AT&T Services, Inc. v. FCC

United States Court of Appeals
FOR THE DISTRICT OF COLUMBIA CIRCUIT

Argued September 17, 2021 Decided December 28, 2021

No. 20-1190

AT&T SERVICES, INC.,
PETITIONER

v.

FEDERAL COMMUNICATIONS COMMISSION AND UNITED
STATES OF AMERICA,
RESPONDENTS

APPLE INC., ET AL.,
INTERVENORS

Consolidated with 20-1216, 20-1272, 20-1274, 20-1281,
20-1284

On Petitions for Review of an Order
of the Federal Communications Commission

Jonathan E. Nuechterlein argued the causes for petitioners
Joint Issues. Mark Reddish argued the causes for petitioner
APCO. With them on the joint briefs were Jeffrey S. Cohen, C.
Frederick Beckner III, Rick Kaplan, Jerianne Timmerman,
Craig A. Gilley, Mitchell Y. Mirviss, Elizabeth C. Rinehart, and
Russell P. Hanser. Michele Farquhar, Brett Kilbourne, Jay
 2
Morrison, Brian W. Murray, Delia D. Patterson, Christopher
T. Shenk, and Ian D. Volner entered appearances.

Trey Hanbury and Jessica L. Ellsworth were on the brief
for amicus curiae Southern Company Services, Inc. in support
of petitioners.

James M. Carr, Counsel, Federal Communications
Commission, argued the cause for respondents. With him on
the brief were Daniel E. Haar and Robert J. Wiggers,
Attorneys, U.S. Department of Justice, Jacob M. Lewis,
Associate General Counsel, Federal Communications
Commission, and Thaila K. Sundaresan, Counsel. Richard K.
Welch, Deputy Associate General Counsel, and Adam Crews,
Counsel, entered appearances.

Christopher J. Wright argued the cause for intervenors. With
him on the joint brief were David Paul Murray, Russell H. Fox,
Robert G. Kidwell, Paul J. Caritj, and Jason Neal in support of
respondents. Rick C. Chessen and Neal M. Goldberg entered
appearances.

Matthew A. Brill and Matthew T. Murchison were on the
brief for amicus curiae Cable Television Laboratories, Inc. in
support of respondents.

Andrew Jay Schwartzman and Harold Feld were on the
brief for amici curiae Public Knowledge, et al. in support of
respondents.
 3
Before: TATEL, MILLETT, and WALKER, Circuit Judges.

Opinion for the Court filed by Circuit Judge TATEL.

TATEL, Circuit Judge: By order dated April 24, 2020, the
Federal Communications Commission opened the 6 gigahertz
(GHz) band of radiofrequency spectrum to unlicensed
devices—routers and the devices they connect to, such as
smartphones, laptops, and tablets. In doing so, the Commission
required that such unlicensed devices be designed and operated
to prevent harmful interference with licensees now using the
6 GHz band, i.e., commercial communications providers,
electric utilities, public safety services, and network
broadcasters. Those licensees, emphasizing that existing uses
of the band involve vital public safety and critical
infrastructure, argue that harmful interference could
nonetheless occur and that the Order therefore runs afoul of
both the Communications Act of 1934 and the Administrative
Procedure Act. But as explained in this opinion, petitioners
have failed to provide a basis for questioning the Commission’s
conclusion that the Order will protect against a significant risk
of harmful interference, just the kind of highly technical
determination to which we owe considerable deference. We
therefore deny the petitions for review in all respects save one.
The exception relates to the petition brought by licensed radio
and television broadcasters using the 6 GHz band. Because the
Commission failed adequately to respond to their request that
it reserve a sliver of that band exclusively for mobile licensees,
we remand to the Commission for further explanation on that
point.

I.
Many users of the radiofrequency spectrum operate by
transmitting information through microwaves—short waves of
890 megahertz (MHz) or higher. See 47 C.F.R. § 101.3
 4
(defining microwave frequencies). To prevent such users from
interfering with one another, the Federal Communications
Commission, pursuant to its authority under the
Communications Act of 1934, 47 U.S.C. § 151 et seq.
(“Communications Act”), awards licenses to operate in specific
frequency ranges, or “bands.” See id. §§ 151, 301 (creating the
Commission to carry out the Act’s provisions and providing for
licensing).

Historically, the 6 GHz band, comprising frequencies
between 5.925 and 7.125 GHz, has been reserved for licensed
users that “support a variety of critical services provided by
utilities, commercial and private entities, and public safety
agencies.” Unlicensed Use of the 6 GHz Band; Expanding
Flexible Use in Mid-Band Spectrum Between 3.7 and 24 GHz,
35 FCC Rcd. 3852, 3855 ¶ 7 (2020) (“Order”). Some of these
licensees transmit signals through a “fixed-microwave
system,” in which “a transmitter on one tower beams 6 GHz
signals to a receiver on another tower within its line of sight.”
Pet’rs’ Br. 3. Fixed microwave systems support “emergency
911 dispatch and other public safety operations,” id.;
“commercial wireless providers,” Order ¶ 7; and “links for
coordination of railroad train movements, control of natural gas
and oil pipelines, management of electric grids, and long-
distance telephone service,” id. In addition to fixed microwave
users, other 6 GHz band licensees operate on a mobile basis.
They employ transmitters and receivers affixed to portable
bases, like news vans and broadcasting cameras, and send
programing from remote locations back to studios. Still others
employ mobile transmitters to support wireless microphones
and backstage communications.

