engaged” in research is either correct based on

“the plain language of the governing regulations” or deserving of deference under Kisor v.

Wilkie,

(2019), Dkt. 23-1 at 15–16; (2) USCIS’s “decision is amply supported by

the record,” id. at 17; and (3) USCIS’s “decision is not unlawful merely because USCIS

previously granted similar petitions,” id. at 18. Open Society, in turn, argues that (1) “USCIS

misconstrued the relevant regulatory language,” and its construction of the regulation is not

entitled to deference, Dkt. 29-1 at 21, 36–40; (2) “USCIS violated the substantial evidence rule”

in its treatment of Open Society’s evidence, id. at 40–43; and (3) “USCIS acted in an arbitrary

and capricious manner through its unexplained departure from [Open Society’s] prior petitions,”

id. at 43–48.

The Court addresses each of these issues in turn.

1. “Primarily Engaged”

The first question turns on the proper interpretation of the requirement that a “nonprofit

research organization” must be “primarily engaged in basic research and/or applied research.” 8

(h)(19)(iii)(C); see also

§ 214.2(h)(8)(iii)(F)(3) (incorporating this definition

for purposes of the cap exemption). Although the regulation does not define what it means to be

“primarily engaged” in research, the AAO construed the phrase to mean that the organization

must be “principally”—as opposed to “incidentally or secondarily”—and directly “employed”

in—as opposed to merely “promot[ing]”—“the activity of research.” Dkt. 41-7 at 6–8 (CAR

781–83). As a result, the AAO needed to “evaluate the proportion of the organization’s activities

dedicated to research against its unrelated trade or business activities.” Dkt. 41-7 at 8 (CAR 783)

(quotation marks omitted). Under that test, an “unrelated business operation” should ordinarily

represent no more than “an incidental or secondary activity.” Id. But a business operation that

constitutes “a substantial part of the organization’s activities . . . may not detract from research

being the primary activity” if that operation is “in furtherance of the organization’s research

activities” and the organization is not “operated for the primary purpose of carrying out that

business operation.” Id.

Open Society objects to this interpretation of the governing regulation, insisting that

under a “‘plain meaning’ reading of the regulatory language . . . nonprofits can be ‘primarily

engaged’ in research if research is among the main focuses of the entity.” Dkt. 29-1 at 21. The

crux of this dispute, according to Open Society, is that the AAO’s interpretation “construes

‘primarily’ to mean more than 50%.” Dkt. 29-1 at 29; see also Dkt. 38 at 8 (criticizing USCIS’s

“‘over 50%’ notion”). The AAO had a specific “quantum of time and/or money in mind that

would constitute ‘primarily engaging’ in research,” Open Society argues, and that approach

conflicts with “the two best sources of understanding” that phrase—“the regulatory history of the

‘primarily engaged’ term” and the agency’s “own rulemaking regarding H-1B cap exemptions.”

Dkt. 29-1 at 28–29.

21 Although USCIS does not squarely address Open Society’s characterization of the

AAO’s interpretation, the Court notes at the outset that it is not at all clear the AAO had a

specific “quantum of time and/or money in mind” when it considered Open Society’s evidence.

Id. at 29. Indeed, the AAO explicitly stated that a business operation could represent “a

substantial part of the organization’s activities” so long as that operation was “in furtherance of

the organization’s research activities” and the organization is not “operated for the primary

purpose of carrying out that business operation.” Dkt. 41-7 at 8 (CAR 783). This language

suggests a less rigid—and less quantitative—approach than the one Open Society describes. The

Court’s obligation, of course, is to evaluate USCIS’s interpretation on its own terms—not to

resolve an interpretative dispute that arises for the first time in the parties’ briefs. The question,

accordingly, is whether USCIS erred by construing the regulatory requirement that a “nonprofit

research organization” is “primarily engaged in basic research and/or applied research,”

(h)(19)(iii)(C), to mean that the organization must be “principally” (as opposed to

“incidentally or secondarily”) and “directly engag[ed] in” (as opposed to merely “promoti[ng])

“the activity of research,” Dkt. 41-7 at 6–8 (CAR 781–83).

The parties have differing views about the level of deference, if any, due to the agency’s

reading of the governing regulation. USCIS maintains that “to the extent the regulations are

ambiguous, the [agency’s] interpretation of its regulations is reasonable and entitled to

deference,” Dkt. 23-1 at 16, while Open Society maintains that “the agency is due no deference

here,” Dkt. 29-1 at 21. Under Auer v. Robbins,

(1997), and Bowles v. Seminole

Rock & Sand Co.,

(1945), the Court must at times defer “to an agency’s

interpretation of its own ambiguous regulation,” Christopher v. SmithKline Beecham Corp.,

(2012). But “Auer deference” applies “only if a regulation is . . . genuinely

22 ambiguous, even after a court has resorted to all the standard tools of interpretation.” Kisor v.

Wilkie,

(2019). This is a high bar. The Supreme Court has warned against

“casually” “declaring [a] regulation ambiguous,” insisting that “a court bring all its interpretive

tools to bear before finding that to be so.”

. It is not enough that both parties are

capable of articulating colorable arguments that the “plain regulatory language supports their

case.”

(quotation marks omitted). Instead, courts “must make a conscientious effort to

determine, based on indicia like text, structure, history, and purpose, whether the regulation

really has more than one reasonable meaning.”

at 2423–24.

