Lisa Guffey v. Roslynn Mauskopf

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued September 20, 2021 Decided August 16, 2022

No. 20-5183

LISA GUFFEY AND CHRISTINE SMITH,
APPELLEES

v.

ROSLYNN R. MAUSKOPF, IN HER OFFICIAL CAPACITY AS
DIRECTOR OF THE ADMINISTRATIVE OFFICE OF THE UNITED
STATES COURTS,
APPELLANT

Consolidated with 20-5208

Appeals from the United States District Court
for the District of Columbia
(No. 1:18-cv-01271)

Weili J. Shaw, Attorney, U.S. Department of Justice,
argued the cause for appellant/cross-appellee. With him on the
briefs were Brian M. Boynton, Acting Assistant Attorney
General, and Scott R. Mcintosh, Attorney.

Scott Michelman argued the cause for appellees/cross-
appellants. With him on the briefs was Arthur B. Spitzer.
 2

Before: HENDERSON and WALKER, Circuit Judges, and
EDWARDS, Senior Circuit Judge.

Opinion for the Court filed by Circuit Judge WALKER.

Dissenting opinion by Circuit Judge HENDERSON.

WALKER, Circuit Judge: Lisa Guffey and Christine Smith
work at the Administrative Office of the United States Courts.
When they are away from work, they want to express support
for their preferred candidates in partisan elections.

AO employees could do that for the first 79 years of the
agency’s history. But since 2018, the AO has forbidden it.

That prohibition violates the Free Speech Clause of the
First Amendment.

I

We begin with some background on the Administrative
Office, the work that Guffey and Smith do there, the AO’s
speech restriction, and the district court proceedings.

A

Congress created the AO in 1939. Pub. L. No. 76-299, 53
Stat. 1223 (Aug. 7, 1939). It is now an 1,100-employee agency
within the Judicial Branch that provides a variety of valuable
support services. For example, its employees:

• Assist judges and courthouse staff with information
technology;
 3
• Help courts connect with visiting judges and coordinate
travel;
• Handle human resources and other support tasks for
courts, probation offices, and federal-defender services;
• Recommend positions to the Judiciary’s internal
policymaking body, the Judicial Conference, on issues
like codes of conduct, court administration, and
defender services;
• Issue press releases and statements on behalf of the
Judiciary;
• Advise judges on reimbursements, recusals, gifts, and
other ethics issues; and
• Represent the Judicial Conference before Congress and
the Executive Branch.

That far-from-exhaustive list shows the array of important
tasks that AO employees handle. But note what is not on that
list. AO employees do not decide cases — only judges do that.
Nor do they make recommendations about the outcomes of
individual cases, as law clerks and other legal advisors inside a
courthouse often do.

B

When this case began, Guffey and Smith worked with the
AO’s Defender Services Office. Guffey still does, but Smith
has since moved to the AO’s Department of Technology
Services.

Guffey makes sure that individual federal-defender offices
and court-panel attorney programs are adequately resourced,
operating effectively, and following administrative policies.
That work includes occasional meetings with judges to report
 4
on federal-defender offices and assess court-panel programs.
In a decade at the AO she has performed work related to an
individual case exactly once, when she researched the
appropriate level of funding for expert witnesses without
making a recommendation.

Until somewhat recently, Smith was the IT Liaison for the
Defender Services Office. She ensured that federal defenders’
IT needs were met and that they had secure systems. She
occasionally met with judges to advance those goals. In her
new role, Smith leads cyber-security assessments.

C

For the first 79 years of the AO’s history, its employees
have been free to engage in certain partisan political expression
outside the office. Both Guffey and Smith have engaged in
partisan political activities away from the job while employed
at the AO. There is nothing in the record to indicate that any
such political activity by Guffey, Smith, or any other AO
employee has had any adverse impact on the operations or
reputation of the AO or the judicial branch. But despite that
history, the AO revised its code of conduct in 2018 to prohibit
partisan political expression by its employees, whether done on
the clock or on their own time.

As is relevant here, the AO’s code of conduct now
prohibits:

1. Publicly expressing opinions about partisan candidates
or political parties, including on social media;
2. Wearing or displaying partisan badges, signs, or
buttons;
 5
3. Contributing money to a party, candidate, or political
action committee;
4. Attending partisan fundraisers;
5. Attending a partisan candidate’s campaign events;
6. Attending party conventions, rallies, and meetings;
7. Being a member of a partisan political organization;
8. Driving voters to the polls on behalf of a party or
candidate; and
9. Organizing events for a partisan candidate.
Those restrictions apply to partisan politics at all levels of
government, from a presidential election to a race for the
county register of deeds.1

D

Guffey and Smith sued the AO, seeking an injunction.
They want to be able to continue engaging in certain partisan
political expression outside the office — when they are in no
way affiliating themselves with the AO. While the suit
proceeded, they also moved for a preliminary injunction.
Guffey v. Duff, 330 F. Supp. 3d 66, 68 (D.D.C. 2018).

The district court granted a preliminary injunction as to the
first seven restrictions listed above (all but the restrictions on
driving voters to the polls and organizing events). Id. at 81.
Then, at the summary-judgment stage, the district court granted

1
See Emily Patrick, All About the Register of Deeds Race in the
Upcoming Election, Citizen Times (Oct. 11, 2016),
https://www.citizen-times.com/story/news/local/2016/10/11/all-
register-deeds-race-upcoming-election/91898522/ (describing the
heated, partisan register of deeds race in a North Carolina county).
 6
Guffey and Smith partial summary judgment and permanently
enjoined the same seven restrictions. Guffey v. Duff, 459 F.
Supp. 3d 227, 232 (D.D.C. 2020). Its injunction covered “all
AO employees except the six high-level ‘designated
employees’” to whom a different set of restrictions applied. Id.
at 256. It granted the AO summary judgment on Guffey and
Smith’s challenge to the driving and organizing restrictions.
Id. at 252.

Both parties appealed their losses.

II

Because the First Amendment prohibits the government
from “abridging the freedom of speech,” the AO’s restrictions
on Guffey and Smith’s off-duty political speech and activities
are unconstitutional. U.S. Const. amend. I.

A

The government has unique interests in its employees’
conduct. See United States v. National Treasury Employees
Union, 513 U.S. 454, 465-66 (1995). Those interests allow it
to regulate some of its employees’ speech, including even
political speech, in ways it could not regulate the general
public. Id.

But the government cannot condition public employment
on the complete surrender of a citizen’s First Amendment
rights. Id. at 465; see also Janus v. American Federation of
State, County, and Municipal Employees, 138 S. Ct. 2448,
2471 (2018). Instead, the government must justify prospective
restrictions on its employees’ off-duty speech by showing that
the speech’s “‘necessary impact on the actual operation’ of the
Government” outweighs the employees’ right to speak and the
 7
nation’s need to hear them. National Treasury Employees, 513
U.S. at 468 (quoting Pickering v. Board of Education, 391 U.S.
563, 571 (1968)).2

In this case, the weight of AO employees’ right to express
their political opinions outside the office is considerable. Time
and again, the Supreme Court has held that political speech
must receive “the highest level of First Amendment
protection.” Williams-Yulee v. Florida Bar, 575 U.S. 433, 443
(2015). That’s in large part because of “the close connection
between our Nation’s commitment to self-government and the
rights protected by the First Amendment. The First
Amendment creates an open marketplace in which differing
ideas about political, economic, and social issues can compete
freely for public acceptance without improper government
interference.” Knox v. Service Employees International Union,
Local 1000, 567 U.S. 298, 308-09 (2012) (cleaned up).