Several decades ago, the Commission, charged by
Congress to “generally encourage the larger and more effective
use of” the spectrum, 47 U.S.C. § 303(g), opened the 2.4 GHz
 5
and certain other bands to unlicensed radiofrequency
transmitters. Today, these devices include routers and the
smartphones, laptops, and tablets they support. Such devices,
however, must refrain from causing “harmful interference”
with licensed users. 47 C.F.R. § 15.5(b)–(c); see also
Additional Spectrum for Unlicensed Devices Below 900 MHz
and in the 3 GHz Band, No. 02-380, FCC-02-328 ¶¶ 3–4 (Dec.
11, 2002) (describing the history of unlicensed operation).
Commission regulations define “harmful interference” as
interference that “endangers the functioning of a radio
navigation service or of other safety services or seriously
degrades, obstructs or repeatedly interrupts a
radiocommunications service.” 47 C.F.R. § 15.3(m). If harmful
interference occurs, the Commission may order the interfering
user to cease operations. Id. § 15.5(c) (“The operator of a radio
frequency device shall be required to cease operating the
device upon notification by a Commission representative that
the device is causing harmful interference.”); see also id.
§ 15.15(c) (“[O]perators [of unlicensed devices] are required to
cease operation should harmful interference occur to
authorized users.”).

The Commission’s opening of the radiofrequency
spectrum for unlicensed uses has taken on new import in recent
years because of a boom in unlicensed devices that use Wi-Fi
and Bluetooth technology. See Order ¶ 1. Such devices include
internet “access points” (e.g., routers) and the myriad “client
devices” that connect to them, like smartphones, tablets, and
laptops. Id. ¶¶ 3, 12. Because these devices transmit large
amounts of data, they require access to wide bands of the
spectrum. “The demand for wireless broadband,” according to
the Commission, “continues to grow at a phenomenal pace;”
by 2024, a smartphone’s average data use is projected to grow
almost sixfold relative to 2018 data levels. Id. ¶ 2.
 6
In 2017, responding to this growing demand, the
Commission announced that it was considering opening a
portion of spectrum between 3.7 and 24 GHz to unlicensed use
and sought public comment. Expanding Flexible Use in Mid-
Band Spectrum Between 3.7 and 24 GHz, 32 FCC Rcd. 6373
(2017). The following year, the Commission proposed a rule
that would open the 6 GHz band to unlicensed devices, again
seeking public comment. Unlicensed Use of the 6 GHz Band;
Expanding Flexible Use in Mid-Band Spectrum Between 3.7
and 24 GHz, 33 FCC Rcd. 10496 (2018). The Commission
chose the 6 GHz band in part because of its proximity and
similarity to the 5 GHz band, portions of which already allowed
unlicensed use. Opening the adjacent 6 GHz band would allow
unlicensed devices to “operate with wider channel bandwidths
and higher data rates with increased flexibility.” Id. ¶ 14; see
also id. ¶ 19. After considering comments, the Commission, at
an open meeting on April 23, 2020, adopted the Order now
before us. See generally Order, 35 FCC Rcd. 3852.

The Order allows unlicensed devices to operate in the
6 GHz band. Because the extent to which a signal may cause
interference depends in part on the signal’s power, the Order
distinguishes between internet access points that use standard
power (like the devices that provide internet to stadiums,
concert halls, and other large areas) and access points that use
low power (like typical residential or office routers).

The Order requires all standard-power access points to use
an automated frequency coordination (AFC) system, a
technology designed to ensure that unlicensed devices do not
cause harmful interference with licensed devices. Id. ¶ 17. But
because the AFC system requires knowing the “exact operating
locations and times” of licensed uses, it offers little protection
to licensed mobile operators, whose location “can change
frequently.” Id. ¶ 93. For that reason, the Order prohibits
 7
unlicensed standard-power access points from using those
6 GHz sub-bands in which mobile licensees operate.

