Concurring opinion filed by Circuit Judge Dyk.
MOORE, Circuit Judge.Erik Brunetti appeals from the decision of the Trademark Trial and Appeal Board (“Board”) affirming the examining attorney’s refusal to register the mark FUCT because it comprises immoral or scandalous matter under 15 U.S.C. § 1052(a) (“§ 2(a)”). We hold substantial evidence supports the Board’s findings and it did not err concluding the mark comprises immoral or scandalous matter. We conclude, however, that § 2(a)’s bar on registering immoral or scandalous marks is an unconstitutional restriction of free speech. We therefore reverse the Board’s holding that Mr. Brunetti’s mark is unregistrable.
Background
I. Section 2(a)’s Bar on Registration of Immoral or Scandalous Marks
Section 2(a) of the Lanham Act provides that the Patent and Trademark Office (“PTO”) may refuse to register a trademark that “[cjonsists of or comprises immoral, deceptive, or scandalous matter; or matter which may disparage or falsely suggest a connection with persons, living or dead, institutions, beliefs, or national symbols, or bring them into contempt or disrepute ..,15-U.S.C. § 1052(a), While § 2(a) identifies “immoral” and “scandalous” subject matter as separate bases to refuse to register a trademark—and are provisions separated by the “deceptive” provision—the PTO generally, applies the bar on immoral or scandalous marks as a unitary provision (“the immoral or scandalous provision”). See TMEP § 1203.01 ("Although the words ‘immoral’ and ‘scandalous’ may have somewhat different connotations, case law has included immoral matter in the same category as scandalous matter.”); In re McGinley, 660 F.2d 481, 485 n.6 (CCPA 1981) (“Because of our holding, infra, that appellant’s mark, is ‘scandalous,’ it is unnecessary to consider whether appellant’s mark is ‘immoral.’ We note the dearth, of reported trademark decisions in which .the. term ‘immoral’ has been directly applied.”); see also Anne Gil-son LaLonde & Jerome Gilson,■ Trademarks Laid Bare: Marks That May Be Scandalous or Immoral, 101 Trademark Rep. 1476, 1489 (2011) (“U.S. courts and the Board have not distinguished between ‘immoral’ and ‘scandalous’ - and have focused on whether marks are scandalous or offensive rather than contrary to some, accepted standard of morality.” (citation-omitted)). The bar on immoral or scandalous marks was first codified in 1905, see Act , of Feb. 20, 190.5, ch. 592, § 5(a), 33 Stat. 724, 725, and re-enacted in the Lan-ham Act in 1946, Pub. L. 79-489, § 2(a), 60 Stat. 427, 428 (codified at 15 .U.S.C. § 1052(a)).. '
To determine whether .a mark should be-disqualified under § 2(a), the, PTO asks whether a “substantial composite of the general public” would-.find the mark scandalous, defined as “shocking- to the sense of truth, decency, or propriety; disgraceful; offensive; disreputable; ... giving offense to the conscience or moral feelings; ... or calling out for condemnation.” In re Fox, 702 F.3d 633, 635 (Fed. Cir. 2012) (alterations omitted) (quoting In re Mavety Media Grp. Ltd,, 33 F.3d 1367, 1371 (Fed. Cir. 1994)). Alternatively, “the PTO, may prove scandalousness by establishing that a mark, is ‘vulgar.’ ” Id. , (quoting In re Boulevard Entm’t, Inc., 334 F.3d 1336, 1340 (Fed. Cir. 2003)), Vulgar marks are “lacking in taste, indelicate, [and] morally crude See McGinley, 660 F.2d at 486 (quoting In re Runsdorf, 171 U.S.P.Q. 443, 443-44 (1971)). The PTO makes a determination as to whether a mark is scandalous “in the context of contemporary attitudes” and “in the context of the marketplace as applied to only the goods described in the application.”' Fox, 702 F.3d at 635 (internal quotation marks and alterations omitted) (quoting Mavety, 33 F.3d at 1371).
Because the scandalousness determination is made in the context of contemporary attitudes, the concept of what is actually. immoral or scandalous changes over time. Early cases often, but not. always, focused on religious words or symbols, See, e.g., In re Riverbank Canning Co., 95 F.2d 327, 329 (CCPA 1938) (MADONNA for wine); Ex parte Martha Maid Mfg. Co., 37 U.S.P.Q, 156 (Comm’r Pat. 1938) (QUEEN MARY for women’s underwear); Ex Parte Summit Brass & Bronze Works, Inc., 59 U.S.P.Q. 22 (Comm'r Pat. 1943) (AGNUS DEI for safes); In re P. J. Valckenberg, Gmbh, 122 U.S.P.Q. 334 (T.T.A.B. 1959) (MADONNA for wine); In re Reemtsma Cigarettenfabriken GM.B.H., 122 U.S.P.Q. 339 (T.T.A.B. 1959) (SENUSSI (a Muslim sect that forbids smoking) for cigarettes); In re Socie-dade Agricola E, Comerical Dos Vinhos Messias, S.A.R.L., 159 U.S.P.Q. 275 (T.T.A.B. 1968) (MESSIAS for wine and brandy). In later cases, the PTO rejected a wider variety of marks as scandalous. See, e.g., Runsdorf, 171 U.S.P.Q. at 443 (BUBBY TRAP for brassieres); McGinley, 660 F.2d at 482 (mark consisting of “a photograph of a nude man -and woman kissing and embracing in a maimer appearing to expose the male genitalia” for a swingers newsletter); In re Tinseltown, Inc., 212 U.S.P.Q. 868 (T.T.A.B. 1981) (BULLSHIT on handbags, purses, and other personal accessories); Greyhound Corp. v. Both Worlds, Inc., 6 U.S.P.Q.2d 1635 (T.T.A.B. 1988), (mark depicting a defecating dog); Mavety, 33 F.3d 1367 (BLACK TAIL for adult entertainment magazines).