In light of the weight of the employees’ interest in political
expression, the AO has a “heavy” burden. National Treasury
Employees, 513 U.S. at 466. It must identify a commensurate
threat to its operations that justifies banning its employees’ off-
duty speech. Id. at 475.

2
Janus questioned whether the Pickering test properly applies to “a
blanket requirement” that burdens the speech of many employees, as
opposed to the one-employee disciplinary context that Pickering
presented. 138 S. Ct. at 2472 (“we have acknowledged that the
standard Pickering analysis requires modification” for blanket rules,
and “[t]he end product of those adjustments is a test that more closely
resembles exacting scrutiny than the traditional Pickering analysis”).
But it did not formally alter National Treasury Employees’s
articulation of the basic standard, so that articulation remains
controlling.
 8
That threat must be “real, not merely
conjectural” — “mere speculation” is not enough. Id. (quoting
Turner Broadcasting System, Inc. v. FCC, 512 U.S. 622, 664
(1994)). And as with any application of heightened scrutiny,
what it takes to show a real threat “will vary up or down with
the novelty and plausibility of the justification raised.” Nixon
v. Shrink Missouri Government PAC, 528 U.S. 377, 391
(2000).

To see the difference between “mere speculation” and a
“real” threat, compare United States v. National Treasury
Employees to Williams-Yulee v. Florida Bar. In National
Treasury Employees, the Supreme Court addressed Congress’s
novel concern that employees accepting honoraria for writing
or speaking would lead them to “misuse or appear to misuse
power.” 513 U.S. at 472. Because that risk was far from self-
evident, Congress needed to provide evidence. Id. at 473. It
didn’t. Id. at 472. So the Court decided that Congress’s
concern could not justify the ban on honoraria. Id. at 477. In
Williams-Yulee, on the other hand, the Court addressed the
“regrettable but unavoidable appearance that judges who
personally ask for money may diminish their integrity.” 575
U.S. at 447 (emphasis added). That concern was neither novel
nor implausible, so “proof by documentary record” was
unnecessary. Id.

Finally, even after it has identified a real threat, a
government employer may impose only those speech
restrictions that are “reasonably necessary to protect the
efficiency of the public service” against the threat. National
Treasury Employees, 513 U.S. at 474.
 9
B

The AO posits three threats to justify its prohibitions on
employees’ political expression outside the office. Each is too
speculative to survive the scrutiny required for a regulation of
political speech.

1

First, the AO argues that political expression by its
employees could undermine the public perception of the
Judiciary as an impartial adjudicative body.

That perception is “a state interest of the highest order.”
Williams-Yulee, 575 U.S. at 446 (quoting Caperton v. A.T.
Massey Coal Co., 556 U.S. 868, 889 (2009)). So the AO may
take steps “reasonably necessary” to preserve it. National
Treasury Employees, 513 U.S. at 474. But the connection
between that interest and the AO’s ban on off-duty speech
looks more like the loose connection in National Treasury
Employees than the narrow tailoring in Williams-Yulee.

For starters, in Williams-Yulee, the regulation allowed
judicial candidates to “discuss any issue with any person at any
time.” 575 U.S. at 452. It just prohibited them from soliciting
campaign contributions and public endorsements. Id. at 439.
That prohibition was based on the “intuitive” notion that
“Judges, charged with exercising strict neutrality and
independence, cannot supplicate campaign donors without
diminishing public confidence in judicial integrity.” Id. at 445.

The connection between the AO’s speech restriction and
the Judiciary’s reputation for deciding cases impartially is not
as intuitive. Guffey and Smith do not decide the outcome — or
even make recommendations about the outcome — of
 10
individual cases. That matters because conduct threatens
judges’ reputation for impartiality when it threatens judges’
reputation for impartially deciding cases. See Williams-Yulee,
575 U.S. at 445-46 (our authority to decide cases “depends in
large measure on the public’s willingness to respect and
follow” our judgments).

To credit the AO’s concern for the perception of judicial
impartiality, we would have to assume that the public is aware
of the AO. There is nothing in the record to suggest that it is.

But even granting the AO that unlikely premise, we would
have to further assume that the public first will learn about the
political activity of AO employees like Guffey and Smith and
then will lose confidence in judges’ adjudication of cases
because those employees support a particular candidate on their
own time. Those two assumptions are novel, implausible, and
unsubstantiated: Even with eight decades of AO history to
draw from, the AO has excavated no instance of off-duty
political conduct by an AO employee that has injured the
Judiciary’s reputation. See Shrink Missouri Government, 528
U.S. at 391; National Treasury Employees, 513 U.S. at 472.
That silent record is strong evidence that AO employees can
speak on matters of public concern without tarnishing the
reputation of the Judiciary.

Without evidence, the AO makes its case for censorship
by (1) speculating that AO employees’ off-duty speech could
be exploited by nefarious foreign actors in novel ways,
(2) analogizing its employees’ speech to the partisan activities
of key investigators at the center of the century’s most high-
profile investigation of an American president, and
(3) conflating judges and AO employees. We’ll address each
in turn.
 11
First, the AO says that nefarious actors like Russian
propaganda agencies could try to attribute AO employees’
private political expression to the Judiciary as a whole, in order
to falsely characterize the Judiciary as partisan. But that is
“mere speculation.” National Treasury Employees, 513 U.S.
at 475. And the speculative prospect of Russian propaganda
does not justify censoring the political speech of American
citizens. Cf. United States v. Alvarez, 567 U.S. 709, 727 (2012)
(plurality opinion) (“The remedy for speech that is false is
speech that is true.”).

Second, the AO speculates that the Judiciary will face
accusations of partisan bias like the criticism Special Counsel
Robert Mueller’s investigative team faced several years ago.
But that analogy doesn’t work. Mueller’s investigators
prepared subpoenas,3 questioned witnesses,4 advised Mueller
on when to bring charges,5 and drafted conclusions about
allegations that could have led to a president’s impeachment.6

3
Special Counsel Robert S. Mueller, III, U.S. Department of Justice,
Report On The Investigation Into Russian Interference In The 2016
Presidential Election: Volume I of II 13 (2019) (the Special
Counsel’s office “issued more than 2,800 subpoenas”).
4
Darren Samuelsohn, What Mueller’s Org Chart Reveals About His
Russia Probe, POLITICO (Nov. 13, 2017),
https://www.politico.com/story/2017/11/13/robert-mueller-russia-
probe-organization-244789.
5
Id. One lawyer advised on “interpreting federal criminal statutes”
and the “special counsel’s own boundaries for pursuing” a “case
against a sitting president.”
6
Josh Gerstein, Justice Dept: Mueller Prepared No Reports to
Congress, POLITICO (Feb. 7, 2020),
https://www.politico.com/news/2020/02/07/justice-department-
mueller-112388 (noting that “the special counsel’s team drafted” a
report that “[s]ome observers compared . . . [to] impeachment
 12
They wielded the Special Counsel’s substantive power, just as
judges wield the judicial power, so they were a predictable
focus of criticism. It would have been quite unexpected,
however, if anyone had attempted to weaponize the political
preferences of the essential support staff who worked with the
Special Counsel.