By contrast, the Order allows unlicensed low-power
access points to operate across the 6 GHz band. But to protect
licensed users from harmful interference, the Order requires
that routers (1) operate below specified maximum power
levels—as relevant here, 5 decibel milliwatts per megahertz (5
dBm/MHz); (2) use a “contention-based protocol,” through
which a device “listens” to a channel to ensure it is free before
transmitting a signal over it; and (3) remain indoors, thus
decreasing the likelihood of interference with licensed outdoor
users. Smartphones, laptops, and other client devices using
these low-power access points must observe an even lower
maximum power limit and employ contention-based protocol
technology. To discourage the outdoor use of low-power
routers, the Order (1) prohibits making them weather-resistant,
(2) requires that they have integrated antennas, and (3) forbids
equipping them with batteries. These multifaceted protections,
the Commission concluded, “eliminate[] any significant risk of
causing harmful interference” with licensed users. Id. ¶ 146.

Petitioners either hold licenses to operate in the 6 GHz
band or represent entities that do. Specifically, petitioners are
commercial communications providers AT&T Services and
Lumen Technologies, electric utilities, the Association of
Public-Safety Communications Officials International
(APCO), and the National Association of Broadcasters.
Petitioners contend that the Order fails to protect licensees
from harmful interference and therefore runs afoul of both the
Communications Act and the Administrative Procedure Act
(APA), 5 U.S.C. § 706(2)(A). They urge us to vacate the Order
and remand to the Commission to implement further
safeguards. Several industry groups and companies, including
Apple, Broadcom, and Cisco Systems, which manufacture
 8
devices or provide services that rely on unlicensed spectrum,
have intervened to defend the Order.

II.
Fundamental and longstanding principles of
administrative law guide our review of petitioners’ challenges
to the Commission’s order. To demonstrate that a regulation is
arbitrary and capricious, a challenger must show that the
agency “relied on factors which Congress has not intended it to
consider, entirely failed to consider an important aspect of the
problem, offered an explanation for its decision that runs
counter to the evidence before the agency, or is so implausible
that it could not be ascribed to a difference in view or the
product of agency expertise.” Motor Vehicle Manufacturers
Ass’n v. State Farm Mutual Automobile Insurance Co., 463
U.S. 29, 43 (1983). Where, as here, the Commission “‘foster[s]
innovative methods of exploiting the spectrum,’ it ‘functions
as a policymaker’ and is ‘accorded the greatest deference by a
reviewing court.’” Mobile Relay Associates v. FCC, 457 F.3d
1, 8 (D.C. Cir. 2006) (quoting Teledesic LLC v FCC, 275 F.3d
75, 84 (D.C. Cir. 2001)). Of course, “we do not hear cases
merely to rubber stamp agency actions.” Natural Resources
Defense Council, Inc. v. Daley, 209 F.3d 747, 755 (D.C. Cir.
2000). But to survive judicial review, the Commission’s
technical judgments need rest upon only “‘a modicum of
reasoned analysis,’ ‘absent highly persuasive evidence to the
contrary.’” Mobile Relay Associates, 457 F.3d at 8 (quoting
Hispanic Information & Telecommunications Network, Inc. v.
FCC, 865 F.2d 1289, 1297–98 (D.C. Cir. 1989)). Moreover,
the Commission’s “‘predictive judgments about areas’” within
its “‘discretion and expertise are entitled to particularly
deferential review, as long as they are reasonable.’” EarthLink,
Inc. v. FCC, 462 F.3d 1, 12 (D.C. Cir. 2006) (emphasis
omitted) (quoting In re Core Communications, Inc., 455 F.3d
267, 282 (D.C. Cir. 2006)).
 9
A.
All petitioners argue that the Commission has understated
the risk of harmful interference. Central to this argument,
petitioners claim that the Commission intended the Order to
eliminate all risk of harmful interference. “[I]n the
[Commission’s] view,” petitioners assert, the Order
“eliminate[s] any ‘significant risk’ that any of the hundreds of
millions of unlicensed 6 GHz devices will cause harmful
interference to any of the nation’s nearly 100,000 licensed
microwave links, at any point in the foreseeable future.” Pet’rs’
Br. 21. Dissecting the studies upon which the Commission
relied and critiquing the Order’s safeguards for licensed users,
petitioners argue that some harmful interference will occur at
some point, thus rendering the Order arbitrary and capricious.

Petitioners mischaracterize the Commission’s goal. It
never claimed that the Order would reduce the risk of harmful
interference to zero. To the contrary, the Commission
repeatedly explained that the Order makes the “potential for
harmful interference to incumbent services operating in the
6 GHz band . . . insignificant.” Order ¶ 104 (emphasis added);
see also id. ¶¶ 110, 122 n.317, 145–46, 245 (repeatedly
characterizing such risk as low or insignificant, not zero). The
Commission acknowledged that it had to “balance unlicensed
device access and incumbent protection,” id. ¶ 63, and
explained that “in the unlikely event that harmful interference
does occur,” “the Commission’s Enforcement Bureau has the
ability to investigate reports of such interference and take
appropriate enforcement action as necessary,” id. ¶ 149. This
aligns perfectly with existing Commission regulations, which
(1) acknowledge that full compliance “will not prevent harmful
interference under all circumstances” and (2) authorize the
Commission to order interfering users to cease operations. 47
C.F.R. § 15.15(c); see also supra at 5.
 10
B.