These principles preclude deference to the agency here, but only because the “text,

structure, history, and purpose,”

, of the disputed regulation unambiguously support

the AAO’s reading. 3 Open Society barely disputes the agency’s textual analysis of the phrase

“primarily engaged,” and for good reason: the agency’s interpretation of “primarily” as meaning

“principally,” as distinct from “incidentally or secondarily,” comports with the dictionary

definition of the term. Dkt. 41-7 at 6–7 (CAR 781–82) (“The word ‘primarily’ is defined in

pertinent part as ‘first of all: FUNDAMENTALLY, PRINCIPALLY.’” (quoting Webster’s Third

New International Dictionary (3d ed. 2002))). Indeed, construing the word “primarily” to mean

“principally” borders on the tautological. See Primarily, Merriam-Webster,

3 Even if the regulation was ambiguous, “[a]mbiguity” is “necessary but not sufficient . . . to afford deference,” Nat’l Lifeline Ass’n v. FCC,

(D.C. Cir. 2020), and the other necessary conditions are not present here. Most notably, for Auer deference to apply the “interpretation . . . must be the agency’s ‘authoritative’ or ‘official position.’” Kisor,

(quoting United States v. Mead Corp.,

, 257–59 & n.6 (2001) (Scalia, J., dissenting)). And USCIS concedes that the agency’s decision in this case “is a non-precedent decision” that “does not create agency guidance or practice.” Dkt. 23-1 at 16; see also Fogo De Chao,

(noting that AAO decisions are “expressly non-precedential” and are binding “only as between [the agency] itself and the [petitioner] to whom it was issued” (alteration in original) (quotation omitted)). 23 https://www.merriamwebster. com/dictionary/primarily (last visited Aug. 12, 2021) (defining

“primarily” as “for the most part: CHIEFLY”); Primarily, Cambridge Dictionary,

https://dictionary.cambridge.org/us/dictionary/ english/primarily (last visited Aug. 12, 2021)

(defining “primarily” as “mainly” and listing “principally” as a synonym). And the principal

reason Kisor insists on a searching analysis of a regulation’s text before resorting to agency

deference is to ensure that the agency “give[s] effect to the regulation’s plain meaning,” if any.

Statewide Bonding, Inc. v. DHS,

(D.C. Cir. 2020) (citing Kisor,

). Here, the AAO did just that, giving effect to the “plain meaning,”

of the term

“primarily.”

The AAO’s understanding of “engaged in” as “denot[ing] direct action,” Dkt. 41-7 at 8

(CAR 783), similarly tracks the ordinary meaning of the word and is further supported by the

regulation’s parallel definition of “a governmental research organization,” which distinguishes

between the “performance” and the “promotion” of basic or applied research,

(h)(19)(iii)(C) (emphasis added). Given that the regulation includes no reference to the

“promotion” of research in its definition of a “nonprofit research organization,”

the AAO

correctly inferred that such an organization must be “directly occupied or employed in research”

to qualify for the exemption, Dkt. 41-7 at 8 (CAR 783); see also Engaged, Webster’s Third New

International Dictionary (3d ed. 2002) (defining “engaged” as “occupied” or “employed”). The

Court, accordingly, has little difficulty concluding that the agency’s reading is supported by the

regulation’s “text [and] structure.” Kisor,

.

The regulatory “history,”

further buttresses USCIS’s interpretation. When the

Department of Labor “worked with the INS [to] develop[] the . . . definition[]” of “nonprofit

research organization,” it considered—and rejected—a suggestion that “nonprofit research

24 organizations engaged in substantial research should be covered . . . whether or not research is

the nonprofit’s primary purpose.” Labor Condition Applications and Requirements for

Employers Using Nonimmigrants on H-1B Visas in Specialty Occupations and as Fashion

Models,

, 80181–83 (Dec. 20, 2000); see also Petitioning Requirements for

the H-1B Nonimmigrant Classification under

Fed. Reg. 65657, 65658

(Nov. 30, 1998). The Department “concluded that the definition of nonprofit research

organization should be limited to organizations primarily engaged in research,” in part because

that definition was “most consistent with the statutory phrase ‘research organization.’” 65 Fed.

Reg. at 80184 (emphasis added). The AAO’s reasoning in this case—that an “unrelated business

operation” should be no more than “an incidental or secondary” activity, Dkt. 41-7 at 8 (CAR

783)—does little more than implement this logic.

In a rulemaking in 2016, moreover, USCIS considered—and rejected—a

recommendation that it “delet[e] the words ‘primarily’ and ‘primary’ in 8 CFR

214.2(h)(19)(iii)(C).” Retention of EB-1, EB-2, and EB-3 Immigrant Workers and Program

Improvements Affecting High-Skilled Nonimmigrant Workers,

, 82446

(Nov. 18, 2016). The agency explained that “[t]hese limitations . . . have been in effect for more

than a decade for purposes of the cap exemptions,” and “maintaining these longstanding

interpretations, which include the ‘primarily’ and ‘primary’ requirements, will serve to protect

the integrity of the cap and fee exemptions as well as clarify for stakeholders and adjudicators

what must be proven to successfully receive such exemptions.”

In pressing its competing view of the regulatory history, Open Society points to the same

2016 rulemaking. Dkt. 29-1 at 29–30. In addition to revisiting the definition of a “nonprofit

research organization,” USCIS “replaced the phrase ‘primary purpose’ with ‘fundamental

25 purpose’” under a different category of cap exemption—for nonprofits affiliated with institutions

of higher education—to clarify that “entities may qualify . . . even if they are engaged in more

than one fundamental activity, any one of which may directly contribute to the research or

education mission of a qualifying college or university.” 81 Fed. Reg. at 82444. But this history

does Open Society no favors. The fact that the agency broadened a related exemption by

replacing the word “primary” with “fundamental” renders Open Society’s more expansive

reading of “primarily required” substantially less, not more, plausible.

