Jaswinder Singh v. Uber Technologies Inc

PRECEDENTIAL

UNITED STATES COURT OF APPEALS
FOR THE THIRD CIRCUIT

______________

No. 21-3234
______________

JASWINDER SINGH, on behalf of himself and
all those similarly situated,
Appellant

v.

UBER TECHNOLOGIES, INC.

______________

No. 21-3363
______________

JAMES CALABRESE; GREGORY CABANILLAS;
MATTHEW MECHANIC,
individually and on behalf of all others similarly situated,
Appellants

v.

UBER TECHNOLOGIES, INC.; RAISER, LLC

____________________
 On Appeal from the United States District Court
for the District of New Jersey
(D.C. Civil Action Nos. 3-16-cv-03044 and 3-19-cv-18371)
District Judge: Honorable Freda L. Wolfson
______________

ARGUED: November 8, 2022

Before: JORDAN, SCIRICA, and RENDELL,
Circuit Judges.

(Filed: April 26, 2023)

Matthew D. Miller
Justin L. Swidler [ARGUED]
Swartz Swidler
9 Tanner Street
Suite 101
Haddonfield, NJ 08033

Counsel for Appellant in No. 21-3234

Catherine E. Anderson
Giskan Solotaroff & Anderson
90 Broad Street
2nd Floor
New York, NY 10004

Roosevelt N. Nesmith [ARGUED]
363 Bloomfield Avenue
Suite 2-C

2
Montclair, NJ 07042

Russell S. Warren, Jr.
473 Sylvan Avenue
Englewood Cliffs, NJ 07632

Counsel for Appellants in No. 21-3363

Samuel E. Eckman
Theane D. Evangelis [ARGUED]
Gibson Dunn & Crutcher
333 South Grand Avenue
Los Angeles, CA 90071

Blaine H. Evanson
Gibson Dunn & Crutcher
3161 Michelson Drive
Suite 1200
Irvine, CA 92612

Paul C. Lantis
Littler Mendelson
1601 Cherry Street
Three Parkway, Suite 1400
Philadelphia, PA 19102

Counsel for Appellees

_________________

OPINION OF THE COURT
_________________

3
SCIRICA, Circuit Judge.

The Federal Arbitration Act (FAA) compels federal
courts to enforce a wide range of arbitration agreements. But it
does not apply to arbitration agreements contained in the
“contracts of employment of seamen, railroad employees, or
any other class of workers engaged in foreign or interstate
commerce.” 9 U.S.C. § 1. These consolidated appeals ask us to
decide whether Uber drivers belong to such a class of workers.
We conclude, as have our sister circuits, that they do not. The
work of Uber drivers is centered on local transportation. Most
Uber drivers have never made a single interstate trip. When
Uber drivers do cross state lines, they do so only incidentally,
as part of Uber’s fundamentally local transportation business.
As a result, they are not “engaged in foreign or interstate
commerce” for the purposes of § 1 of the FAA. The District
Court reached this conclusion in a detailed and carefully
reasoned opinion. We will affirm.

I.

The FAA, enacted “in response to a perception that
courts were unduly hostile to arbitration,” Epic Sys. Corp. v.
Lewis, 138 S. Ct. 1612, 1621 (2018), requires courts to
“‘rigorously enforce’ arbitration agreements according to their
terms,” Am. Express Co. v. Italian Colors Rest., 570 U.S. 228,
233 (2013) (citation omitted). But the FAA’s scope is not
limitless. Expressly exempted from its coverage are arbitration
agreements within the “contracts of employment of seamen,
railroad employees, or any other class of workers engaged in
foreign or interstate commerce.” 9 U.S.C. § 1; Circuit City
Stores, Inc. v. Adams, 532 U.S. 105, 109 (2001). Our decision

4
addresses the scope of that final phrase—“any other class of
workers engaged in foreign or interstate commerce”—
otherwise known as the “residual clause.”

Two principles guide our analysis. First, the FAA’s
statutory context and purpose compel us to give § 1 “a narrow
construction.” Circuit City, 532 U.S. at 118. Second, the scope
of the residual clause is “controlled and defined by reference
to the enumerated categories” of seamen and railroad workers
designated in the statute. Id. at 115.

A.

This consolidated appeal arises out of two cases filed
against Uber by its drivers—Singh v. Uber Technologies and
Calabrese v. Uber Technologies. In each case, Uber
successfully moved to compel arbitration under the terms of its
agreements with the drivers. We described the facts of Singh’s
case in our previous decision, Singh v. Uber Technologies,
Inc., 939 F.3d 210 (3d Cir. 2019), and briefly review them here.

Plaintiffs are current or former Uber drivers from many
different states—New Jersey, New York, Ohio, Pennsylvania,
Missouri, and Nevada. At one time or another, each agreed to
a contract Uber calls its “Technology Services Agreement” as
a condition of using Uber’s platform. The content of the
relevant provisions of the agreement is not in dispute.

The agreement requires drivers to “resolve disputes
with [Uber] on an individual basis through final and binding
arbitration unless [the driver] choose[s] to opt out.” JA3, 168
(emphasis omitted). This includes “every claim or dispute that
lawfully can be arbitrated,” save a few specific exceptions.