Petitioners argue that the Order falls short in other ways.
We start with petitioners’ contention that the Commission
should have estimated the frequency and cost of harmful
interference. In support, they cite Competitive Enterprise
Institute v. National Highway Traffic Safety Administration, in
which our court remanded a fuel efficiency rule to the agency
for failing to consider the standard’s impact upon car size and
safety. 956 F.2d 321 (D.C. Cir. 1992). But unlike the agency in
Competitive Enterprise, the Commission “conduct[ed] a
serious analysis of the data,” revealing the likelihood of harm
to be vanishingly low. Id. at 327. Moreover, even if harmful
interference does occur, its victims may petition the
Commission for relief. See Order ¶ 149 & n.397 (discussing
the Commission’s ability to respond to complaints); see also
supra at 5.

Accusing the Commission of a “textbook APA violation,”
petitioners argue that the agency failed to explain why it did
not require low-power devices to use an AFC system, as must
standard-power devices. Pet’rs’ Reply Br. 6–7. Oral Arg. Tr.
4–5. We disagree. In the Order, the Commission explained that
the requirements for low-power devices (power limits,
contention-based protocol, and indoor operation), together
with Commission enforcement authority, “reduce[] the
possibility of harmful interference to the minimum that the
public interest requires.” Order ¶ 146; see also id. ¶¶ 147–50
(acknowledging comments seeking an AFC system and
explaining why the Order’s other requirements reduce the risk
of harmful interference to an acceptably low level). Put
differently, the Commission concluded that even without an
AFC system, “the restrictions and requirements . . .
establish[ed] for indoor use of low-power access points
 11
eliminates any significant risk of harmful interference.” Id.
¶ 146; see also id. ¶¶ 147–50.

Petitioners also raise several technical objections to the
principal study on which the Commission relied. A simulation
submitted by Cable Television Laboratories (“CableLabs”), a
nonprofit supporting broadband providers, the study models
the likelihood that hypothetical unlicensed 6 GHz devices
scattered across New York City will interfere with transmission
from a local microwave tower. To simulate a city filled with
unlicensed routers, the study uses what is known as Monte
Carlo analysis. Id. ¶ 117. Developed by scientists working on
the Manhattan Project, Monte Carlo analysis differs from more
traditional mathematical models in how it accounts for
variables. Nick T. Thomopoulos, Essentials of Monte Carlo
Simulation: Statistical Methods for Building Simulation
Models 1 (2013). Here the variables are those factors that affect
a router’s transmission, such as its power, location, and
frequency range. While traditional models select a single value
(e.g., an average) for each variable, Monte Carlo analysis uses
a range of possible values for each variable, runs hundreds of
simulations, and produces a range of possible outcomes. In
situations where “interactions between the possible outcomes
become [exceptionally] complex,” Monte Carlo analysis can
provide a “more complete view of potential outcomes and their
associated likelihoods.” Federal Judicial Center & National
Research Council, Reference Manual on Scientific Evidence
469 (3d ed. 2011) (first quote); CableLabs Amicus Br. 7–8
(second quote).

The CableLabs study uses sales projections and statistical
distributions drawn from real-world data and industry
standards to analyze the effect of approximately 800,000
hypothetical routers on a microwave tower in New York City.
CableLabs ran 1,500 simulations, generating data on more than
 12
1.2 billion hypothetical routers. Not one of the hypothetical
routers caused harmful interference.

Calling the study a “black box,” petitioners argue that the
Commission should have made available “spreadsheets,
formulas, detailed datasets, and transparent explanations of
how those datasets were obtained.” Pet’rs’ Br. 14. In support,
they cite American Radio Relay League, Inc. v. FCC, in which
we faulted the Commission for cherry-picking the data it
disclosed. 524 F.3d 227, 237 (D.C. Cir. 2008). Here, by
contrast, the Commission disclosed all data in its possession.
The Commission, moreover, emphasizes that CableLabs’s
submission was “typical [of] FCC proceedings”—that is, it
“presented the results of its study by describing the sample size,
simulation parameters, methodology, and results.”
Commission Br. 46. According to the Commission, such
information, not raw data, “allows parties to meaningfully
comment.” Id. at 46–47. Indeed, our court has explained that
“requiring agencies to obtain and publicize the data underlying
all studies on which they rely would be impractical and
unnecessary.” American Trucking Ass’ns, Inc. v. EPA, 283
F.3d 355, 372 (D.C. Cir. 2002) (internal quotation marks
omitted).