Open Society’s other historical evidence is no more persuasive. It focuses on USCIS’s

predecessor agency’s reference to IRS regulations when first defining “nonprofit research

organization.” Dkt 29-1 at 29. Because, according to Open Society, certain IRS regulations

permit a broader understanding of what it means to be “primarily” engaged in an activity, the

governing regulation is best read to permit similar flexibility. Id. (quoting

(c)(3)-1(c)(1)). Open Society is correct that during that rulemaking the INS drew from

“definitions contained in the regulations of the Internal Revenue Service.” 63 Fed. Reg. at

65658. And though it is true that the AAO looked to an IRS regulation in its decision in this

proceeding, see Dkt. 41-7 at 6 (CAR 781), nothing in that original rulemaking or the AAO’s

decision supports Open Society’s contention that “primarily” means a “fundamental” or

“important” part, much less that the AAO committed a “massive conceptual” error. Dkt. 29-1 at

28–31. Rather, the INS (and, later, the AAO) looked to the IRS regulations as only one source of

guidance, after concluding that the definition of “nonprofit” should be drawn from “generally

accepted definitions of the term,” 63 Fed. Reg. at 65658, a proposition that is difficult to square

with Open Society’s contention that the regulation uses the phrase “primarily engaged” as a

“term of art” divorced from common usage, Dkt. 29-1 at 31–32.

26 But even if the Court focuses on the cited IRS regulation, which includes more expansive

language than found in the USCIS regulation, see, e.g.,

(c)(3)-1(c)(1) (“An

organization will not be so regarded if more than an insubstantial part of its activities is not in

furtherance of an exempt purpose.”), that test does not differ in material respects from the test the

AAO applied. Under the Internal Revenue Code, a Section 501(c)(3) organization must be

“organized and operated exclusively for” one of the enumerated exempt activities.

(c)(3). The regulation on which Open Society relies, in turn, provides that “[a]n

organization will be regarded as operated exclusively for one or more exempt purposes only if it

engages primarily in activities which accomplish one or more of such exempt purposes specified

in section 501(c)(3).”

(c)(3)-1(c)(1). “An organization will not be so regarded

if more than an insubstantial part of its activities is not in furtherance of an exempt purpose.”

The triple negative in this sentence may cause the reader momentary pause, but all that the IRS is

saying is that an organization is ineligible for the exemption if a substantial portion of its

activities are for a nonexempt purpose. That is almost exactly how USCIS articulated the

applicable test here. See Dkt. 41-7 at 7 (CAR 783) (providing that an “unrelated business

operation” must ordinarily represent no more than an “incidental or secondary activity”). As a

result, Open Society’s protestations aside, the regulatory history supports USCIS’s reading.

That leaves the regulatory “purpose,” Kisor,

, which—although

contested by Open Society—fails to militate strongly in either direction. Open Society argues

that the AAO erred by making “unfounded contentions regarding Congressional intent with

respect to the H-1B cap exemptions.” Dkt. 29-1 at 22. But all the agency said is that it “view[s]

the grant of a cap exemption . . . as a matter of legislative grace” in light of “how narrowly

Congress tailored the section 214(g)(5) exemptions to only three types of organizations within

27 the broader statutory scheme to limit the number of initial H-1B visas,” and in light of the goal of

protecting “the wages and working conditions of U.S. workers.” Dkt. 41-7 at 7 (CAR 782).

That argument turns on the words and structure of the statute rather than any unsupported

speculations about legislative intent. Consistent with this “general legislative purpose,”

discerned from the text and structure of the statute, the AAO “narrowly interpret[ed] the

requirement that research organizations be ‘primarily engaged’ in basic or applied research.”

Open Society objects that “[i]f Congress wished for the ‘nonprofit research organization’

exemption in particular to be narrow, it provided no overt indication.” Dkt. 29-1 at 23. And,

Open Society points out, another of the cap exemptions authorizes up to 20,000 additional H-1B

visas, nearly “a third” of the otherwise-applicable cap of 65,000 visas.

(citing

(g)(5)(C)). Based on this statutory language and its review of the legislative history, Open

Society maintains that the AAO’s understanding of congressional intent is unfounded.

at 23–

24. But that contention disregards the AAO’s actual reasoning, which sensibly inferred from

Congress’s decision to impose a statutory cap on H-1B visas, and then authorize three “narrowly

. . . tailored” exemptions, that Congress did not intend for the exemptions to overwhelm the

“numerical limitations in section 214(g)(1)(A).” Dkt. 41-7 at 7 (CAR 782). And, as USCIS

notes, Open Society offers no evidence that Congress intended a different result.

There is at least some evidence, moreover, to support the AAO’s conclusion that

Congress did not intend for the INS or USCIS to broadly construe the exemptions. Dkt. 41-7 at

7–8 (CAR 782–83). When Congress considered the law that first created those exemptions—the

American Competitiveness and Workforce Improvement Act of 1998,

Stat. 2681—that bill “met strong opposition in the House of Representatives from traditionally

pro-labor Democrats.” Jung S. Hahm, Note, American Competitiveness and Workforce

28 Improvement Act of 1998: Balancing Economic and Labor Interests Under the New H-1b Visa

Program,

, 1684 (2000) (collecting and summarizing the legislative

history). It was only “[a]fter months of wrangling and intense negotiations,” that “the White

House and the congressional supporters of the bill finally reached a compromise,” and even then

the final legislation “includ[ed] additional protective measures for American workers.”

. To be sure, much of this debate focused on the legislation’s provisions that “raise[d] the

H-1B visa cap,”

rather than the exemptions. But this history at least suggests that the AAO

was not far from the mark when it read its own regulation in a manner intended to cohere with

the intent of the statute. And, more importantly, the AAO’s decision did not rest on this

assessment of congressional intent alone. Indeed, as the AAO stressed, its reading of the

regulation does just what USCIS did in 2016, when it rejected proposals “to remove the term[]”

“primarily” from the regulation. Dkt. 41-7 at 7 (CAR 782).