Third, the AO speculates that scrutiny of employees’
political speech might one day resemble the scrutiny of the
Judicial Conference’s Code of Conduct Committee after it
proposed guidance about judges’ membership in the American
Bar Association, the American Constitution Society, and the
Federalist Society. Before that proposal was abandoned, a
congressman questioned the “biases and motivations of the
opinion’s drafters” — but he asked about the committee
members’ membership in those groups, not their political
affiliations.7 More importantly, those committee members
were judges, not AO employees. That is why, as the AO
recognizes, criticism of the committee’s guidance focused on
the judges’ backgrounds, not the AO employees who assisted
them.8 And even if we assume that the AO can limit the off-
duty political speech of employees who make policy
recommendations to that committee, it is not “reasonably
necessary” to impose the same limits on AO employees whose

fodder . . . although the articles of impeachment . . . largely
overlooked” it).
7
Appellant’s Brief at 37 (quoting Letter from Rep. Jim Jordan to
Sheryl Walter (May 18, 2020), https://republicans-
judiciary.house.gov/wp-content/uploads/2020/05/2020-05-18-JDJ-
to-Judicial-Conference-re-Draft-Advisory-Opinion.pdf).
8
Appellant’s Brief at 36 (One article cited “the fact that a judge on
the committee had donated (prior to joining the bench) to a Senator
who had spoken out on the issue.”).
 13
work is nothing like that. National Treasury Employees, 513
U.S. at 474.

2

The AO also argues that if Congress learns of AO
employees’ political views, it will have less faith in the
employees’ ability to be neutral messengers for the Judiciary.9

The AO has identified no other case that has considered
the weight of the AO’s concern for the Judiciary’s relationship
with Congress. That relationship is undoubtedly important.
The Judiciary relies on Congress for essentials like funding, the
creation of judgeships, and the construction of courthouses.
And Congress looks to the Judiciary for advice on subjects like
death-penalty reform, the propriety of nationwide injunctions,
and other court-facing issues.

But even assuming the AO’s new ban on off-duty political
speech is defensible when applied to employees who work with
the other branches, the AO has offered no reason to think that
most of its employees do such work. Instead, the AO censored
all of its 1,100 employees because it says that “it is difficult to
predict with certainty which AO employees will be required, as
part of their job responsibilities, to represent the Judiciary to
the other branches of government.” JA 142 ¶ 21.

That difficulty does not satisfy the AO’s “heavy” burden.
National Treasury Employees, 575 U.S. at 466; cf. Janus, 138
S. Ct. at 2477 (logistical concerns about preventing free riders
could not justify a “heavy burden” on “First Amendment

9
Although the AO also mentions its relationship with the Executive
Branch, it offers little explanation of, or argument about, that
relationship. So like the AO, we will focus on Congress.
 14
interests”). Some employees who will appear before Congress
are obvious. For example, the Office of Legislative Affairs
“carries out the Judiciary’s legislative liaison activities with
Congress and other government entities.” JA 138, ¶ 14.d.
Perhaps the AO can impose its new rule on employees there
who might one day meet with members of Congress or their
staffs. But many other employees — like Guffey and
Smith — have no reason to think that they will ever interact
with Congress. Censoring their off-duty political speech is not
“reasonably necessary” to address the AO’s concern for the
Judiciary’s relationship with the other branches. National
Treasury Employees, 513 U.S. at 474.10

3

The AO’s last concern is that judges will not trust AO
employees who engage in partisan political expression while
off-duty. But even assuming that judges will learn of AO
employees’ private activity, this concern is implausible.

The AO’s argument focuses on the 30 or so employees
who advise judges on sensitive matters like recusals and
participation in outside activities. That alone indicates that
imposing the restrictions on all 1,100 AO employees — from
the HR professional who processes a travel reimbursement to
the IT-help-desk worker who assists a judge with a forgotten
password — is not “reasonably necessary to protect the

10
Of course, none of this should be read to imply that any AO
employee’s work is more important than any other’s. We are simply
highlighting the disconnect between the speculative threats that the
AO identified and the work that most of its employees do. All AO
employees provide the Judiciary essential and much appreciated
support.
 15
efficiency of the public service.” National Treasury
Employees, 513 U.S. at 474.

Even as to those 30 employees, though, there is some
reason to doubt the foundation of the AO’s predicted harm. It
conflicts with “the powerful and realistic presumption that the
federal work force consists of dedicated and honorable civil
servants.” National Treasury Employees, 513 U.S. at 476.
And we have no reason to doubt the AO when it tells us that its
employees fit that bill by performing “their job duties and tasks
without regard for partisan considerations.” R.24-1 ¶ 28.

III

The above analysis applies equally to all nine speech
restrictions, from publicly stating political opinions to
organizing political events. But the district court treated two
of those nine restrictions differently: driving voters to the polls
and organizing events for a partisan candidate. Guffey, 459 F.
Supp. 3d at 251. That differential treatment turned largely on
the district court’s belief that the Supreme Court’s approval of
similar restrictions on Executive Branch employees in the
Hatch Act “strongly supports their legality” here. Id. at 252.

Although the district court’s analysis was thoughtful and
thorough, we disagree.

In the Hatch Act context, the Court held that speech
restrictions were justified by a federal interest in ensuring “that
the rapidly expanding Government work force should not be
employed to build a powerful, invincible, and perhaps corrupt
political machine.” United States Civil Service Commission v.
National Association of Letter Carriers, 413 U.S. 548, 565
(1973). Importantly, Congress sought to protect employees
from “pressure and from express or tacit invitation to vote in a
 16
certain way or perform political chores in order to curry favor
with their superiors rather than to act out their own beliefs.” Id.
at 566. It had extensive evidence of such pressures. See, e.g.
84 Cong. Rec. 9598 (1939) (statement of Rep. Taylor)
(describing Works Progress Administration supervisors
forcing WPA workers to place part of their paycheck “under
the Democratic donkey paperweight” on their supervisor’s
desk).

Thus, the Hatch Act’s restrictions passed muster because
they “aimed to protect employees’ rights, notably their right to
free expression, rather than to restrict those rights.” National
Treasury Employees, 513 U.S. at 471. That is not the case here.
The AO makes no claim that its censorship protects its
employees’ rights. Instead, the AO restricts its employees’
expression to fend off speculative harms to its own operational
interests. So the Hatch Act’s balance between employees’
speech rights and protecting employees from political pressure
offers little guidance here.

Absent the belief that precedent directs it, there is no
reason to treat driving voters to the polls and organizing
political events differently from the other seven prohibited
modes of political expression. They all implicate core First
Amendment rights. See Monitor Patriot Co. v. Roy, 401 U.S.
265, 272 (1971). And, again, the AO has failed to show that
they present any non-speculative threat to its operations.

IV

We turn now to the scope of relief. The district court
reasoned that because analyzing a prospective restriction on
employee speech “requires the court to go beyond the facts of
the particular case before it,” granting an injunction to all AO
employees was appropriate. Guffey, 459 F. Supp. 3d at 256
 17
(quoting Sanjour v. EPA, 56 F.3d 85, 92 (D.C. Cir. 1995)). But
the analysis of a right and the choice of a remedy are distinct
concepts. Brown v. Sessoms, 774 F.3d 1016, 1021 (D.C. Cir.
2014).