Petitioners also criticize the CableLabs study for ignoring
those rare cases when a router’s signal might experience zero
“building loss,” a variable that measures the extent to which a
building’s characteristics, such as insulation and wall
thickness, weaken router signals. Pet’rs’ Br. 45–48. Although
the Commission acknowledged that “it would be more
appropriate for CableLabs to have used the full statistical
distribution” of building loss values, it nonetheless concluded
that the range used in the CableLabs study was “not different
enough from the [full] statistical distribution to materially alter
the likelihood of harmful interference occurring.” Order ¶ 122.
 13
Petitioners have offered no reason for us to depart from our
court’s longstanding practice of according “considerable
deference” to the Commission’s expertise on such a “highly
technical question.” American Radio Relay, 524 F.3d at 233
(internal quotation marks omitted).

Petitioners next criticize the CableLabs study for assuming
an “average activity factor of 0.4%,” meaning that routers
“transmit only one minute out of every 250.” Pet’rs’ Br. 43
(internal quotation marks omitted). “Where,” they ask, “does
that strikingly low figure come from?” Id. Answering that
question in the Order, the Commission explained that the
CableLabs study uses “a distribution of airtime utilization
based on data taken from 500,000 Wi-Fi access points to model
how often each access point in the simulation transmits”—the
average activity factor was 0.4%. Order ¶ 117; see also
CableLabs Ex Parte Letter on AT&T’s Comment Letter (Feb.
14, 2020), at 1–2 (explaining the source of this figure). True,
this does seem low, but as intervenors explained at oral
argument, routers, especially those operating in the 6 GHz
band, transmit huge amounts of data in “really tiny burst[s].”
Oral Arg Tr. 52. For example, counsel explained, in “a matter
of seconds or less,” a router can receive the data necessary to
enable “watching [a video] for two hours.” Id. Determining a
router’s activity factor “is precisely the type of technical issue
on which we defer to the Commission’s expertise,” Keller
Communications, Inc. v. FCC, 130 F.3d 1073, 1077 (D.C. Cir.
1997), especially “absent highly persuasive evidence” from
petitioners that routers have a higher activity factor than the one
used by CableLabs, Mobile Relay Associates, 457 F.3d at 8
(internal quotation marks omitted).

Last, petitioners fault the CableLabs study for assuming
that “1500 snapshots in time provide[] a sample sufficient for
drawing a statistically sound conclusion that harmful
 14
interference will never occur.” Pet’rs’ Br. 48. But as explained
above, the Commission never said that no harmful interference
would occur; it concluded only that, given the Order’s
safeguards, “the potential for harmful interference to
incumbent services operating in the 6 GHz band is
insignificant.” Order ¶ 104 (emphasis added).

Petitioners’ next set of challenges centers on the
Commission’s rerun of a study prepared and submitted by
petitioner AT&T. That study identifies several buildings with
direct lines of sight to various microwave towers and assumes
that the buildings contain 6 GHz routers that might interfere
with the towers’ signals. Unlike the Monte Carlo analysis used
in the CableLabs study, the AT&T study, as originally
designed and submitted to the Commission, selects single,
worst-case values for all but one variable—that is, values likely
to cause harmful interference. The AT&T study concludes that
hypothetical routers could interfere with a microwave tower in
every case.

The Commission discounted that conclusion because the
AT&T study uses worst-case scenarios and so does not “rebut
the persuasive showing by CableLabs based on a reliable
probabilistic assessment derived from measurements
associated with hundreds of thousands of actual Wi-Fi [access
points].” Id. ¶ 130. Notwithstanding the Commission’s
preference for Monte Carlo analyses, it reran the AT&T study
“to show that even under AT&T’s preferred mode of analysis
. . . the likelihood of harmful interference [is] insignificant.”
Id. ¶ 127 n.331. To accomplish this, the Commission revised
the AT&T study in several respects, two of which are relevant
here. First, it modified how the AT&T study deals with
building loss, the one variable for which that study uses a range
of values. Because the Commission believed that treating
building loss differently from all other variables undermined
 15
the study’s accuracy, it replaced the range with a single,
average value. Second, for two of the six scenarios, the
Commission substituted what it believed to be more reasonable
values for “clutter loss,” signal attenuation caused by terrain,
trees, and other structures. Id. ¶ 124. Thus modified, the AT&T
study demonstrates that only one of the six scenarios could
result in “a nontrivial possibility of harmful interference,” and
the Commission discounted even that because it did “not
believe this one case poses a significant potential for actual
harmful interference.” Id. ¶ 131.

Petitioners criticize the Commission for using an average
value instead of a statistical distribution for building loss and
for failing to respond to comments on this subject. But the
Commission did respond, explaining that treating only building
loss “as a probabilistic quantity while not considering all the
other [relevant] statistical quantities” exaggerated the
likelihood of interference. Order ¶ 127. Petitioners quibble
with this conclusion, but they have given us no real basis for
second-guessing the Commission’s analysis, which, as in much
of this case, “requires a high level of technical expertise”
meriting deference to the Commission’s “informed discretion.”
Marsh v. Oregon Natural Resources Council, 490 U.S. 360,
377 (1989) (internal quotation marks omitted).