The Court, accordingly, concludes that the regulation at issue here does not “ha[ve] more

than one reasonable meaning,” Kisor,

, and that the correct meaning is the one

embraced by the AAO. Although that conclusion precludes Auer deference, no such deference is

needed to uphold the AAO’s correct interpretation of an unambiguous regulatory term.

2. USCIS’s Evaluation of Open Society’s Evidence

Open Society next argues that “USCIS committed numerous violations of the substantial

evidence rule in casting aside Plaintiff’s evidence.” Dkt. 29-1 at 40. USCIS responds that the

AAO’s “decision is amply supported by the record and the evidence supplied by” Open Society.

Dkt. 23-1 at 17. As explained below, the Court is unconvinced by Open Society’s argument.

As an initial matter, the parties disagree about the proper test. Open Society urges the

Court to apply the “substantial evidence rule,” Dkt. 29-1 at 40, while USCIS argues that “the

29 higher legal standard under the ‘substantial evidence’ rule should not be applied to this case,”

Dkt. 35 at 17. That rule, USCIS points out, “comes from

(2)(E),” which directs

courts to “‘hold unlawful and set aside’” agency decisions that are “‘unsupported by substantial

evidence in a case subject to sections 556 and 557 of this title or otherwise reviewed on the

record of an agency hearing provided by statute.’”

(quoting

(2)(E)). And

because the AAO’s decision was not “subject to sections 556 or 557” of title 5, and because

“there is no statute providing for a hearing in visa petition adjudications,” USCIS contends that

the substantial evidence rule is inapplicable.

at 17–18. “[I]nstead,” according to USCIS, “the

appropriate legal standard for review of this case is whether the final agency decision was

‘arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.’”

at 17

(quoting

(2)(A)).

Under D.C. Circuit precedent, this debate is of little consequence. As the D.C. Circuit

has observed, “[w]hen the arbitrary or capricious standard is performing that function of assuring

factual support, there is no substantive difference between what it requires and what would be

required by the substantial evidence test.” Ass’n of Data Processing Serv. Orgs., Inc. v. Bd. of

Governors of Fed. Rsrv. Sys.,

, 683–84 (D.C. Cir. 1984). This is because “it is

impossible to conceive of a ‘nonarbitrary’ factual judgment supported only by evidence that is

not substantial in the APA sense.” Id. at 684. For this reason, “in those situations where

paragraph (E)” and its substantial evidence test “has no application,” the arbitrary and capricious

standard provided for in “paragraph (A) takes up the slack, so to speak, enabling the courts to

strike down, as arbitrary, agency action that is devoid of needed factual support.” Id. at 683.

Because “the arbitrary and capricious standard,” at least in this context, “does not

substantively differ from the substantial evidence test,” Phoenix Herpetological Soc’y, Inc. v.

30 U.S. Fish & Wildlife Serv.,

(D.C. Cir. 2021), the question before the Court

under either standard is whether a “reasonable mind might accept a particular evidentiary record

as adequate to support a conclusion,” Dickinson v. Zurko,

, 162 (1999) (quotation

marks omitted); see also, e.g., Verizon v. FCC,

, 643–44 (D.C. Cir. 2014);

Environmental Integrity Project v. McCarthy,

(D.D.C. 2015). An

“agency’s factual findings may be supported by substantial evidence even though a plausible

alternative interpretation of the evidence would support a contrary view.” Dillmon v. Nat’l

Transp. Safety Bd.,

(D.C. Cir. 2009) (quotation marks omitted). The Court

must, accordingly, affirm the AAO’s treatment of the record if, based on the evidence before it,

“the agency could fairly and reasonably find the facts as it did.” Chritton v. Nat’l Transp. Safety

Bd.,

, 856 (D.C. Cir. 1989) (quotation marks and alterations omitted).

The Court is persuaded that the AAO’s decision satisfies that standard. The AAO

concluded that Open Society’s evidence failed to meet its burden of showing that it was

“principally” (as opposed to “incidentally or secondarily”) and “directly” engaged in the activity

of research. Dkt. 41-7 at 6–8 (CAR 781–83). In reaching that conclusion, the AAO considered,

among other things, Open Society’s self-classification on its H-1B visa petition, Dkt. 41-7 at 9

(CAR 784); the organization’s mission statement, id.; three affidavits regarding the

organization’s research efforts, Dkt. 41-7 at 10–11 (CAR 785–86); a list of ongoing research

projects, Dkt. 41-7 at 11–12 (CAR 786–87); and the organization’s operating budget, as

evidenced by tax forms, Dkt. 41-7 at 12 (CAR 787). Although this evidence showed that Open

Society engaged in some research activity, USCS concluded, “[b]ased on the totality of the

evidence in the record, and considering its research activities in proportion to its other activities,”

31 that “the record [did] not demonstrate that [Open Society] is directly and principally engaged in

research.” Dkt. 41-7 at 13 (CAR 788).

That is a more than a “fair[] and reasonabl[e],” Chritton, 888 F.2d at 856, assessment of

the record, and, indeed, Open Society points to no evidence contradicting the AAO’s central

conclusion: that Open Society had failed to carry its burden of showing that, when considered “in

proportion to its other activities,” Open Society is “principally” engaged in research that it

“directly” conducts. Dkt. 41-7 at 13 (CAR 788). Instead of meaningfully engaging on this issue,

Open Society asserts that a “full accounting of the panoply of [the agency’s] errors would require

a separate round of briefing.” Dkt. 29-1 at 41. But USCIS has filed its own motion for summary

judgment, and to avoid an adverse decision on that motion, Open Society must show that the

AAO violated the APA. In the present context, that means that Open Society must show that a

“reasonable mind” could not find the existing administrative “record . . . adequate to support

[USCIS’s] conclusion,” Dickinson, 527 U.S. at 162 (quotation marks omitted); see also, e.g.,

Fulbright v. McHugh,

(D.D.C. 2014). The few arguments that Open

Society makes do not satisfy that demanding standard.