5
JA153. Under the agreement, an arbitrator—not “a court or
judge”—is to decide any dispute “relating to interpretation or
application” of the provision, including its “enforceability,
revocability or validity.” JA4, JA82. Drivers who do not wish
to be bound by the arbitration provisions may opt out by
sending Uber an email or letter to that effect.

Singh’s case began six years ago as a putative class
action in New Jersey state court. Singh, 939 F.3d at 215. Singh
alleged Uber had violated New Jersey wage and hour laws by
misclassifying drivers as independent contractors, had failed to
pay them the minimum wage, and had failed to reimburse them
for business expenses. Uber removed the action and then
successfully moved to dismiss the case and compel arbitration
pursuant to the terms of its arbitration agreement with Singh.
Id. at 216. The District Court held that § 1 applied only to
transportation workers who move goods, not those who carry
passengers. Id. at 216-17. Singh appealed to this Court and we
reversed, holding that the exception applies equally to
“transportation workers who transport passengers, so long as
they are engaged in interstate commerce or in work so closely
related thereto as to be in practical effect part of it.” Id. at 214.
We reaffirmed our longstanding view of the residual clause as
including those classes of workers “actually engaged in the
movement of interstate or foreign commerce or in work so
closely related thereto as to be in practical effect part of it.” Id.
at 220 (quoting Tenney Eng’g, Inc. v. United Elec. Radio &
Mach. Workers of Am., (U.E.) Local 437, 207 F.2d 450, 452
(3d Cir. 1953) (en banc)).

We remanded to the District Court to determine whether
Singh belonged to a class of transportation workers “engaged
in interstate commerce.” Id. at 226-27. Because this question

6
could not be answered from the complaint alone, we directed
that “discovery must be allowed before entertaining further
briefing on the question.” Id. at 226. We encouraged the
District Court to consider “various factors,” including but not
limited to “the contents of the parties’ agreement(s),
information regarding the industry in which the class of
workers is engaged, information regarding the work performed
by those workers, and various texts—i.e., other laws,
dictionaries, and documents—that discuss the parties and the
work.” Id. at 227-28.

Calabrese filed his case in the District of New Jersey in
September 2019, just a few weeks after our decision in Singh.
The District Court consolidated the case with Singh’s and
ordered joint discovery. Like Singh, Calabrese claimed that
Uber had violated various labor and employment laws. He also
sought to proceed collectively under the Fair Labor Standards
Act.1

The District Court ordered that joint discovery before
again ruling in Uber’s favor and compelling arbitration. The
court defined the relevant class as Uber drivers nationwide.
Based on the record developed in discovery and the factors
listed above, the court determined that neither the arbitration
agreement nor the total number of cross-border trips was

1
Several additional plaintiffs opted in to Calabrese’s case. The
suit originally included himself (James Calabrese), Gregory
Cabanillas, and Matthew Mechanic as plaintiffs. Several more
plaintiffs opted-in later: Bulent Tasdemir, Salvador Delgado,
Vernon Small, Shane Golden, Shyidah Johnson, Corey Wims,
Denis Odom, Robin Rienerth, and Scott Tucker. We refer to all
these plaintiffs collectively as “Calabrese.”

7
dispositive. More significant, the court found, was “evidence
that [interstate] rides constitute just 2% of all rides, resemble
in character the other 98% of rides, and likely occur due to the
happenstance of geography.” JA32.

The District Court compelled arbitration in a thorough
and well-reasoned opinion.

B.

Two other appeals courts have concluded that rideshare
drivers2 do not constitute a “class of workers engaged in
foreign or interstate commerce” under § 1. In Capriole v. Uber
Technologies, Inc., 7 F.4th 854, 865 (9th Cir. 2021), the Ninth
Circuit found that “interstate movement” was not a “central
part of [Uber drivers’] job description.” The court noted that
Uber drivers primarily made “short and local” trips and crossed
state lines “infrequently.” Id. Even these infrequent trips across
state lines were still “inherently local in nature”—any
interstate component was a mere “happenstance of geography”
which did not alter Uber drivers’ fundamentally “intrastate
transportation function.” Id. at 864 (emphasis added) (citations
omitted).

The First Circuit charted a similar course in
Cunningham v. Lyft, Inc., 17 F.4th 244 (1st Cir. 2022). The
court pointed out that “not all Lyft drivers engage in any

2
By “rideshare drivers,” we mean those who use “a mobile app
or website” in order “to collect and transport a fare-paying
customer to a chosen destination.” Ride-Share, Oxford English
Dictionary (3d ed. updated Mar. 2022),
https://www.oed.com/view/Entry/165647#eid179399275.

8
interstate transportation.” Id. at 252. The court recognized that
the residual clause must be given a “narrow construction,” and
that its meaning was controlled by the enumerated categories
of “seamen” and “railroad workers,” who were “primarily
devoted to the movement of goods and people beyond state
boundaries.” Id. at 253. Lyft drivers, the court found, were not.
Id. Lyft as a company was “primarily in the business of
facilitating local, intrastate trips.” Id. The court concluded for
these reasons that “Lyft drivers are not among a class of
transportation workers engaged in interstate commerce within
the meaning of section 1 as narrowly construed.” Id.

II.3

To decide whether Plaintiffs are members of a “class of
workers engaged in foreign or interstate commerce,” a court
must first define the “class of workers” at issue. We agree with
the District Court that the class should be defined as
nationwide Uber drivers.4 See Capriole, 7 F.4th at 862.