Next, petitioners fault the adjustments the Commission
made for clutter loss. According to AT&T, it selected the six
case studies precisely because the towers all had a direct line
of sight to at least one building assumed to have one or more
routers, which meant that “clutter loss approached zero.”
Pet’rs’ Br. 34. The Commission, however, explained that it
found the assumption of zero clutter loss unrealistic for the two
scenarios in which the tower and the router were more than one
kilometer apart. “Based on [its] experience,” the Commission
explained, AT&T’s model “drastically underpredicts [clutter]
 16
loss for longer distances because, as a practical matter, there is
almost always interaction with the environment that reduces
the signal level.” Order ¶ 67. Besides, as noted above, the
Commission explained that because the AT&T study uses
worst-case scenarios, it does not “rebut the persuasive showing
by CableLabs” that the likelihood of harmful interference is
insignificant. Id. ¶ 130.

C.
In addition to challenging the CableLabs study and the
Commission’s rerun of the AT&T study, petitioners challenge
the Order’s requirements for low-power access points—that
they not exceed a power limit of 5 dBm/MHz, that they be
equipped with contention-based protocol technology, and that
they operate only indoors.

We begin with power limits, which the Order sets at
5 dBm/MHz. According to petitioners, the Commission
“plucked [that figure] out of thin air” and failed to “cite
evidence . . . [for] pegging the power level to 5, rather than
(say) 3 or 1.” Pet’rs’ Br. 51. Quite to the contrary, the
Commission chose 5 dBm/MHz “[b]ased on [its] experience
with unlicensed operations and interference analyses,”
including using that precise power limit when it reran the
AT&T study and found an insignificant risk of harmful
interference. Order ¶ 110. Relying on its “engineering
judgment,” the Commission concluded that 5 dBm/MHz “will
both adequately protect all incumbents in the band from
harmful interference as well as offer enough power to
unlicensed devices, commensurate with the levels in . . . other
. . . bands.” Id.

Petitioners claim that contention-based protocol
technology allows low-power devices to detect only devices
like themselves that “transmit in all directions at once” and thus
 17
offers no protection to licensed fixed microwave users that
send “focused point-to-point beam[s].” Pet’rs’ Reply Br. 27.
But the Commission never claimed that contention-based
protocol would directly protect microwave towers from
interference. Instead, it explained, “our rule requiring that low-
power indoor access points employ a contention-based
protocol ensures that none of these unlicensed devices will
employ continuous transmissions,” thus making the occurrence
of harmful interference “even less likely.” Order ¶ 141; see
also id. n.374.

Petitioners contend that even if power limits and
contention-based protocol technology could protect licensees
from indoor low-power devices, these precautions will fall
short when such devices inevitably operate outdoors—for
example, when people take their routers outside to conduct
Zoom calls on their balconies. Fully aware of that risk, the
Commission imposed several requirements to make outdoor
use “impractical and unsuitable.” Id. ¶ 108. Specifically, it
required that routers have incorporated antennas, no batteries,
and no weather-resistant capability. Petitioners insist that
“[a]lthough these measures might help discourage outdoor use
. . . they cannot possibly prevent it.” Pet’rs’ Br. 53. But again,
petitioners are measuring the Order against a standard the
Commission never embraced; as explained above, the Order
does not seek to reduce the risk of harmful interference to zero.
Rendering outdoor router use impractical, as petitioners
concede the Order does, rather than impossible, promotes the
Commission’s goal of making the risk of harmful interference
“insignificant.” Order ¶ 104.

Petitioners argue that client devices, like smartphones and
laptops, will interfere with licensed users when operating
outdoors. Equally aware of this risk, the Commission imposed
power limits on client devices to “ensure that [they] remain in
 18
close proximity to the indoor access points.” Id. ¶ 103. By
doing so, the Commission “authorize[d] indoor unlicensed
devices with adequate power to be useful to the public while
also protecting the licensed services in the 6 GHz band from
harmful interference.” Id.

D.
Petitioners’ remaining arguments are equally without
merit. They contend that the Commission arbitrarily rejected
two studies that analyze situations with low clutter loss. The
Commission, however, offered perfectly reasonable
explanations for rejecting each. See id. ¶ 133 (“We have
conducted a similar analysis of the [Cellular
Telecommunications Industry Association] study as we did
with AT&T’s study and arrived at similar results.”); id. ¶ 154
(disagreeing with assumptions in the National Association of
Broadcasters’ study, including that devices will have direct
lines of sight to news gathering receivers and that the threshold
for harmful interference is -10 dB). Petitioners disagree with
the Commission’s view of worst-case assumptions in these and
other studies, but disagreement by itself is insufficient to
demonstrate that the Commission failed to “examine the
relevant data and articulate a satisfactory explanation for its
action including a ‘rational connection between the facts found
and the choice made.’” State Farm, 463 U.S. at 43 (quoting
Burlington Truck Lines v. United States, 371 U.S. 156, 168
(1962)).