Open Society first argues that it “submitted a handful of published reports from its Public

Health Program staff as evidence that [it is] engaged in that type of research,” yet despite its

qualification that these material were just a “‘partial selection’” of materials, the AAO

improperly “concluded that only the five individuals who authored the [six] specific reports were

engaged in research.” Dkt. 29-1 at 41 (emphasis removed). But that contention misunderstands

the burden that Open Society bore in the administrative proceeding—it was not the AAO’s

burden to disprove the organization’s qualification. And, even if the AAO should have

reasonably inferred that there were others who conducted research on behalf of Open Society,

32 the AAO was still left with an evidentiary vacuum beyond the limited and inconclusive

information Open Society provided. See, e.g., Chippewa Dialysis Servs. v. Leavitt,

(D.C. Cir. 2007) (providing that courts “set aside” an agency’s “factual findings only if

[they are] unsupported by the record as a whole” (emphasis added)).

In addition, Open Society takes issue with the AAO’s findings because the agency

“discounted one of those reports” on the ground that the “acknowledgements section” thanks

“various individuals,” who appear to work at other organizations, “for research assistance and

comments on the draft report.” Dkt. 29-1 at 41; Dkt. 41-7 at 10 (CAR 785). Open Society, for

its part, does not take issue with the AAO’s premise that some of those individuals work at other

organizations. The AAO’s premise was nonetheless problematic in Open Society’s view

because it ignores the reality that “researchers at one organization . . . typically collaborate with

researchers at another,” and the AAO’s approach “would rob both organizations of any claim to

doing research unless they kept detailed time records.” Dkt. 29-1 at 42 (emphasis removed).

But that vastly overstates what the AAO said. The agency merely provided this

acknowledgement as an example of the difficulty in discerning from the face of the reports who

did what. To be sure, the Court might not have singled out the acknowledgment as the best

example of the lacuna in Open Society’s petition. But the Court’s review of the agency’s

factfinding is highly deferential, see Dillmon,

; Chritton, 888 F.2d at 856, and

the AAO’s point—even if conveyed in less than ideal terms—remains valid: “The publications

do not sufficiently establish that [Open Society] is directly and principally engaged in research,”

and Open Society did not otherwise “provide sufficient information for a comparison of the time

and resources spent on the published articles to the rest of the activities and operations” of the

organization. Dkt. 41-7 at 10 (CAR 785).

33 Relatedly, Open Society argues that “no amount of evidence would possibly have

convinced USCIS that [Open Society’s] public health division was engaged in research.” Dkt.

29-1 at 42. That, of course, misapprehends the relevant inquiry—the question is not whether

Open Society was “engaged in research,” but whether it was “directly and principally engaged in

research.” Dkt. 41-7 at 8 (CAR 783). But assuming that Open Society means that no amount

evidence would have persuaded the AAO under that test, its argument still fails. The

organization seems to maintain that it was doomed to fail because the reports that it submitted

thank “collaborator[s] outside of” the organization, and because it submitted many reports to no

avail. Dkt. 29 at 42. What Open Society fails to address, however, is the AAO’s conclusion that

it lacked sufficient evidence to gauge what proportion of the organization’s efforts—however

measured—were devoted to research, as opposed to grant making “to promote social, legal and

economic reforms.” Dkt. 41-7 at 13 (CAR 788). Although evidence of the type that Open

Society offered is undoubtedly probative, what it did not do is place that evidence in context with

more definitive affidavits or declarations from knowledgeable Open Society personnel

addressing Open Society’s overall activities and how its research activities fit within that broader

picture.

Open Society comes closer to mark, but still fails, in arguing that the AAO failed “to

consider or credit swathes of evidence together, such as the listing of research products contained

in a sworn affidavit.” Dkt. 29-1 at 42. In evaluating that affidavit and its attachments, the AAO

commented that Open Society “provided copies of six reports produced by the Public Health

Program staff,” Dkt. 41-7 at 10 (CAR 785), while the affidavit identifies “four recent reports

produced by Public Health Program [] staff,” along with six “peer-reviewed academic journals

and book chapters” and five “book chapters and other similar publications,” Dkt. 41-4 at 104–06

34 (CAR 411–13). Further complicating matters is the fact that the record includes some, yet not

all, of the publications listed in that affidavit. See Dkt. 41-4 at 107–43 (CAR 414–50); Dkt. 41-5

at 1–14 (CAR 450–64). But even if the AAO overlooked (or failed to discuss) some of the

research publications referenced in, or attached to, the affidavit submitted by the Public Health

Program’s Acting Co-Director, see CAR Dkt. 41-4 at 104–06 (411–13), any confusion as to

precisely how many research publications Open Society submitted or referenced is of no

moment. “In administrative law, as in federal civil and criminal litigation, there is a harmless

error rule.” Nat’l Ass’n of Home Builders v. Defs. of Wildlife,

, 659–60 (2007)

(quoting PDK Lab’ys Inc. v. DEA,

(D.C. Cir. 2004)). This means that “[i]f

the agency’s mistake did not affect the outcome, if it did not prejudice the petitioner, it would be

senseless to vacate and remand for reconsideration.” PDK Lab’ys Inc.,

.