3
The District Court had jurisdiction over Plaintiffs’ claims
under 28 U.S.C. §§ 1367, 1453, and 29 U.S.C. § 216. We have
appellate jurisdiction over the final judgment of the District
Court compelling arbitration under 28 U.S.C. § 1291. We
review the court’s order compelling arbitration de novo. Singh,
939 F.3d at 217. As “the party resisting arbitration,” Plaintiffs
bear the burden of showing “that the claims at issue are
unsuitable for arbitration.” Green Tree Fin. Corp.-Ala. v.
Randolph, 531 U.S. 79, 91-92 (2000).
4
Companies other than Uber employ rideshare drivers. The
parties’ arguments in this case have understandably focused on
Uber, and they have not discussed the practices of other
companies offering similar services. The District Court defined

9
 We begin, as the Supreme Court has instructed, by
examining the types of workers specifically mentioned in the
FAA’s text—“seamen” and “railroad employees.” See Circuit
City, 532 U.S. at 115. “As those terms contain ‘no geographic
limitations,’ ‘the most natural inference is that Congress
intended those terms to encompass all seamen and railroad
employees nationwide.’” Capriole, 7 F.4th at 862 (quoting
Osvatics v. Lyft, Inc., 535 F. Supp. 3d 1, 15 (D.D.C. 2021)
(Brown Jackson, J.)). We give the residual clause the same
national scope. The parties do not dispute this approach.
Accordingly, the relevant “class of workers” must cover the
whole country.

In addition, the class of workers must include all Uber
drivers.5 In interpreting the FAA, the Supreme Court has
repeatedly emphasized the Act’s use of the term “workers.”
Sw. Airlines Co. v. Saxon, 142 S. Ct. 1783, 1788 (2022)
(quoting New Prime, Inc. v. Oliveira, 139 S. Ct. 532, 540-41
(2019)). This word “directs the interpreter’s attention to ‘the
performance of work,’” and, when coupled with the word
“engaged,” “emphasizes the actual work that the members of

the class as “Uber drivers nationwide,” JA12, but elsewhere in
its opinion referred more generally to “rideshare drivers
nationwide,” JA15. The parties’ arguments—and our
decision—encompass those who do substantially similar work
for different companies. The precise framing of the class as
either Uber drivers or rideshare drivers makes no difference to
our opinion. See Osvatics v. Lyft, Inc., 535 F. Supp. 3d 1, 16
n.8 (D.D.C. 2021) (Brown Jackson, J.) (following the same
practice for Lyft drivers).
5
The Calabrese Plaintiffs, for their part, do not appear to
contest either part of this definition.

10
the class, as a whole, typically carry out.” Id. This work must
be defined specifically. See Circuit City, 532 U.S. at 118
(emphasizing that the § 1 exception must be given “a narrow
construction”).6

6
Singh argues that we would be wrong to construe the § 1
exception narrowly, as the “Supreme Court has since
abandoned the rule that statutory exemptions should be
narrowly construed.” Singh Br. 20. But Circuit City did not
depend on a general principle that all statutory exemptions
should be narrowly construed. Rather, the Court determined
that § 1 should be narrowly construed because of the FAA’s
“statutory context” and “purpose.” 532 U.S. at 118. More
specifically, the statute’s designation of “specific categories of
workers” before the residual clause “undermines any attempt
to give [the residual clause] a sweeping, open-ended
construction.” Id. The Court also relied on the FAA’s purpose,
which “seeks broadly to overcome judicial hostility to
arbitration agreements.” Id. (quoting Allied-Bruce Terminix
Cos. v. Dobson, 513 U.S. 265, 272-73 (1995)). These FAA-
specific considerations, and not any general principle, drove
the result in Circuit City. Singh’s cases, which attack only the
latter, are therefore inapplicable. See BP P.L.C. v. Mayor &
City Council of Balt., 141 S. Ct. 1532, 1538-39 (2021)
(rejecting a general principle of narrow construction for
statutory exceptions). Regardless, we would be obligated to
follow the specific holding of the Court in Circuit City even if
its justifications were undermined by “some other line of
decisions.” Rodriguez de Quijas v. Shearson/Am. Express, Inc.,
490 U.S. 477, 484 (1989). Fairly construed, the § 1 exception
has a narrow scope.

11
 Accordingly, we reject Singh’s proposed class of
“motor carrier workers” as too broad to be sustained. The
Supreme Court recently rejected a similar class of airline
employees. Saxon, 142 S. Ct. at 1788-89. Like “airline
employees,” “motor carrier workers” is too general a
description to explain much about class members’ actual work.
See Saxon, 142 S. Ct. at 1788.7

After defining the proper scope of the class at issue, we
next consider what it means for a class of workers to be
“engaged in interstate commerce” for purposes of § 1.
Plaintiffs argue that rare engagement is enough. We instead
conclude, in line with our sister circuits, that a class of workers
comes within the exception only if “interstate movement of
goods” or passengers is “a central part” of the job description
of the class. Wallace v. Grubhub Holdings, Inc., 970 F.3d 798,
803 (7th Cir. 2020) (Barrett, J.); accord Capriole, 7 F.4th at
864 (adopting the same standard); Cunningham, 17 F.4th at
253 (similar); see also Saxon, 142 S. Ct. at 1789-90 (explaining
that the residual clause only applies to “transportation
workers”—those who are “actively engaged in transportation”
of goods or people “via the channels of foreign or interstate
commerce”). Put another way, the class must either be
“actually engaged in the movement of interstate or foreign
commerce or in work so closely related thereto as to be in
practical effect part of it.” Singh, 939 F.3d at 220 (quoting
Tenney, 207 F.2d at 452).