That’s why, in applying the same test in National Treasury
Employees, the Supreme Court still emphasized the age-old
principle that “we neither want nor need to provide relief to
nonparties when a narrower remedy will fully protect the
litigants.” 513 U.S. at 478. Applying that principle makes
especially good sense here — the AO may believe that
employees who do different jobs than Guffey and Smith should
be subject to different restrictions. We cannot assess that belief
on the record before us. And an injunction that applies only to
Guffey and Smith will fully protect their rights. That narrower
remedy is therefore the right one.

Further, the AO is a government entity with an
independent duty to uphold the Constitution. We trust that
upon receipt of our judgment, it will reconsider the contested
restrictions for employees whose work is comparable to (or less
sensitive than) the work Guffey and Smith do.

* * *

The AO wants to maintain the Judiciary’s reputation for
independence from politics. That is among the worthiest of
goals. And no one in this case takes issue with the AO’s
longstanding prohibition of political speech by its employees
when they are at the office. But the AO cannot prohibit
political speech by Guffey and Smith when they are away from
work and in no way affiliating themselves with the Judiciary.

We affirm the district court’s grant of summary judgment
to Guffey and Smith, but we limit its injunction against the first
 18
seven restrictions to apply only to Guffey and Smith. We
reverse its grant of summary judgment to the AO on the other
two restrictions, and we remand for it to enjoin their application
to Guffey and Smith as well.

So ordered.
 KAREN LECRAFT HENDERSON, Circuit Judge, dissenting: I
agree with my colleagues that Lisa Guffey and Christine Smith
both have a strong interest in freely participating in the political
process, a right fundamental to our democracy. See
McCutcheon v. FEC, 572 U.S. 185, 191 (2014) (plurality
opinion). And few, if any, could deny that the challenged
restrictions on the partisan political activities of Administrative
Office of the United States Courts (AO) employees encroach
on that right. On the other side of balance, however, is the
“genuine and compelling” interest in safeguarding the “public
perception of judicial integrity,” Williams-Yulee v. Fla. Bar,
575 U.S. 433, 446, 447 (2015), “a state interest of the highest
order,” id. at 446 (quoting Caperton v. A.T. Massey Coal Co.,
556 U.S. 868, 889 (2009)). “The guarantee of an independent,
impartial judiciary enables society to ‘withdraw certain
subjects from the vicissitudes of political controversy, to place
them beyond the reach of majorities and officials and to
establish them as legal principles to be applied by the courts.’”
Republican Party of Minn. v. White, 536 U.S. 765, 804 (2002)
(Ginsburg, J., dissenting) (quoting W. Va. Bd. of Ed. v.
Barnette, 319 U.S. 624, 638 (1943)); see also United States v.
Will, 449 U.S. 200, 217–18 (1980) (“A Judiciary free from
control by the Executive and the Legislature is essential if there
is a right to have claims decided by judges who are free from
potential domination by other branches of government.”). With
neither “sword” nor “purse” to safeguard its independence, see
The Federalist No. 78, at 465 (Alexander Hamilton) (Clinton
Rossiter ed., 1961), the Judiciary’s position of authority within
our constitutional framework “depends in large measure on the
public’s willingness to respect and follow its decisions,” see
Williams-Yulee, 575 U.S. at 445–46. Yet recent evidence
indicates that public confidence in the Judiciary is in decline.
See Brandon L. Bartels & Christopher D. Johnston, On the
Ideological Foundations of Supreme Court Legitimacy in the
American Public, 57 Am. J. Pol. Sci. 184, 185–86 (2013)
(“[I]ndividuals grant or deny the [Supreme] Court legitimacy
based on the ideological tenor of the Court’s” decisions.); see
 2
also Megan Brenan, Americans’ Trust in Government Remains
Low, Gallup News (Sept. 30, 2021). 1

In revising its employees’ code of conduct to impose the
nine challenged restrictions on partisan political conduct, 2 the
AO has attempted to safeguard this vital faith in the Judiciary
as a body of independence and impartiality. Yet the majority
gives short shrift to its attempt. Because I find the AO has
adequately justified its restrictions, as required under United
States v. National Treasury Employees Union (NTEU), 513
U.S. 454 (1995), I would reverse the district court’s partial
invalidation of the restrictions and allow all nine to take effect. 3
Accordingly, and with respect, I dissent.

I. Administrative Office’s Burden

I first depart from my colleagues regarding the AO’s
evidentiary burden under the NTEU framework. As discussed
infra, the AO need provide only a reasonable ground to

1
Available at https://news.gallup.com/poll/355124/americans-
trust-government-remains-low.aspx.
2
The restrictions prohibit AO employees from (1) publicly
expressing opinions about partisan candidates or political parties,
including on social media; (2) wearing or displaying partisan badges,
signs or buttons; (3) contributing money to a party, candidate or
political action committee; (4) attending partisan fundraisers; (5)
attending a partisan candidate’s campaign events; (6) attending party
conventions, rallies and meetings; (7) belonging to a partisan
political organization (other than registering as a member of a
political party for voting); (8) driving voters to the polls on behalf of
a party or candidate; and (9) organizing events for a partisan
candidate.
3
I agree with my colleagues that the restrictions precluding AO
employees from driving voters to the polls and organizing events for
a partisan candidate should not be treated differently from the other
seven; all nine rise or fall together. See Majority Op. at 15–16.
 3
conclude that the harms it fears will occur. But documentary
evidence of past harm is only one way to support its
restrictions. And precedent like Williams-Yulee v. Florida Bar,
575 U.S. 433 (2015), highlights how preserving the perception
of judicial impartiality is a compelling interest—one that
extends beyond adjudication of discrete cases.

To impose a prospective, generally applicable restriction
on employee speech, the government is required to “show that
the interests of both potential audiences and a vast group of
present and future employees in a broad range of present and
future expression are outweighed by that expression’s
‘necessary impact on the actual operation’ of the Government.”
NTEU, 513 U.S. at 468 (quoting Pickering v. Bd. of Educ., 391
U.S. 563, 571 (1968)). In other words, the government must
show that the harms it aims at are “real, not merely conjectural”
and that “the regulation will in fact alleviate the[] harms in a
direct and material way.” Id. at 475 (quoting Turner Broad.
Sys., Inc. v. FCC, 512 U.S. 622, 664 (1994)). The government
must also demonstrate that the restriction’s “sweep” is
“reasonably necessary to protect the efficiency of the public
service.” Id. at 466, 474.