Here, the AAO’s findings turned not on the precise number of research projects listed in

the affidavit or that Open Society submitted along with its petition, but on the proportion of those

projects to the balance of the organization’s activities. The AAO concluded its analysis of Open

Society’s Public Health Program by noting that Open Society had “not provide[d] sufficient

information for a comparison of the time and resources spent on the published articles to the rest

of its activities and operations.” Dkt. 41-7 at 10 (CAR 785) (emphases added). Similarly, the

AAO faulted another piece of evidence—Open Society’s list of research projects from its Justice

Initiative—because that list “d[id] not sufficiently establish what proportion of Petitioner’s

organization is engaged in these projects.” Dkt. 41-7 at 11 (CAR 786) (emphasis added).

Because the agency’s analysis focused on the comparison of Open Society’s research work to its

non-research activities, the distinction between six publications and fifteen (or more) is—without

further context—immaterial to the AAO’s conclusion. “Thus, the agency’s mistake, if any, ‘did

35 not affect the outcome’ nor ‘prejudice the petitioner,’ and ‘it would be senseless to vacate’ the

[AAO’s] decision.” Stand Up for California! v. U.S. Dep’t of Interior,

, 319

(D.D.C. 2016) (quoting PDK Lab’ys Inc., 262 F.3d at 799).

Finally, Open Society argues that USCIS’s treatment of its evidence “obviate[s] the broad

regulatory definition of ‘research’ in the regulation it construes.” Dkt. 29-1 at 32. “A nonprofit

research organization is an organization that is primarily engaged in basic research and/or

applied research,”

(h)(19)(iii)(C), and, Open Society maintains, the regulatory

definition of “basic and/or applied research” “clarifies that applying . . . traditional notions of

research—written reports or similar—in a real-world context is research,” Dkt. 29-1 at 33. The

AAO overlooked this portion of the regulatory definition, according to Open Society, by

discounting its efforts to “app[ly] the fruits of [its] research in the real world to effectuate

change.”

USCIS, in turn, responds that “the AAO did not discuss or consider whether [Open

Society] ‘conducted basic and/or applied research’ because the AAO determined [Open Society]

had not met the threshold requirement of being ‘primarily engaged’ in research.” Dkt. 35 at 10.

Although Open Society presents this argument as part of its critique of USCIS’s

interpretation of its own regulation, Dkt. 29-1 at 32–34, the arguments that follow criticize the

AAO’s treatment of its evidence and are therefore best addressed as part of Open Society’s

substantial-evidence argument. Unlike the “primarily engaged” requirement, see Dkt. 41-7 at 6–

8 (CAR 781–83), the AAO did not interpret the regulatory definition of “basic and/or applied

research” but, instead, applied that definition in a way that Open Society now says is incorrect.

See Dkt. 41-7 at 8–13 (CAR 783–88). The parties dispute, for example, whether the AAO

correctly found that Open Society’s mission statement weighed against a finding that the

organization is “primarily engaged” in research. Dkt. 29-1 at 33; Dkt. 35 at 11. The AAO noted

36 that Open Society’s stated goals included “seek[ing] to strengthen the rule of law,” before

concluding that “the mission of [Open Society] does not support a finding that it directly and

principally engages in research.” Dkt. 41-7 at 9 (CAR 784). Open Society argues that this

portion of its mission statement encompasses its efforts to “research[] a legal regime and then

propos[e] solutions to strength the rule of law”—“exactly the type of ‘investigation’ and then

real-world application that the ‘applied research’ definitions seems to contemplate.” Dkt. 29-1 at

33. But regardless of whether that is how Open Society interprets its mission statement, it is far

from evident that “seek[ing] to strength the rule of law,” Dkt. 41-7 at 9 (CAR 784), includes the

kind of applied research that Open Society now describes. The AAO also relied, moreover, on

other goals within Open Society’s mission statement, such as “help[ing] to shape public

policies,” “implement[ing] initiatives,” “build[ing] alliances,” and “protecting and improving the

lives of people in marginalized communities.”

Taken in its totality, this mission statement

does not so obviously demonstrate that Open Society is “primarily engaged” in research that a

“reasonable mind” could not find otherwise, Dickinson, 527 U.S. at 162.

In summary, even Open Society’s limited efforts to undermine the agency’s treatment of

the record are unpersuasive. The Court, accordingly, concludes that USCIS “could fairly and

reasonably find the facts as it did,” Chritton, 888 F.2d at 856, such that its decision is not

arbitrary and capricious.

3. USCIS’s Approval of Open Society’s Prior Petitions

Open Society’s final argument posits that the AAO’s denial of its petition was arbitrary

and capricious in light of USCIS’s approval of its prior petitions. See Dkt. 29-1 at 43–48; Dkt.

38 at 15–16. The parties agree that USCIS had previously determined that Open Society was

qualified for cap-exempt H-1B visas as a nonprofit research organization, even if they offer

37 different figures as to the precise number of such approvals. Compare Dkt. 29-1 at 40 (claiming

USCIS granted “fourteen prior petitions based on the same evidence”), with Dkt. 23-1 at 18

(“USCIS approved eleven petitions that [Open Society] filed claiming the exemption cap for a

nonprofit research organization.”). The AAO also acknowledged these prior petitions in its

decision, Dkt. 41-7 at 14 (CAR 789), as did the California Service Center in its initial decision,

see Dkt. 41-2 at 20–21 (CAR 19–20).

The parties disagree, however, regarding the significance of this precedent. Open Society

argues that such an “abrupt change in patterned decision making obligates the agency to provide

a more reasoned explanation than potential serial error.” Dkt. 29-1 at 47. USCIS responds that

its prior approvals “do[] not entitle [Open Society] to an automatic approval of future petitions”

and that the APA requires nothing more than what the agency provided in this case—a

reasonable explanation of its decision based on the record before it. Dkt. 23-1 at 19–20.