7
On remand, the District Court collected decisions from
district courts that likewise defined the class as Uber (or Lyft)
drivers nationwide.

12
 We have suggested a non-exhaustive list of factors to
structure the residual clause inquiry: “the contents of the
parties’ agreement(s), information regarding the industry in
which the class of workers is engaged, information regarding
the work performed by those workers, and various texts—i.e.,
other laws, dictionaries, and documents—that discuss the
parties and the work.” Id. at 227-28. As the District Court
recognized, these factors are useful guides but do not change
the core question of whether interstate commerce is central to
the class’s job description.

Our focus on the centrality of interstate work is
compelled by the principle that the scope of the residual clause
is “controlled and defined by reference to the enumerated
categories” of seamen and railroad workers. Circuit City, 532
U.S. at 115. Seamen and railroad workers are “primarily
devoted to the movement of goods and people beyond state
boundaries.” Cunningham, 17 F.4th at 253. Their jobs are
“centered on the transport of goods in interstate or foreign
commerce.” Wallace, 970 F.3d at 802. The residual clause
must be similarly limited.

The FAA’s text and structure lead to the same result.
Compare the language of the FAA’s operative provision,
contained in § 2, with the language of the § 1 exception at issue
here. Section 2 applies to any “contract evidencing a
transaction involving commerce.” 9 U.S.C. § 2. This broad
language “reach[es] to the limits of Congress’ Commerce
Clause power.” Allied-Bruce Terminix Cos. v. Dobson, 513
U.S. 265, 270 (1995). By contrast, § 1 uses the narrower
formulation “engaged in interstate commerce.” See Circuit
City, 532 U.S. at 118. This formulation limits the clause’s
scope to workers employed “in the channels” of interstate

13
commerce. Great W. Mortg. Corp. v. Peacock, 110 F.3d 222,
227 (3d Cir. 1997); Wallace, 970 F.3d at 802. Accordingly, the
clause only applies to those whose jobs are centered on
“interstate transportation routes through which persons and
goods move.” United States v. Morrison, 529 U.S. 598, 613 n.5
(2000) (citation omitted).

Workers “who are actually engaged in the movement of
interstate or foreign commerce”—for example, interstate
truckers—easily qualify under the residual clause. Singh, 939
F.3d at 220 (quoting Tenney, 207 F.2d at 452); New Prime, 139
S. Ct. at 536, 539 (observing that interstate truck drivers are
plainly a class of workers engaged in interstate commerce). But
the exception is not limited to such workers.

A class of workers that does not regularly cross state
lines will qualify if their work is “so closely related” to
interstate commerce “as to be in practical effect part of it.”
Singh, 939 F.3d at 220 (quoting Tenney, 207 F.2d at 452).
Work meets this standard if it is a “constituent part” of the
interstate movement of goods or people rather than a “part of
an independent and contingent intrastate transaction.”
Immediato v. Postmates, Inc., 54 F.4th 67, 77 (1st Cir. 2022)
(citing Cunningham, 17 F.4th at 251).

It is not always easy to tell whether work is a
“constituent part” of the flow of interstate commerce or occurs
outside of it. See Immediato, 54 F.4th at 79 (observing that “[i]t
may be possible that goods can change hands several times
during transport without exiting the flow of interstate
commerce” and that a class of workers need not be “employed
by a company of any particular size or geographic scope”). Our
analysis focuses on “practical, economic continuity.” Osvatics,

14
535 F. Supp. 3d at 18-19 (quotation marks omitted) (quoting
Gulf Oil Corp. v. Copp Paving Co., 419 U.S. 186, 195 (1974)).

Some recent decisions help clarify the line between
classes of workers who qualify under the residual clause and
classes of workers that do not. On one side of the line are those
whose work occurs within the flow of interstate commerce. In
Saxon, for example, the Supreme Court held a class of
Southwest Airlines baggage handlers fell under § 1. As loaders
of interstate cargo, these baggage handlers performed work
which was part of an unbroken stream of interstate commerce.
Saxon, 142 S. Ct. at 1790. Similarly, in Waithaka v.
Amazon.com, Inc., 966 F.3d 10, 13 (1st Cir. 2020), the First
Circuit held that Amazon delivery drivers who “locally
transport[ed] goods on the last legs of interstate journeys,” fell
under § 1 because their work occurred “within the flow of
interstate commerce.” On the other side of the line are workers
who engage in primarily local economic activity with only
tangential interstate connections. Food delivery drivers, for
example, can be distinguished from Amazon delivery drivers,
as the former deliver food only after it has left the stream of
interstate commerce. Rittmann v. Amazon.com, Inc., 971 F.3d
904, 916 (9th Cir. 2020); Wallace, 970 F.3d at 802-03.
Similarly, Chicago taxi drivers provide “independent local
service” which is “not an integral part of interstate
transportation.” United States v. Yellow Cab Co., 332 U.S. 218,
233 (1947), overruled on other grounds by Copperweld Corp.
v. Indep. Tube Corp., 467 U.S. 752 (1984).