In support of the challenged restrictions, the AO asserts
three interests centered on protecting the perception of the
Judiciary as a nonpartisan, impartial body: first, in the public
view (Public Perception Interest); second, in the view of
members of the two elected branches (Inter-Branch Interest);
and finally, in the view of the Judiciary itself (Intra-Branch
Interest). See Guffey v. Duff (Guffey II), 459 F. Supp. 3d 227,
236–37 (D.D.C. 2020); see also Duff Decl. ¶ 23, Joint
Appendix (J.A.) 143. In describing the AO’s burden, the
district court focused on NTEU’s requirement that the
government’s recited harms be “real, not merely conjectural,”
Guffey II, 459 F. Supp. 3d at 241 (quoting NTEU, 513 U.S. at
 4
475), and concluded that the AO must “point to documentary
evidence showing that employees’ activities have eroded
public confidence in the past and will continue to do so if left
unrestricted,” id. at 243 (quoting Guffey v. Duff (Guffey I), 330
F. Supp. 3d 66, 76 (D.D.C. 2018)). The district court then
seemed to “relax[]” the burden for the Public Perception
Interest and allowed the AO to rely on “realistic hypotheticals
of how partisan activity restricted under the Code could lead
the public to believe that the judiciary is not behaving
impartially.” Id. at 244 (quoting Guffey I, 330 F. Supp. 3d at
76). It did so in light of Williams-Yulee’s caution that “the
concept of public trust in judicial impartiality ‘does not easily
reduce to precise definition, nor does it lend itself to proof by
documentary record.’” Guffey I, 330 F. Supp. 3d at 76 (quoting
Williams-Yulee, 575 U.S. at 447). But the district court
declined to relax the burden for the AO’s Inter-Branch and
Intra-Branch Interests by limiting its analysis to public trust in
the Judiciary’s adjudicative function and ignoring the
Judiciary’s administrative role. See Guffey II, 549 F. Supp. 3d
at 244.

As I see it, the district court imposed too high an
evidentiary burden. Notwithstanding NTEU’s mandate, see
also Nixon v. Shrink Mo. Gov’t PAC, 528 U.S. 377, 392 (2000)
(“We have never accepted mere conjecture as adequate to carry
a First Amendment burden . . . .”), it does not follow that “real”
evidence requires “documentary” evidence in all situations. As
the majority rightly notes, see Majority Op. at 8, the
government’s evidentiary burden in the First Amendment
context “will vary up or down with the novelty and plausibility
of the justification raised.” Shrink Mo. Gov’t PAC, 528 U.S. at
391. This holds true even for restrictions reviewed under strict
scrutiny. See Lorillard Tobacco Co. v. Reilly, 533 U.S. 525,
555 (2001) (“[W]e have permitted litigants . . . even[] in a case
applying strict scrutiny, to justify restrictions based solely on
 5
history, consensus, and simple common sense.” (internal
quotation marks omitted) (quoting Fla. Bar v. Went For It, Inc.,
515 U.S. 618, 628 (1995)); see also Williams-Yulee, 575 U.S.
at 448 (government is “not require[d] . . . to tolerate . . . risks”
to public perception of judicial integrity (emphasis added));
Blount v. SEC, 61 F.3d 938, 939–40, 943–45 (D.C. Cir. 1995)
(applying strict scrutiny to SEC restrictions on municipal
finance professionals’ campaign contributions to state and
local officials but rejecting argument that agency must show
“specific instances of quid pro quos”).

In NTEU itself, the Supreme Court, after referencing the
“real, not merely conjectural” harm requirement, noted that the
government can rely on the fear or risk of harm so long as there
is “‘reasonable ground to fear that serious evil will result if free
speech is practiced.’” 513 U.S. at 475 (quoting Whitney v.
California, 274 U.S. 357, 376 (1927) (Brandeis, J.,
concurring)). Similarly, in Weaver v. United States Information
Agency, 87 F.3d 1429 (D.C. Cir. 1996), we upheld a policy
requiring prepublication review of employees’ “speaking,
writing, and teaching material on matters of official concern”
in furtherance of the agency’s “compelling” interest in
safeguarding classified information, id. at 1431, 1441 (internal
quotation marks omitted), without requiring documentary
evidence of past harm. Instead, the agency supported its review
of speech by personnel without direct access to classified
information via an affidavit explaining that such employees
may unknowingly come into contact with classified
information. Id. at 1441. 4

4
Granted, in Weaver we upheld a review of employee speech,
not an outright prohibition, in order to avoid “constitutional
difficulties.” 87 F.3d at 1440. Weaver nevertheless indicates that
simply invoking the NTEU test does not suffice to make
documentary evidence necessary.
 6
Williams-Yulee is the Supreme Court’s most recent
elaboration on the governmental interest in preserving
confidence in the Judiciary. Williams-Yulee involved a
provision of the Florida Supreme Court’s Code of Judicial
Conduct that forbade elected judges from personally soliciting
campaign funds. 575 U.S. 439–40. The Court upheld the
restriction, recognizing that “public perception of judicial
integrity is ‘a state interest of the highest order.’” Id. at 446
(quoting Caperton, 556 U.S. at 889). Important here, the Court
observed that, although “[t]he concept of public confidence in
judicial integrity does not easily reduce to precise definition,
nor does it lend itself to proof by documentary record,” the
need to safeguard that confidence “is genuine and compelling.”
Id. at 447. There, the Court applied strict scrutiny—a standard
of review higher than that applied in NTEU—but did not
require documentary evidence of past harm. Cf. Williams-
Yulee, 575 U.S. at 447–48 (government is “not require[d] . . .
to tolerate the[] risks” to public perception of judicial integrity
(emphasis added)).

The district court limited Williams-Yulee to the AO’s
Public Perception Interest and to the Judiciary’s adjudicative
function, see Guffey II, 459 F. Supp. 3d at 243–44, but I believe
that limitation is unwarranted. There is little reason to think that
the AO’s interests in preserving the Judiciary’s reputation for
impartiality among members of the elected branches—with
which it routinely interacts—as well as within the Judiciary and
its Administrative Office inter se are any less compelling or
less susceptible of “precise definition” than preserving its
reputation with the general public. Williams-Yulee, 575 U.S. at
447. Further, I doubt that public perception can be isolated
from elected officials’ perception. The latter are elected, and
therefore influenced, by the former. See McCutcheon, 572 U.S.
at 227 (plurality opinion) (“Representatives are not to follow
constituent orders, but can be expected to be cognizant of and
 7
responsive to those concerns. Such responsiveness is key to the
very concept of self-governance through elected officials.”); cf.
Gravel v. United States, 408 U.S. 606, 661 (1972) (Brennan, J.,
dissenting) (“The dialogue between Congress and people has
been recognized, from the days of our founding, as one of the
necessary elements of a representative system.”). And political
conflicts within and between governmental branches and
institutions necessarily spill into the public sphere. See
generally David R. Mayhew, America’s Congress: Actions in
the Public Sphere, James Madison Through Newt Gingrich 1–
28 (2000).