USCIS is correct that “‘[t]he mere fact that the agency, by mistake or oversight,

approved’ a visa petition on ‘one occasion does not create an automatic entitlement to the

approval of a subsequent petition.’” Fogo de Chao,

(alteration in original)

(quoting Royal Siam Corp. v. Chertoff,

(1st Cir. 2007)). “Agencies are, of

course, allowed to change their views over time.” Nw. Immigrant Rts. Project v. USCIS,

, 75 (D.D.C. 2020). But Open Society does not maintain that USCIS was prohibited

from changing its mind; it merely argues that “[t]he law demands that the agency explain a shift

in conduct,” given at least twelve prior approvals. Dkt. 29-1 at 48. That contention, moreover,

is not obviously wrong. In Fogo de Chao, the D.C. Circuit observed that “it may be that a

pattern of visa grants of sufficient magnitude could obligate the agency to provide a ‘reasoned

explanation for . . . treating similar situations differently’—or at least something more reasoned

38 than confessing a decade-long pattern of ‘material and gross error.’”

(alteration

in original) (quoting ANR Pipeline,

). But, as noted above, the D.C. Circuit had no

occasion to resolve that question given the plaintiff’s failure in that case to establish “that the

prior petitions were factually equivalent” such that the court could “conclude that [the agency] in

fact treated ‘similar situations differently.’”

(quoting ANR Pipeline,

).

Although Fogo de Chao’s dicta is neither definitive nor binding, it is well-established

that “an agency may not . . . depart from a prior policy sub silentio” and must, instead, “display

awareness that it is changing policy” and “show that there are good reasons for the new policy.”

FCC v. Fox Television Stations, Inc.,

(2009). An agency generally need not,

however, “provide a more detailed justification than what would suffice for a new policy created

on a blank slate,” or “demonstrate to a court’s satisfaction that the reasons for the new policy are

better than the reasons for the old one.”

The touchstone is that the agency’s explanation

must “enable” a reviewing court to conclude that the agency’s action “was the product of

reasoned decisionmaking.” Motor Vehicle Mfrs. Ass’n of U.S., Inc. v. State Farm Mut. Auto. Ins.

Co.,

(1983).

In the present context, the AAO did not run afoul of these principles. As an initial matter,

there is some question whether the agency actually changed course, such that it would need to

“display awareness that it is changing position . . . [and] provide[] a reasoned explanation for its

decision,” Nat’l Ass’n of Home Builders v. EPA,

(D.C. Cir. 2012). Open

Society’s prior approvals, after all, were each granted by USCIS’s California Service Center, and

the petition underlying this case offered the first opportunity for the AAO to weigh in on Open

Society’s eligibility as a “nonprofit research organization,”

(h)(19)(iii)(C). And

it is only when an agency “depart[s] from established precedent” that it must “acknowledge and

39 provide an adequate explanation for [that] departure.” Dillmon v. Nat’l Transp. Safety Bd.,

, 1089–90 (D.C. Cir. 2009).

But, as the D.C. Circuit has observed, the obligation to articulate coherent reasons for a

change in position attaches when an agency “depart[s] from precedents or practices,” Physicians

for Soc. Resp. v. Wheeler,

(D.C. Cir. 2020) (emphasis added), and so surely

something short of binding precedent is enough. Fogo de Chao, moreover, at least suggests that

“a pattern of visa grants of sufficient magnitude could obligate the agency to provide a reasoned

explanation for treating similar situations differently,” although that case involved a substantially

larger and longer prior pattern of decisionmaking—251 visa grants over the course of a decade.

(emphasis added) (alteration and quotation marks omitted). Because the D.C.

Circuit had no occasion to determine exactly how many prior grants an obligation to explain a

break in that pattern, or even whether any such pattern would definitively trigger such an

obligation, see

at 1144–46, the applicability of Fox Television remains an open question in the

present context.

The Court need not decide that issue to resolve the present dispute, however. Even

assuming that the California Service Center’s prior approvals gave rise to a requirement that the

AAO “display awareness that it [was] changing policy” and “show that there [were] good

reasons for the new policy,” Fox Television,

, the Court is persuaded that the

AAO satisfied that requirement. There is no question that the AAO acknowledged that its

decision conflicted with prior approvals from USCIS’s California Service Center, Dkt. 41-7 at 14

(CAR 789), as did the California Service Center in its decision, Dkt. 41-2 at 20–11 (CAR 19–

20). The AAO likewise offered “good reasons for the new policy.” Fox Television,

. It considered Open Society’s petition at length, and it offered a detailed analysis of the

40 regulatory requirements and the evidence that Open Society proffered to the California Service

Center (if not the AAO). See Dkt. 41-7 at 2–14 (CAR 777–89). Under Fox Television and its

progeny, no more is required. See 556 U.S at 515 (“[T]he agency need not always provide a

more detailed justification than what would suffice for a new policy created on a blank slate.”).

To be sure, the AAO did not engage in a lengthy discussion of the California Service

Center’s prior approvals, stating only that “if the prior approvals were premised on a similar

dearth of evidence, the approvals would have been issued in error.” Dkt. 41-7 at 14 (CAR 789).

That explanation, moreover, is at least arguably like the reasoning—“material and gross error”—

that the D.C. Circuit described as likely insufficient in Fogo de Chao, see

(“[A]

pattern of visa grants of sufficient magnitude could obligate the agency to provide a ‘reasoned

explanation for . . . treating similar situations differently’—or at least something more reasoned

than confessing a decade-long pattern of ‘material and gross error.’” (quoting ANR Pipeline,

)). Open Society argues with some force that it is entitled to a more complete

explanation for why the same evidence that passed muster repeatedly no longer suffices. See,

e.g., Dkt. 29-1 at 47.