Plaintiffs argue along various lines to reach the
conclusion that even a trivial amount of interstate
transportation work suffices to bring a worker within the
exception. This conclusion must be rejected. It is contrary to

15
the principles described above and would contravene the basic
policy of the FAA, which is to broadly place arbitration
agreements on equal footing with other contracts. See Circuit
City, 532 U.S. at 115, 118-19. Plaintiffs’ interpretation would
cover even “a pizza delivery person who delivered pizza across
state lines to a customer in a neighboring town.” Hill v. Rent-
A-Center, Inc., 398 F.3d 1286, 1290 (11th Cir. 2005). There is
no evidence to suggest that Congress meant to cover such
workers. See Wallace, 970 F.3d at 802-03; Immediato, 54 F.4th
at 77-78; Archer v. Grubhub, Inc., 190 N.E.3d 1024, 1031-33
(Mass. 2022). Congress’ use of the enumerated categories of
“seamen” and “railroad employees,” when coupled with the
narrow construction due the exception, convinces us that the
residual clause includes only those workers whose jobs are
centered on interstate commerce.

We are unpersuaded by Singh’s argument that there is
no way to know that the key shared characteristic of “seamen”
and “railroad employees” is having a job centered on interstate
commerce. Congress meant to identify engagement in
interstate commerce as the enumerated categories’ key shared
characteristic. The FAA’s text makes it explicit—the residual
clause requires that a class of workers is “engaged in interstate
commerce.” See 9 U.S.C. § 1. This text is “[t]he best evidence
of Congress’ intent.” United States v. Schneider, 14 F.3d 876,
879 (3d Cir. 1994).

This approach is consistent with Saxon. Singh
emphasizes a single sentence—the Court’s statement that “any
class of workers directly involved in transporting goods across
state or international borders falls within § 1’s exemption.”
Saxon, 142 S. Ct. at 1789. As the rest of the opinion makes
clear, this does not mean that rare border crossings are enough

16
to make interstate transportation central to a class of workers’
job description. Rather, we consider the “actual work” that
class members “typically carry out.” Id. at 1788. Incidental
border crossings are insufficient if a class of workers is not
typically involved with the channels of interstate commerce.
Wallace, 970 F.3d at 800 (“[S]omeone whose occupation is not
defined by its engagement in interstate commerce does not
qualify for the exemption just because she occasionally
performs that kind of work.”); Waithaka, 966 F.3d at 25
(noting that crossing state lines is not the “touchstone of the
exemption’s test”).8

III.

We now turn to the key question: Is engagement with
interstate commerce central to the work of Uber drivers? The
District Court found that it was not. We agree. As a class, Uber
drivers are in the business of providing local rides that
sometimes—as a happenstance of geography—cross state
borders. Remove interstate commerce from the equation, and

8
Both Singh and Calabrese urge us to follow International
Brotherhood of Teamsters Local Union No. 50 v. Kienstra
Precast, LLC, 702 F.3d 954, 956 (7th Cir. 2012), which they
read to hold that crossing a state line even once means that a
worker is engaged in interstate commerce for purposes of § 1.
As explained above, this position is at odds with Saxon, our
precedents, and the majority approach of our sister courts. The
duties of seamen and railroad employees are defined by
interstate commerce—remove interstate commerce from the
equation, and the fundamental character of their work changes.
So too must the work of any class of workers covered by § 1.
See, e.g., Waithaka, 966 F.3d at 22-24.

17
the work of Uber drivers remains fundamentally the same.
Plaintiffs have not shown that drivers’ infrequent interstate
trips are, on the whole, an essential part of their job. Indeed,
their statistics demonstrate that most Uber drivers have never
made a single interstate trip. Neither have Plaintiffs shown that
drivers’ intrastate duties, such as driving riders to and from
airports, are a “constituent part” of the interstate movement of
goods or people. Immediato, 54 F.4th at 77. As a result, we
conclude that Uber drivers are not a class of workers engaged
in interstate commerce and, accordingly, that they do not fall
under the § 1 exception.

The other appeals courts to consider this question have
reached the same conclusion. Plaintiffs, however, encourage us
to disregard these decisions on the grounds that those courts
had insubstantial evidentiary records before them. See Singh,
939 F.3d at 226-27. Without the benefit of more evidence,
Plaintiffs argue, courts have routinely placed undue emphasis
on a single statistic: that 2.5 percent of Uber trips are interstate.
See Capriole, 17 F.4th at 252-53 (noting only 2 percent of Lyft
trips cross state lines). Through discovery, Plaintiffs have
developed their own statistics which they say provide a more
accurate picture of Uber drivers’ engagement with interstate
commerce. But the District Court considered Plaintiffs’ new
evidence and still found the 2.5 percent statistic persuasive.
The core problem with Plaintiffs’ evidence, which the District
Court identified, is that it stresses the total volume of interstate
trips to the exclusion of other types of evidence.

Singh focuses directly on the “141.5 million total
interstate trips” Uber drivers made from 2010 through May
2020. Singh Br. 30. This statistic, however, cannot shed much
light on Uber drivers’ typical duties. A high number of

18
interstate trips does not mean that a class of workers is engaged
in interstate commerce for purposes of § 1 if a small proportion
of the class is responsible for most of the trips. Rather, to be
central to a class of workers’ job description, engagement with
interstate commerce must be typical of the work that class
members generally do.