Moreover, although Williams-Yulee does not speak to the
Judiciary outside its adjudicative function, or to its relationship
with the Congress, other decisions do. For example, in
Mistretta v. United States, 488 U.S. 361 (1989), the Supreme
Court considered the propriety of federal judges’ membership
on the United States Sentencing Commission—what the Court
deemed “judiciary involvement in the making of policy”—and
whether participation in policy-making risked “undermin[ing]
public confidence in the disinterestedness of the Judicial
Branch,” id. at 407. The Court has also considered how its
“exercise of judicial power”—its adjudicative function—
“affects relationships between the coequal arms of the National
Government,” aware that “when employed unwisely or
unnecessarily,” that power can pose a “threat to the continued
effectiveness of the federal courts.” Valley Forge Christian
Coll. v. Ams. United for Separation of Church and State, Inc.,
454 U.S. 464, 473 (1982). This precedent recognizes the
Judiciary’s awareness of its significant role outside the
courtroom. Indeed, the AO regularly interacts with the
Congress to secure funding and influence policies affecting the
Judiciary, including, inter alia, new judgeships, the federal
rules of procedure and evidence, federal jurisdiction and
pretrial and probationary services. See Duff Decl. ¶ 11, J.A.
 8
134–36. The AO has a designated Office of Legislative Affairs
to facilitate this relationship, see id. at ¶ 14, J.A. 138–39, and
former legislative staffers attest to the essential role the AO
plays in advocating for the Judiciary and advising the Congress
on judicial matters, see Cooney Decl. ¶ 9, J.A. 164; see also
Weich Decl. ¶ 11, J.A. 176–77. Lacking either “sword” or
“purse,” see The Federalist No. 78, at 465, the Judiciary’s
reputation—and, of necessity, the AO’s reputation—for
nonpartisanship and impartiality is the currency with which it
is able to negotiate on behalf of its interests. No less than with
its adjudicative function, the Judiciary’s status as a co-equal
branch to set its administrative and policymaking priorities
“depends in large measure” on the Congress’s “willingness to
respect” its position as an impartial, nonpartisan branch. See
Williams-Yulee, 575 U.S. at 445–46. The alternative is a
Judiciary “drawn into political disputes with other branches,”
which serves only to “diminish its legitimacy before [the]
general public as well.” Appellant’s Br. 33; cf. Valley Forge
Christian Coll., 454 U.S. at 474 (1982) (Powell, J., concurring)
(“[R]epeated and essentially head-on confrontations between
the life-tenured branch and the representative branches of
government will not, in the long run, be beneficial to either.”
(alteration in original) (quoting United States v. Richardson,
418 U.S. 166, 188 (1974) (Powell, J., concurring))).

II. Administrative Office’s Interests

The next questions are whether the AO’s “recited harms
are real, not merely conjectural” and whether the challenged
restrictions “will in fact alleviate these harms in a direct and
material way.” NTEU, 513 U.S. at 475 (quoting Turner Broad.
Sys., 512 U.S. at 664).
 9
A. Public Perception Interest

The AO’s Public Perception Interest is premised on the
concern that its employees’ partisan political activities will
undermine the public perception of the Judiciary as an
apolitical, impartial body. As the majority notes, the AO has
not isolated a specific instance of this fear’s materialization.
See Majority Op. at 10. But here, “materialization” may be
provable only by assessing the risk of harm. The AO has
pointed out instances of entities leveraging the political
activities of government employees to impeach their
employer’s reputation and further a partisan agenda and has
submitted affidavits attesting to the risk that similar partisan
attacks could be aimed at the AO without the challenged
prohibitions. Its showing provides, at least to me, a “reasonable
ground” to support the AO’s Public Perception Interest, in
furtherance of which a ban on partisan political activity is the
most “direct and material” means of eliminating that risk.
NTEU, 513 U.S. at 475 (quoting Turner Broad. Sys., 512 U.S.
at 664).

The AO has described instances of opposition research
firms and partisan-aligned groups seeking information
regarding politically focused emails sent and campaign
contributions made by employees of the National Oceanic and
Atmospheric Administration, the Environmental Protection
Agency and the Internal Revenue Service—information sought
in order to cast doubt on the reputation and policies of the
employing agency. See Appellant’s Br. 41; Appellant’s Reply
Br. 20–21; see also; Phillip Bump, What campaign
contributions tell us about the partisanship of government
employees, Wash. Post (Dec. 27, 2018) 5; Timothy P. Carney,

5
Available at https://www.washingtonpost.com/politics/2018/
12/27/is-trumps-dismissal-unpaid-government-employees-
democrats-accurate/.
 10
The IRS Is Deeply Political—and Very Democratic, Wash.
Exam’r (May 15, 2013). 6 It did not matter to these entities
whether the targeted employees in fact wielded any significant
influence in the policy matters at issue. Similarly, it does not
matter if a particular AO employee lacks significant power
over the AO’s policy-making functions or the Judiciary’s
adjudication of cases, see Majority Op. at 9–10, as the AO
itself, through its employees, routinely engages in potentially
contentious debates relating to the nation’s judicial system, see
infra p. 13–14.

The AO’s examples track with its submitted declarations,
including one from a lobbyist and former congressional staffer
who attested that he was “aware of firms whose business
models include the dissemination of this type of political
information in order to seek to portray particular groups or
individuals as supporting or opposing a particular partisan
agenda.” Cooney Decl. ¶ 16, J.A. 168. This information, the
lobbyist noted, can then be packaged and presented “to a media
organization or publication for dissemination to the general
public.” Id. If this sort of targeted aggregation and
dissemination of political activity routinely occurs with
executive-branch employees and agencies, it is no stretch to
conclude it can—and will—happen with AO employees.

The district court nevertheless faulted the AO’s
hypotheticals premised on these sorts of partisan attacks for
being based on an untenable chain of inferences:

[M]embers of the public would [1] need to
observe an AO employee engaged in partisan
activity, [2] somehow come to know that the
person in the photo or social media post is an

6
Available at https://www.aei.org/articles/the-irs-is-deeply-
political-and-very-democratic/.
 11
AO employee, [3] understand that AO
employees work with federal judges, [4] but
mistakenly believe that they play a role in
handling individual cases, and [5] assume—
based on ordinary expressions of political
preference—that the AO employee is so
politically biased that she would be willing to
violate her professional ethical obligations by
attempting to sway the outcome of a case.

Guffey II, 459 F. Supp. 3d at 249. The majority in effect echoes
the district court. See Majority Op. at 10 (labeling the
assumptions “novel, implausible, and unsubstantiated”).

Assuming Williams-Yulee should not be limited to the
Judiciary’s adjudicative function as opposed to its
administrative and policy-making functions, see supra p. 6–8,
I believe the chain of inferences is not implausible. For one
thing, not all members of the public are neutral observers and
we cannot ignore that partisanship is an unavoidable fact of
21st-century society. For another, to identify an individual who
engages in political expression as an AO employee is not
difficult for an even mildly motivated actor. Individuals,
including government employees, regularly post their
headshots and job titles on networking websites like LinkedIn.
Further, the increased use of “doxxing”—“publicly
identify[ing] or publish[ing] private information about
(someone) especially as a form of punishment or revenge” 7—
makes public identification of a person all the more likely. See
Nellie Bowles, How ‘Doxxing’ Became a Mainstream Tool in
the Culture Wars, N.Y. Times (Aug. 30, 2017) 8; see also

7
Dox, Merriam-Webster, https://www.merriam-webster.com/
dictionary/dox (last visited July 25, 2022).
8
Available at https://www.nytimes.com/2017/08/30/
technology/doxxing-protests.html.
 12
Callum Borchers, Doxxed Trump Donors Have an Unlikely
Defender in this Democratic Congressional Candidate, Wash.
Post (Apr. 30, 2017) (highlighting Twitter account tweeting
“the names, hometowns, occupations and employers of people
who contributed . . . to [former President] Trump’s
campaign”). 9 As a more concrete example, an individual’s
political contributions, alongside his employment information,
is publicly available through the Federal Election
Commission’s database. See Cooney Decl. ¶ 16, J.A. 168–69;
Weich Decl. ¶ 18, J.A. 180; see also Federal Election Comm’n,
Campaign Finance Data (last visited July 25, 2022). 10 Finally,
once a motivated actor compiles the identifying information, is
it likely he/it will pause to learn whether those AO employees
are in fact involved in the Judiciary’s adjudicative function
before packaging the information for consumption by media
and the general public? These examples involving executive-
branch employees indicate that hesitation is unlikely.