Petitioners criticize the Commission for declining to
impose a mandatory maximum activity factor for unlicensed
devices. As indicated above, however, the Commission
explained that “requiring [such] devices to use a contention-
based protocol . . . will prevent [them] from transmitting”
continuously, even though “the adopted rules do not [directly]
limit the activity factor.” Order ¶ 120.
 19
Finally, petitioners argue that the Order fails to “create[]
an effective mechanism for immediately detecting, identifying,
and turning off any device that . . . cause[s] harmful
interference to licensed operations.” Pet’rs’ Br. 74. According
to petitioners, the Commission’s “post-hoc enforcement
mechanisms are designed to locate interference caused by
pirate radio transmitters or enterprise-grade machinery, not the
types of portable, sporadically transmitting consumer devices
bought by hundreds of millions of” people and kept on private
property. Id. at 75. The Commission disagreed, assuring
licensed users that its “Enforcement Bureau has the ability to
investigate reports of such interference and take appropriate
enforcement action.” Order ¶ 149. To accomplish this, the
Commission will rely on field agents with “fixed, vehicular-
mounted, and portable commercial and specialized spectrum
equipment to conduct investigations” and “work[] with entities
at the federal, state, county, and local levels of government” to
stop interference. Order ¶ 149 n.397. Petitioners have given us
no basis for second-guessing this “predictive judgment[]. . .
within the agency’s field of discretion and expertise.”
EarthLink, 462 F.3d at 12 (internal quotation marks omitted).
Should it turn out that the Enforcement Bureau is not up to the
task, petitioners can return to the Commission for relief.

III.
In addition to petitioners’ shared challenges to the
adequacy of the Order’s safeguards against harmful
interference, three groups of petitioners bring individual
claims.

A.
APCO, representing public safety operators, argues that
the Commission failed to consider the Order’s impact on 911
dispatch and other public safety services. In support, it cites our
 20
court’s recent decision in Mozilla Corporation v. FCC, where
we faulted the Commission for its “failure to consider the
implications for public safety of its changed regulatory posture
in [a] 2018 Order.” 940 F.3d 1, 59 (D.C. Cir. 2019). There,
however, the Commission failed entirely to consider public
safety. Here, the Commission expressly acknowledged that
public safety services use the 6 GHz band and adopted many
of the very safeguards APCO sought. See Order ¶ 7 (noting that
public safety services operate in the 6 GHz band); see also id.
¶¶ 30, 39–40, 46, 81–83, 187–88 (adopting various safeguards
that APCO suggested). APCO insists that the Commission
could have done more, but it identifies no “‘fail[ure] to
consider an important aspect of the problem.’” Mozilla, 940
F.3d at 59 (quoting State Farm, 463 U.S. at 43).

APCO also challenges the Order’s approach to unlicensed
standard-power devices—access points that provide broadband
to large areas like stadiums, concert halls, and shopping malls.
The Order requires that such devices, prior to transmitting,
consult a centralized AFC system to determine available
frequencies and maximum permissible power levels. APCO
complains that this system will be effective only 95% of the
time. The Commission, however, determined that based on its
experience with other devices, an AFC system with a 95%
confidence level will sufficiently protect licensees. “Our
experience with this rule,” the Commission explained,
“confirms that [such a confidence level] reliably ensures
protection against harmful interference, at reasonable cost.”
Order ¶ 41. This is just the kind of “predictive judgment[]
about areas . . . within the agency’s field of discretion and
expertise . . . entitled to particularly deferential review.”
EarthLink, 462 F.3d at 12 (internal quotation marks and
emphasis omitted).
 21
APCO next argues that AFC systems are unable to protect
public safety services that operate on a temporary emergency
basis in the wake of hurricanes and other major disasters. The
Order, however, contains provisions designed to protect
against that very possibility. Specifically, it instructs such
licensees to “register the details of their [temporary]
operation,” including temporary “transmitter and receiver
location,” so that the AFC system can “protect [such licensees]
from harmful interference.” Order ¶ 32. “Because temporary
fixed links are not mobile and intended to operate at a specified
location for up to a year,” the Commission explained, “we do
not believe this registration requirement poses a significant
burden on licensees.” Id.

Finally, APCO doubts that the Commission’s enforcement
authority is adequate to protect licensees from interference
from standard-power access points. But the Order includes
several measures designed to ensure that the Commission can
detect and end just such interference. See Order ¶ 83 (listing
requirements for AFC operators to facilitate enforcement). We
have no more basis for questioning the Commission’s
judgment about its ability to stop harmful interference from
standard-power access points than we did with respect to its
ability to stop interference from low-power devices. See supra
at 19.