Calabrese suggests a different statistic: the percentage
of drivers who “provide 50 or more trips in a year.” Calabrese
Br. 4. Calabrese justifies his focus on drivers who make more
than 50 trips by arguing that the § 1 analysis must take turnover
into account. Since a minority of Uber drivers use the Uber app
for more than six months, he contends, an accurate picture of
Uber drivers’ engagement with interstate commerce requires
adjusting the raw statistics for turnover. He proposes we
consider the 40 percent of drivers who work long enough to
complete at least 50 trips in a year. Of this 40 percent, 35.1
percent have made at least one interstate trip.

We need not address how, or whether, the § 1 analysis
should take turnover into account. Calabrese’s statistics taken
at face value undermine his point. His numbers reveal that even
among the most active Uber drivers, a majority—nearly 65
percent—have never made a single interstate trip. On such
evidence, it is easy to conclude that interstate trips are not a
typical feature of class members’ work.

We stress that this statistic is not dispositive. An
occurrence may be central to a worker’s job description even
if it is rare. See, e.g., Islam v. Lyft, Inc., 524 F. Supp. 3d 338,
351 (S.D.N.Y. 2021) (observing that criminal trials are central
to the work of district court judges even though they are
infrequent occurrences). Uber drivers’ interstate trips,

19
however, are incidental—take away interstate trips, and the
fundamental character of Uber drivers’ work remains the same.
One trip may influence another tangentially, but each is
discrete. See Cunningham, 17 F.4th at 251 (distinguishing
engagement with the flow of interstate commerce from
participation in independent transactions). Because it is not the
act of crossing a state border alone that qualifies as engagement
with interstate commerce for purposes of § 1, and drivers’
interstate trips are largely unrelated to one another, Plaintiffs
cannot show that drivers’ rare trips across state lines are
anything more than incidental to their intrastate work.

In addition to driver and trip data, Plaintiffs offer
evidence of Uber’s policies from the Technology Services
Agreement. An employer’s policies can be relevant to the § 1
analysis if they tend to show that the employer directed a
single, unbroken stream of interstate commerce. Plaintiffs
argue that Uber did just that. Plaintiffs’ argument has an
intriguing implication: that millions of discrete interstate trips,
directed by one employer, can together form an unbroken
stream of commerce. But § 1’s focus is on a “class of workers,”
not employers. An employer’s policies are only relevant
insofar as they illuminate something about a class of workers’
typical duties.

Accordingly, Plaintiffs’ evidence of Uber’s policies
misses the mark. Stating that interstate trips are “integral to
Uber’s business and the fulfillment of its mission goals,”
Calabrese Br. 9, without more, does not explain much about
drivers’ actual work. A business undeniably engaged in
interstate commerce may employ workers who are not so
engaged. See Saxon, 142 S. Ct. at 1789 n.1 (leaving open the
question of whether a class of workers who only supervise

20
cargo unloading would be exempt under § 1). If an individual
worker is not personally engaged in interstate commerce, that
worker must belong to a class of workers “whose occupation
is . . . defined by its engagement in interstate commerce” in
order to be exempt under § 1. See Wallace, 970 F.3d at 800.

Plaintiffs’ evidence that Uber organizes drivers into
multistate “territories” based on where they live and does not
allow drivers to opt out of interstate trips shows that Uber
anticipates at least some drivers crossing state lines. It does not
demonstrate, however, that interstate trips are essential to
drivers’ activities.

When drivers sign up with Uber, they are assigned to
either a multistate or a single-state territory, depending on
where they live. Uber created these territories in response to
variation in state and local regulations. Drivers assigned to
multistate territories can pick up passengers in other states,
while drivers assigned to single-state territories can only pick
up passengers in that state. Drivers in a single-state territory
will not receive ride requests while driving outside that state,
unlike drivers in a multistate territory. All drivers must accept
a ride request before learning the trip destination. Although
drivers may cancel a trip, Uber may deactivate a driver’s
account if her cancellation rate is higher than average for her
area. Drivers may not opt out of receiving ride requests that
require interstate travel.

The existence of multistate territories and the lack of an
opt-out feature could show that interstate travel is essential to
drivers’ work if paired with evidence that these policies impact
drivers’ actual work. Plaintiffs do not provide that extra
evidence. It is unclear, for example, whether the day-to-day

21
work of drivers in multistate territories differs from that of
drivers in single-state territories. It is also unclear whether the
lack of an opt-out feature pressures drivers into taking
interstate trips, given that only 17 percent of all Uber drivers
completed one or more interstate trips in 2019.

In addition to arguing that interstate transport is integral
to Uber’s business, Plaintiffs challenge the idea that interstate
trips can be local for § 1 purposes. Plaintiffs point to the fact
that Uber authorizes would-be passengers to request a trip
exceeding 100 miles. But they do not seriously contest Uber’s
claim that the average trip is far shorter—6.1 miles for all trips
and 13.5 miles for interstate trips. Although average trip length
is not dispositive, a short average trip length makes it more
likely that drivers serve local communities that may, by
happenstance of geography, cross state lines. See Capriole, 7
F.4th at 864 (citing Rogers v. Lyft Inc., 452 F. Supp. 3d 904,
916 (N.D. Cal. 2020) (“Interstate trips that occur by
happenstance of geography do not alter the intrastate
transportation function performed by the class of workers.”));
Omaha & Council Bluffs St. Ry. Co. v. Interstate Com.
Comm’n, 230 U.S. 324, 335-36 (1913) (holding that street
railroads are not engaged in interstate commerce because they
“are local . . . and for the use of a single community, even
though that community be divided by state lines”).