To the extent the chain of inferences is deemed
implausible because the AO is a small or unfamiliar office, see
Appellees’ Br. 38–39, this conclusion is also unwarranted. It is
far from remote that members of the public might fail to
differentiate employees of the Administrative Office of the
United States Courts from those employees located in
courthouses; nor is it implausible for a partisan group to gloss
over any such distinction before packaging its message for
dissemination to the public, especially given that courthouse
employees have long been barred from partisan political
activity, see U.S. Courts, Guide to Judiciary Policy, vol. 2, pt.

9
Available at https://www.washingtonpost.com/news/the-
fix/wp/2017/04/30/doxxed-trump-donors-have-an-unlikely-
defender-in-this-democratic-congressional-candidate/.
10
Available at https://www.fec.gov/data/.
 13
A, ch. 3, § 320 (last revised Mar. 21, 2022), 11 making AO
employees the sole Judiciary employees who can similarly, if
not prohibited, dilute its apolitical essence. 12 Further, the AO
as a body is involved in contentious debates related to the
federal judicial system—including, but not limited to, new
judgeships and judicial vacancies to be filled by the elected
branches, 13 pro se litigation 14 and judicial ethics 15—that have
11
Available at https://www.uscourts.gov/sites/default/files/
guide-vol02a-ch03.pdf.
12
On this point, I note that the restrictions at issue parallel those
applicable to federal judicial employees located in courthouses—
other than law clerks, who are barred from partisan and nonpartisan
activity—restrictions that were drafted and approved by a committee
of federal judges as well as the Judicial Conference, comprised of
federal judges and presided over by the Chief Justice of the United
States. See Duff Decl. ¶¶ 17–18, J.A. 140.
13
See John Gramlich, Federal judicial picks have become more
contentious, and Trump’s are no exception, Pew Rsch. (Mar. 7,
2018), https://www.pewresearch.org/fact-tank/2018/03/07/federal-
judicial-picks-have-become-more-contentious-and-trumps-are-no-
exception/ (explaining “rising discord in the federal judicial
nominations process”); see also Tara Leigh Grove, Sacrificing
Legitimacy in a Hierarchical Judiciary, 121 Colum. L. Rev. 1555,
1593–96 (2021) (explaining evolving partisan tensions characteristic
of federal judicial nominations).
14
See Andrew Hammond, The Federal Rules of Pro Se
Procedure, 90 Fordham L. Rev. 2689, 2700–02 (2022) (describing
opposing public attitudes towards pro se litigation—seen as either
essential to judicial system or frivolous waste of judicial resources—
and Judicial Conference’s role in shaping relevant policies).
15
The AO points to the public scrutiny of a Judicial Conference
proposal—with AO support—regarding federal judges’ membership
in law-related organizations like the Federalist Society and American
Constitution Society. See Appellant’s Br. 36; see also Editorial
Board, Judicial Political Mischief, Wall St. J. (Jan. 21, 2020),
https://www.wsj.com/articles/judicial-political-mischief-
11579652574; cf. Charles G. Geyh, The Architecture of Judicial
 14
political aspects and may invite partisan attacks, regardless of
the AO’s perceived obscurity.

B. Inter-Branch Interest

Next, I believe the AO’s Inter-Branch Interest—that AO
employees’ partisan political activities will negatively affect
the Judiciary’s reputation with members of the Congress—is
sufficiently grounded in reality.

The AO’s evidence shows that the AO’s effectiveness as
the Judiciary’s representative vis-à-vis the elected branches is
inextricably linked to its reputation for nonpartisanship.
Specifically, the AO provided affidavits from former
congressional staffers attesting to the “reputation for non-
partisanship and professionalism that the AO and its employees
hav[e] built up among members [of Congress] and their staff.”
Cooney Decl. ¶ 9, J.A. 164; see also Weich Decl. ¶ 11, J.A.
176–77 (“In my experience, the AO and its personnel . . .
advanced the Judiciary’s interests in a qualitatively different
manner, without any political or policy agenda other than
promoting the administration of justice in federal courts,
protecting the Judiciary’s independence from the political
branches, and upholding the rule of law.”). And these same
affidavits submit that, if AO employees were perceived to be
overtly partisan, the AO’s and Judiciary’s long-cultivated
reputation for nonpartisanship would be jeopardized. See
Cooney Decl. ¶ 10, J.A. 165 (“Were individual AO employees,
or the AO generally, to be perceived as engaged in partisan
political activity, I believe it could lead to increased
congressional skepticism toward the positions of the AO and
the Judicial Conference and possibly to reduced congressional
willingness to advance legislation in the best interests of the

Ethics, 169 U. Penn. L. Rev. 2351, 2353–55 (documenting public
debates over judicial ethics).
 15
federal judiciary, all to the detriment of the institution of the
Judiciary.”); Weich Decl. ¶ 12, J.A. 177 (“It would have been
difficult for members of Congress and their staffs, including
me, to maintain the perception of the AO and its employees as
non-partisan actors if AO employees were known to have
publicly engaged in partisan political activity, even in their
non-professional capacities.”). It is as likely that efforts by
partisan-aligned individuals to tarnish the public perception of
the AO and the Judiciary would similarly affect the perception
of members of the Congress and its staff.

Further, the harm resulting from a loss of congressional
confidence may be more tangible than that resulting from a
diminution of public confidence, as there is substantial and
direct interaction between the AO and the Congress. The AO,
like any federal agency, depends on the Congress for the
latter’s approval of its budget and accordingly provides
budgetary briefings to congressional members and their staff.
See Duff Decl. ¶ 14(d), J.A. 138–39; Cooney Decl. ¶ 4, J.A.
155–56. In addition, the AO regularly consults with and
testifies before the Congress—sometimes at the Congress’s
behest—on matters relating to the Judiciary, including
proposed legislation on matters ranging from cameras in
courtrooms to sentencing reform to new judgeships. See Duff
Decl. ¶ 14(d), J.A. 138–39; see also Weich Decl. ¶ 9, J.A. 175;
Cooney Decl. ¶¶ 6–7, J.A. 162–63. It is not implausible to fear,
as the AO does, that members of the Congress and their staff,
upon learning of AO employees’ partisan political activities,
will question the neutrality of the AO and the policies it
proposes. For example, the Congress may question whether the
Judiciary’s requests for additional judgeships to address
increasing caseloads are in fact intended to benefit the
particular political party or candidate that would fill new
 16
judgeships. 16 See Baugher Decl. ¶ 10, J.A. 158–59; see also
Duff Decl. ¶ 25, J.A. 144 (observing political parties’ goals of
“making Judicial appointments that advance the parties’
partisan objectives”). Or legislators may move to delay or
reduce the Judiciary’s budgetary requests in retaliation for
perceived partisanship on the part of the AO in setting its policy
priorities. See Baugher Decl. ¶ 10, J.A. 158–59. The
effectiveness and efficiency of the AO both as advocate for the
Judiciary and as neutral advisor to the Congress could be
placed at significant risk if its internal administrative and
policy aims were hindered by a distrustful relationship with the
Congress resulting from AO employees’ overtly partisan
political activities—and the examples discussed supra make
this fear entirely plausible.