B.
Electric utility petitioners argue that the Commission
unreasonably dismissed two studies on which they relied to
show that unlicensed low-power devices will interfere with
licensed users. With respect to one of the studies, submitted by
Southern Company Services, we agree with petitioners that the
Commission seems to have mischaracterized the study’s
treatment of clutter loss and ignored their clarifying comments.
The Order, however, cites other perfectly sound reasons for
 22
rejecting the study—in particular, the Commission’s
preference for Monte Carlo analyses. Order ¶ 135 & n.345. The
other study, a Critical Infrastructure Industry analysis, does
employ Monte Carlo methodology, but, as the Commission
explained, it relied on several unreasonable assumptions about
the demand unlicensed devices place on the 6 GHz band. See
id. ¶ 138 (listing unreliable assumptions, including that “every
man, woman, and child living in the Houston area” would be
using their own access points at the same time for a 4K video
streaming service).

Petitioners next argue that the Commission failed to
respond to comments about the Southern and Critical
Infrastructure Industry studies that they submitted in response
to a draft order that the Commission circulated three weeks
before its April 23, 2020 open meeting. See supra at 6. As for
the Southern Study, the Commission acknowledged
petitioners’ comments and explained why it nonetheless found
the study less reliable than Monte Carlo simulations. Order
¶ 135 n.345. As for the Critical Infrastructure Industry study,
the Commission explains in its brief that the utilities’ April
comments “merely repeated arguments [they] had made in
‘technical submissions’ that were previously placed in the
record” and to which the Commission had already responded.
Commission Br. 70–71; see also Order ¶ 138 n.364
(responding to utilities’ earlier submission); Thompson v.
Clark, 741 F.2d 401, 408 (D.C. Cir. 1984) (“The failure to
respond to comments is significant only insofar as it
demonstrates that the agency’s decision was not based on a
consideration of the relevant factors.” (internal quotation marks
omitted)).

C.
The National Association of Broadcasters argues that
because mobile operators frequently work indoors, the
 23
provisions of the Order designed to restrict low-power routers
to indoor operation offer mobile licensees little protection.
Moreover, the Association informs us, after the Commission
allowed unlicensed access in the 2.4 GHz band, “a contention-
based protocol . . . failed to protect . . . licensed users[,] . . .
rendering that band partially unusable.” Pet’rs’ Br. 71.

The Association and others raised these concerns in
comments to the Commission and requested that it reserve a
sliver of 6 GHz band for licensed mobile operation. The
Commission, however, never responded to their complaints
about interference in the 2.4 GHz band. Although the
Commission cited a study to support its conclusion that the
Order sufficiently protects mobile operators, that study does
not rebut the Association’s claims about interference in the 2.4
GHz band. As we have explained, “the opportunity to comment
is meaningless unless the agency responds to significant points
raised by the public.” Sherley v. Sebelius, 689 F.3d 776, 784
(D.C. Cir. 2012) (quoting Home Box Office, Inc. v. FCC, 567
F.2d 9, 35–36 (D.C. Cir. 1977)). We shall thus grant the
Association’s petition for review on this point and remand to
the Commission for it to respond to the Association’s concerns
about interference in the 2.4 GHz band.

The Association urges us to go further and vacate the
Order. “The decision whether to vacate depends on [(1)] the
seriousness of the order’s deficiencies (and thus the extent of
doubt whether the agency chose correctly) and [(2)] the
disruptive consequences of an interim change that may itself be
changed.” Allied-Signal, Inc. v. Nuclear Regulatory
Commission, 988 F.2d 146, 150–51 (D.C. Cir. 1993) (internal
quotation marks omitted). Here both factors favor remand
without vacatur. “It is conceivable that the Commission may
be able to explain” why its experience in the 2.4 GHz band
supports its ability to protect licensed mobile operators from
 24
harmful interference. Id. at 151. “At the same time, the
consequences of vacating may be quite disruptive.” Id. At oral
argument, Commission counsel explained that “vacating this
order would be incredibly disruptive given the fact that devices
have already started to be deployed” and assured us that “it’s
well within the Commission’s power to provide [more]
explanation” if needed. Oral Arg. Tr. 44. Given the
Commission’s failure to respond to the Association’s concerns
about harmful interference in the 2.4 GHz band, further
explanation is called for.

IV.
We end where we began, with the principles that guide our
review of petitioners’ challenges. As explained in the foregoing
pages, petitioners commercial communications providers,
electric utilities, and APCO have failed to demonstrate that the
Commission “relied on factors which Congress has not
intended it to consider, entirely failed to consider an important
aspect of the problem, offered an explanation for its decision
that runs counter to the evidence before the [Commission], or
is so implausible that it could not be ascribed to a difference in
view or the product of [Commission] expertise.” State Farm,
463 U.S. at 43. This failure is especially significant because in
issuing the Order, the Commission was acting to “foster[]
innovative methods of exploiting the spectrum,” thus requiring
our “greatest deference.” Mobile Relay Associates, 457 F.3d at
8 (internal quotation marks omitted). We therefore deny their
petitions for review. But for the reasons set forth above, we
grant the National Association of Broadcasters’ petition in part
and remand for further proceedings consistent with this
opinion.

So ordered.