Finally, we reject Plaintiffs’ argument that drivers who
ferry passengers to and from airports are part of an integrated
interstate transport effort. Plaintiffs point out that Uber has
agreements with major airports authorizing drivers to drop off
and pick up passengers at terminals. They also argue that
airport trips are so closely related to interstate commerce as to
bring rideshare drivers within the ambit of § 1. They connect

22
these arguments to the Supreme Court’s decision in United
States v. Yellow Cab Co., which found that certain station-to-
station taxi rides implicated interstate commerce. See 332 U.S.
at 228-29.

We find this analogy unconvincing. The rides in Yellow
Cab were part of an exclusive contract between a taxi service
and the railroad—passengers bought a single ticket which
included both the train and taxi portions of their journey. Id. at
228 Plaintiffs have pointed to no examples of a rideshare app
which allows passengers to buy a single ticket that includes
both flight and rideshare. Rather, rideshare trips to airports are
done as part of drivers’ “independent local service.” Yellow
Cab, 332 U.S. at 232-33. Such rides are not “part of interstate
transportation.” Id. at 233. Yellow Cab therefore seriously
undermines Plaintiffs’ argument, as our fellow courts have
found. See Capriole, 7 F.4th at 863-64 (citing Yellow Cab, 332
U.S. at 228-29); Cunningham, 17 F.4th at 250-52 (same);
Osvatics, 535 F. Supp. 3d at 19 (same); Immediato, 54 F.4th at
79 (same).

IV.

Plaintiffs also object to the District Court’s decision to
compel arbitration on various contractual grounds. We reject
these arguments.

A.

Singh argues at length that “no contract to arbitrate” was
formed between himself and Uber. Singh Br. 41-47. Singh
could have raised this issue in his first appeal but did not. See
Singh Br. 2, 5-6, 27, Singh, 939 F.3d 210 (No. 17-1397) (Aug.

23
13, 2018). In fact, Singh told us that Uber “required” him to
accept the agreement. Id. at 27; see also id. at 5 (“Singh had to
click a button that said, ‘YES, I AGREE.’”). Having explicitly
conceded the point in his first appeal, Singh may not now
challenge the formation of the arbitration agreement. See
Beazer E., Inc. v. Mead Corp., 525 F.3d 255, 263 (3d Cir.
2008) (“It is elementary that where an argument could have
been raised on an initial appeal, it is inappropriate to consider
that argument on a second appeal following remand.” (quoting
Nw. Ind. Tel. Co. v. FCC, 872 F.2d 465, 470 (D.C. Cir. 1989))).

Singh objects that Uber has failed to produce admissible
evidence that he assented to the arbitration agreement. But
Singh has admitted that he was presented with the agreement,
and the court found that he accepted it. We discern no error in
this finding, and no abuse of discretion in the court’s
evidentiary rulings.

B.

None of Plaintiffs’ challenges to the validity of the
arbitration clause are cognizable in this court. The contract
says that all disputes “relating to interpretation or application
of this Arbitration Provision, including the enforceability,
revocability or validity of the Arbitration Provision . . . shall be
decided by an Arbitrator and not by a court or judge.” JA182-
83. Courts call this type of provision a delegation clause—“an
agreement to arbitrate threshold issues concerning the
arbitration agreement.” Rent-A-Center, W., Inc. v. Jackson,
561 U.S. 63, 68 (2010). In the presence of a delegation clause,
we “cannot reach the question of the arbitration agreement’s
enforceability” unless the clause itself “is not enforceable.”
MacDonald v. CashCall, Inc, 883 F.3d 220, 226 (3d Cir. 2018).

24
“A party contesting the enforceability of a delegation clause,”
as Singh does, “must ‘challenge the delegation provision
specifically.’” Id. (quoting Rent-A-Center W., 561 U.S. at 70,
72).

Singh specifically challenges the delegation clause on
two bases. We reject both. First, he argues the agreement has
not made a “clear and unmistakable delegation of authority to
the arbitrator,” as is required. Singh Br. 57. This argument is
based on the agreement’s separate forum selection clause,
which provides that disputes arising out of the agreement “shall
be subject to the exclusive jurisdiction” of San Francisco’s
courts. Singh Br. 56; JA180. The Ninth Circuit, construing
these precise provisions, rejected this argument. Mohamed v.
Uber Techs., 848 F.3d 1201, 1208-09 (9th Cir. 2016). So do
we. The language in the two provisions is easily reconciled,
and any conflict is “artificial.” Id. at 1209. “It is apparent” that
the forum selection clause here “was intended” to identify the
proper venue for “an action in court to enforce” the agreement,
and “to identify the venue for any other claims that were not
covered by the arbitration agreement.” Id. “That does not
conflict with or undermine the agreement’s unambiguous
statement identifying arbitrable claims and arguments.” Id.