The Court is ultimately unconvinced, however, because the purpose of the Fox Television

inquiry is “to ensure the agency’s prior policies and standards are being deliberately changed, not

casually ignored.” Dillmon,

(quotation marks omitted); see also Ramaprakash

v. FAA,

(D.C. Cir. 2003) (“[A]gency action is arbitrary and capricious if it

departs from agency precedent without explanation.”). In making that assessment, moreover, the

Court must bear in mind that a long line of cases requires courts to “uphold a decision of less

than ideal clarity if the agency’s path may reasonably be discerned.’” Pub. Citizen, Inc. v. FAA,

(D.C. Cir. 1993) (quoting Bowman Transp., Inc. v. Ark.-Best Freight Sys.,

41 Inc.,

(1974)). Here, the AAO acknowledged the California Service Center’s

prior approvals. It addressed the meaning of the governing regulation and defined the key terms.

And, it explained the insufficiency of Open Society’s evidence in light of the burden of proof

and controlling standard. In this posture, it is difficult to imagine what else the AAO might have

done to explain why it concluded that, even in the face of contrary precedent from the California

Service Center, Open Society’s petition was unavailing. The AAO is an appellate body and is

entitled to disagree with a regional service center when the law and the facts dictate. To the

extent the AAO disagreed with prior decisions of the California Service Center, it did all that

was required of it to assure the Court that the agency’s “policies and standards” were

“deliberately changed, not casually ignored,” Dillmon,

(quotation marks

omitted). 4

Finally, as noted above, the Court is able to reach this conclusion without the benefit of

the extra-record evidence Open Society seeks to introduce into the administrative record. See

Dkt. 30. The parties agree that the agency previously approved Open Society for cap-exempt H-

1B visas as a “nonprofit research organization,” see, e.g., Dkt. 23-1 at 18; Dkt. 29-1 at 40, and

Open Society’s argument is premised on the proposition that the petition underlying this case

included the evidence in those prior petitions that USCIS found sufficient, see Dkt. 29-1 at 13–

4 This conclusion disposes of Open Society’s primary argument—that USCIS’s purportedly “abrupt change in patterned decision making obligates the agency to provide [a] more reasoned explanation.” Dkt. 29-1 at 47. But at times Open Society intimates that even “a more reasoned explanation” would be insufficient, and that full notice-and-comment rulemaking is required before USCIS can deny its H-1B visa petitions. See id. at 44 (“The denial of [Open Society’s] petition is USCIS’s attempt to rewrite its regulation without any indicia of formality and sidestep the notice and comment procedures required by the APA.”). It is well-established, however, that an agency “is free to alter its past rulings and practices even in an adjudicatory setting,” Airmark Corp. v. FAA,

, 691–92 (D.C. Cir. 1985), and Open Society cites to no authority for the proposition that USCIS’s further articulation of the “primarily engaged” requirement itself required notice-and-comment rulemaking. See Fogo De Chao,

. 42 14. USCIS does not dispute this evidentiary point, and the Court concludes that the AAO’s

explanation was adequate even assuming the agency did, in fact, “treat[] similar situations

differently,” Fogo De Chao,

(quoting ANR Pipeline,

). But, in

any event, even if the Court were to consider Open Society’s proposed extra-record evidence,

that would not advance Open Society’s cause. Not only were the prior approvals on which Open

Society relies issued by USCIS’s California Service Center, but those approvals were entirely

unreasoned. See Dkt. 41-13 at 2–20 (SAR 2–20). Those decisions therefore included no

rationale with which to grapple, and thus the AAO in this case needed to offer only an adequate

explanation for why Open Society’s evidence failed to establish its eligibility. And, as already

explained, the agency did just that.

The Court therefore concludes that USCIS’s denial of Open Society’s petition was

neither arbitrary nor capricious in light of its prior approvals, regardless of whether the Court

considers Open Society’s proposed extra-record evidence.

4. Open Society’s Status as a “Nonprofit”

There is one final issue for the Court to address. In its motion for summary judgment,

Open Society notes that, “in the context of another petition,” the California Service Center

concluded that Open Society was not a qualifying “nonprofit” for purposes of the H1-B visa cap

exemption for nonprofit research organizations. Dkt. 29-1 at 48–49. Open Society “requests

that the Court . . . enter declaratory relief” confirming Open Society’s nonprofit status. Id. at 49.

In response, USCIS correctly observes that, “if [Open Society] disagrees with a separate decision

from the California Service Center[],” the “proper remedy is to appeal the decision to the AAO,

not to seek relief from this Court” in a case addressing an entirely different petition. Dkt. 35 at

20. Open Society moved for summary judgment on the claims within its amended complaint,

43 Dkt. 11, and that complaint does not challenge the finding Open Society now asks the Court to

declare unlawful. The AAO decision challenged in this action found that Open Society

“satisfie[d] the definition of a ‘nonprofit organization.’” Dkt. 41-7 at 6 (CAR 781) (quoting

(h)(19)(iv)). The Court’s review is limited to the agency action at issue here, as

“[i]t is well established that a party may not amend its complaint or broaden its claims through

summary judgment briefing.” Trudel v. SunTrust Bank,

(D.D.C. 2018)

(quotation marks omitted).

The Court will not address the propriety of a decision that is not before it.

CONCLUSION

For the foregoing reasons, it is hereby ORDERED that USCIS’s motion for summary

judgment, Dkt. 23, is GRANTED. It is further ORDERED that Open Society’s motion for

summary judgment, Dkt. 29, and its motion to supplement the administrative record, Dkt. 30, are

DENIED. The Clerk of Court is accordingly directed to close the case.

SO ORDERED.

/s/ Randolph D. Moss RANDOLPH D. MOSS United States District Judge

Date: September 7, 2021

44