C. Intra-Branch Interest

Finally, as with the first two interests, I believe the AO’s
Intra-Branch Interest—that partisan political activity risks
adversely affecting individual judges’ perceptions of the AO as
an impartial body—is sufficiently grounded to support the
challenged restrictions.

The AO serves as the Judiciary’s advisor, advocate and
administrator, independent of the other branches. See Chandler
v. Judicial Council of the Tenth Circuit of the U.S., 398 U.S.
74, 102 (1970) (Harlan, J., concurring) (describing AO as “an
arm of the judicial branch of government and under the direct
control of the Supreme Court and the Judicial Conference of
the United States”). AO employees work closely with scores of

16
Much ink has been spilled to document the rising—and
regrettable—politicization of the judicial nomination process,
including inferior court nominees. See supra note 13; see generally
Keith E. Whittington, Partisanship, Norms, and Federal Judicial
Appointments, 16 Geo. J.L. & Pub. Pol’y 521 (2018).
 17
individual federal judges, providing myriad services and
support. AO employees provide policy-making support for the
Judicial Conference—and its many committees, working
groups and advisory councils comprised of federal judges—on
subjects like court rules, probation and pretrial services,
defender services, PACER access, judicial security and judicial
salaries. See Duff Decl. ¶¶ 9–10, J.A. 132–34. They also draft
proposed legislation, amendments to federal rules of procedure
and budgetary proposals. Id. at ¶ 11, J.A. 134–36. They manage
the mechanisms for filing judicial misconduct and disability
complaints and update the codes of conduct applicable to
federal judges, id. at ¶ 10–11 J.A. 133–36, and supervise and
provide guidance regarding federal judges’ financial disclosure
reports, id. at ¶¶ 9, 11, J.A. 132, 134–36. The Judicial
Conference provides oversight of AO employees and can make
hiring, promotion and assignment recommendations. Id. at
¶¶ 11, 13, J.A. 134, 137.

This close working relationship on matters central to the
functioning of the Judiciary underscores the necessity of trust.
See Baugher Decl. ¶ 12, J.A. 160 (“As a longtime employee of
the Judicial Branch, I know that the AO’s work with judges is
built on the understanding and trust that AO employees are
offering objective, unbiased advice and assistance to the
judges.”). Loss of that trust stemming from concerns over an
AO employee’s political activity risks weakening the core
mission of the AO—to provide advice, support and
recommendations to and on behalf of the Judiciary. It cannot
benefit the Judiciary for judges to question, based on perceived
political leanings, the AO employees who provide advice on
judges’ attendance at seminars organized by outside
organizations. See Duff Decl. ¶ 33, J.A. 149. Or question the
neutrality of AO employees with roles in processing
complaints of judicial misconduct. See id. at ¶ 134, J.A. 149.
Or AO employees assigned to appear before the Congress on
 18
policy matters approved by the Judicial Conference and/or its
various committees comprised of federal judges. See id. at
¶ 135, J.A. 149–50.

Further, I do not agree with my colleagues’ implication
that we have to “assum[e] that judges will learn of AO
employees’ private activity.” Majority Op. at 14. With the
frequency with which AO employees and federal judges
interact—together with the increasing reach of social media,
where much political conduct occurs—it is not implausible that
judges will learn of a particular AO employee’s partisan
political activity. See Duff Decl. ¶¶ 28–30, J.A. 146–48.

Assuming the AO could properly provide evidence
reflecting the views of individual federal judges, as Guffey and
Smith suggest, see Appellees’ Reply Br. 6, it cannot be faulted
for declining to do so. I doubt that it is possible to gauge
accurately the effect on the Intra-Branch Interest that
implementation of the challenged provisions will have, short of
perhaps polling substantial members of the federal judiciary, a
proposal not only improbable but undoubtedly improper.
Nevertheless, the assurance that federal judges have enjoyed
for decades that partisanship is absent within the AO will
undoubtedly be lost if the challenged restrictions are
invalidated.

III. Sweep of the Challenged Restrictions

Finally, the “sweep” of the restrictions on protected
activity must be “reasonably necessary to protect the efficiency
of the public service.” Weaver, 87 F.3d at 1439 (quoting NTEU,
513 U.S. at 474); see also Sanjour v. EPA, 56 F.3d 85, 95 (D.C.
Cir. 1995) (en banc) (examining “fit between government’s
purported interest and sweep of its restrictions” (internal
quotation marks omitted)). I doubt my colleagues would
dispute that even an employee’s routine political activities—
 19
including campaign contributions and social media posts—can
be leveraged to mount a partisan attack against his employer.
See supra p. 9–10. Because I believe all of the outlawed
activities under the revised Code of Conduct, if allowed, could
threaten the Judiciary’s reputation, restricting all such activity
is reasonably necessary to protect that reputation.

I find the reasoning relied upon by the district court and
offered by Guffey and Smith unpersuasive. First, the district
court concluded that the AO failed to demonstrate that the Code
of Conduct’s existing prohibition on using “position, title, or
authority in connection with” partisan political activity,
coupled with after-the-fact disciplinary proceedings, is
insufficient to address threats to the Judiciary’s reputation.
Guffey II, 459 F. Supp. 3d at 250–51. But that vague
proscription does not specify prohibited conduct nor provide
sufficient means of redress. As the AO points out, disciplinary
proceedings would serve only to police “clearly inappropriate
conduct” but ignore “normal partisan conduct,” like campaign
contributions within legal limits or social media posts, both of
which can have similarly deleterious effects on the perception
of the Judiciary as an impartial, nonpolitical body. See
Appellant’s Br. 56–57; Appellant’s Reply Br. 42–45. Further,
as the AO notes, an ad hoc disciplinary investigation “may
itself be cast as partisan.” Appellant’s Br. 57; see also Cooney
Decl. ¶ 18, J.A. 170 (“[I]t is not unlikely that . . . a disciplinary
proceeding itself could become politicized by members of
Congress who agree with the employee’s partisan message and
attempt to exploit the AO’s disciplinary action as evidence that
the AO is itself promoting a particular partisan view.”).

Second, the district court found the restrictions
underinclusive because they do not resolve the potential
consequences of an employee’s past partisan activity like pre-
employment campaign contribution disclosures or social media
 20
posts. Guffey II, 459 F. Supp. 3d at 251. But I fail to see how
the AO could legitimately police employees’ pre-employment
activity; further, it is well-settled that the AO “need not address
all aspects of a problem in one fell swoop,” especially ones
unrelated to its employees’ conduct while they are employed
by it. Williams-Yulee, 575 U.S. at 449; see also Burson v.
Freeman, 504 U.S. 191, 207 (1992) (“We do not, however,
agree that the failure to regulate all speech renders the statute
fatally underinclusive.”). And Guffey and Smith’s suggestion
that the revised Code of Conduct is underinclusive in that it
excludes political activity such as reading political books on
the bus or publicly viewing a political film, see Appellees’ Br.
53–54, is ludicrous on its face. 17

***

In a city that is ground zero for political expression, it may
be folly to attempt to restrict—let alone prohibit—
it. Notwithstanding the Administrative Office’s location,
however, the Constitution demands that the apolitical essence
of the Judiciary and all members of its workforce—wherever
sited—remain uncompromised.

For the foregoing reasons, I respectfully dissent.

17
I find their over-inclusiveness, see id. at 50–51, and Hatch
Act, see id. at 51–52, arguments equally meritless.