Second, Singh claims the delegation clause is invalid
because it is “subject to unilateral modification” and is
“illusory.” Singh Br. 45. The agreement, however, provides
that any modifications will be conveyed in writing to the driver
and become effective only if the driver consents by continuing
to use the Uber app. These limitations on Uber’s right to
modify the agreement are sufficient to save it from being
illusory. Jaworski v. Ernst & Young U.S. LLP, 119 A.3d 939,
947-49 (N.J. Super. Ct. App. Div. 2015) (holding that an

25
employee’s continued employment after the amendment of an
arbitration policy constituted consent to the policy); Blair v.
Scott Specialty Gases, 283 F.3d 595, 604 (3d Cir. 2002)
(approving arbitration agreement as not illusory when
employer’s right to modify was conditional on “putting the
change in writing, providing a copy to the employees, and
allowing the employees to accept the change by continuing
employment”).9

9
Whether or not a contractual provision is illusory—and the
other contractual issues raised by the parties—are questions of
state law. See Collins v. Mary Kay, Inc., 874 F.3d 176, 182 (3d
Cir. 2017); DIRECTV, Inc. v. Imburgia, 577 U.S. 47, 54 (2015)
(“[T]he interpretation of a contract is ordinarily a matter of
state law . . . .”). But which state’s law applies? Singh argues
that California law applies to the arbitration agreement. But his
brief often makes arguments—including on this issue—based
solely on the law of other states with no reference to California
law. And in his previous appeal, Singh wrote to this Court that
“New Jersey law controls” this case. Letter of Jan. 28, 2019,
Singh (17-1397). Uber argues that the applicable law is the law
of the various states where each Plaintiff lived and worked:
Missouri, Nevada, New York, Ohio, Pennsylvania, and New
Jersey.

To resolve this dispute, we apply the choice-of-law rules of
New Jersey. See Klaxon Co. v. Stentor Elec. Mfg. Co., 313 U.S.
487, 496 (1941) (holding federal courts should apply the
choice of law rules of the forum state). Under those rules, “the
first step is to determine whether an actual conflict exists”
between the potentially applicable laws. P.V. ex rel. T.V. v.
Camp Jaycee, 962 A.2d 453, 460 (N.J. 2008). The parties have
pointed out no relevant differences in state law with respect to

26
 C.

Calabrese argues that some of his fellow FLSA
plaintiffs “opted out of arbitration with Uber” and thus cannot
be compelled to arbitrate. True enough, three plaintiffs did
purport to exercise their right to opt out of arbitration under
contracts with Uber in 2019, and one plaintiff did so in 2019
and 2020. But both of those agreements made clear that
Plaintiffs would be “bound by an existing arbitration
agreement” with Uber if they had accepted one in the past.
Uber Br. 9; JA357-59. All three of the opt-out plaintiffs
previously agreed to arbitration with Uber in 2015. They
remain bound by that agreement notwithstanding their
subsequent opt-out. See Capriole, 7 F.4th at 859 n.2.10

this issue. Instead, they all argue from general principles,
typically with reference to federal and New Jersey cases. Our
own examination of the cases similarly reveals no relevant
distinctions. See Baker v. Bristol Care, Inc., 450 S.W.3d 770,
776-77 (Mo. 2014); Baldonado v. Wynn Las Vegas, LLC, 194
P.3d 96, 105-06 & n.39 (Nev. 2008); Bassett v. Elec. Arts, Inc.,
93 F. Supp. 3d 95, 106-08 (E.D.N.Y. 2015) (applying New
York law); Jones v. Carrols, LLC, 119 N.E.3d 453, 464-65
(Ohio Ct. App. 2019); Blair v. Scott Specialty Gases, 283 F.3d
595, 603-04 (3d Cir. 2002) (applying Pennsylvania law);
Jaworski v. Ernst & Young U.S. LLP, 119 A.3d 939, 948-49
(N.J. Super. Ct. App. Div. 2015). As such, we see “no choice-
of-law issue to be resolved.” Camp Jaycee, 962 A.2d at 460.
10
We take no position on whether plaintiffs must arbitrate
claims arising after they exercised their right to opt out. As
plaintiffs are bound in some sense by the 2015 agreement,
which delegates the question of arbitrability to the arbitrator,

27
 The parties’ contract forecloses Calabrese’s argument.
The agreement says that the driver can “opt out of this
Arbitration Provision,” but may be “bound by an existing
agreement to arbitrate disputes.” Calabrese Br. 23-24; JA450
(emphasis added). Calabrese argues that our interpretation
makes opting out of arbitration illusory. We disagree. Plaintiffs
had a meaningful right to opt out of every agreement that they
were presented with. We are sympathetic to Plaintiffs’ point.
No doubt Uber’s requirement that they opt out of each new
agreement is “more burdensome” than a permanent opt-out
right. Mohamed, 848 F.3d at 1211. Ultimately, though, we
agree with the Ninth Circuit that “the contract bound Uber to
accept opt-outs from those drivers who followed the procedure
it set forth. There were some drivers who did opt out and whose
opt-outs Uber recognized. Thus, the promise was not illusory.”
Id.

***

We will affirm the judgment of the District Court.

we must compel arbitration and leave the determination of
whether any particular dispute is within the scope of the
agreement to the arbitrator.

28