II. Facts of This Case
Mr. Brunetti owns the clothing brand “fuct,” which he founded in 1990. In 2011, two individuals filed an intent-to-use application (No. 85/310,960) for the mark FUCT for various items of apparel. The original applicants assigned -the application to Mr. Brunetti, who amended it to allege use of the mark. The examining attorney refused to register the mark under § 2(a) of the Lanham Act, finding it comprised immoral or scandalous matter. The examining attorney reasoned that FUCT is the past tense of the verb “fuck,” a vulgar word, and is therefore scandalous. J.A. 203. ■
Mr. Brunetti requested reconsideration and appealed to the Board. The examining attorney denied reconsideration, and the Board affirmed. In its decision, the Board stated the dictionary definitions in the record uniformly characterize the word “fuck” as offensive, profane, or vulgar. The Board noted that the word “fuct” is defined by Urban Dictionary as the past tense of the verb “fuck” and pronounced the samé as the word “fucked,” and therefore found it is “recognized as a slang and literal equivalent of the word ‘fucked,’ ” with “the same vulgar meaning.” J.A. 6-7 & n.6. Based on the examining attorney’s Google Images search results, the Board stated Mr. Brun-etti used the mark in the context- of “strong, and often explicit, sexual imagery that objectifies women and offers degrading examples 'of extreme misogyny,” .with a theme “of extreme nihilism—displaying' ah unending succession of anti-social imagery of executions, despair, violent and bloody scenes including dismemberment, hella-cious or apocalyptic events, and dozens of examples of other imagery lacking .in taste.” J.A. 8-9. The Board explained that Mr. Brunetti’s use of, the mark “will be perceived by his targeted market segment as the phonetic equivalent of the wor[d] ‘fucked.’-” J.A. 9. In light of the record, it found Mr.. Brunetti’s assertion that the mark “was chosen as an invented or coined term stretches credulity.” Id, It concluded that the mark is vulgar and therefore un-registrable under § 2(a) of. the Lanham Act. Mr. .Brunetti appealed. We have jurisdiction under 28 U.S.C. § 1295(a)(4).
"Discussion. ,,
Mr. Brunetti argues substantial evidence does not support the Board’s finding the mark FUCT is vulgar under § 2(a) qf the -Lanham Act. He argues even,if the mark is vulgar, § 2(a) does not expressly prohibit the registration -of vulgar marks and' a mark should be approved for registration when there is doubt as to its meaning, as he alleges there is here. Alternatively, Mr, Brunetti challenges the constitutionality of § 2(a)’s bar on immoral or scandalous marks,
-I. The Mark FUC.T is Vulgar and Therefore Scandalous
The determination that a mark is scandalous Us a conclusion of law. based upon underlying factual inquiries. Fox, 702 F.3d at 637, We review the Board’s factual findings1 for substantial evidence, and its ultimate conclusion de novo. Id. Substantial evidence- is “more than a mere scintilla” and “such relevant evidence as a reasonable mind might accept as adequate” to support a conclusion. Consol. Edison v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938).
It is undisputed that the word “fuck” is vulgar. Dictionaries in the record characterize the word as “taboo,” “one of the most offensive” English words, “almost universally considered vulgai’,” and an “extremely offensive expression.” J.A. 5-6; J.A. 206 (Collins Online Dictionary); J.A. 209 (Vocabulary.com); J.A. 211 (Wikipedia.com); J.A. 351 (MacMillan Dictionary). Mr. Brunetti argues that the vulgarity of “fuck” is irrelevant to whether the mark FUCT is vulgar. We do not agree.
Substantial evidence supports’ the Board’s finding that “fuct” is a “phonetic twin” of “fucked,” the past tense of the word “fuck.” J.A. 10. Urban Dictionary defines “fuct” as the “past tense of the verb fuck.” J.A. 83. MacMillan Dictionary indicates that the word “fucked” is pronounced phonetically as /fAkt/, which the Board found sounds like “fukt” or “fuct.” J.A. (3 & n.6. This evidence linking the two terms is sufficient to render the vulgarity of the word “fuck” relevant to the vulgarity of Mr. Brunetti’s mark.
Evidence of the use of Mr. Brunetti’s mark in the marketplace further buttresses the Board’s finding of a link between the mark and the word “fuck.” The Board found the term “fuct” is used on products containing sexual imagery and that consumers perceive the mark as having “an unmistakable aura of negative sexual connotations.” J.A. 9. One T-shirt—captioned the “FUCT Orgy” shirt—depicts a group sex scene. J.A. 346. Another T-shirt contains the word “FUCK” in yellow letters, with a “T” superimposed over the “K” such that the word FUCK is still visible. J.A. 325. A third T-shirt has the brand name FUCT depicted above the slogan “1970 smokin dope & fucking in the streets.” J.A. 312. Because one meaning of “fuck” is “to have sex with someone,” the placement of the mark on products containing sexual imagery makes it more likely that the mark will be perceived as the phonetic equivalent of the word “fucked.” J.A. 9.
Mr. Brunetti challenges the evidence on which the Board relied in making the vulgarity finding. He argues that Urban Dictionary is not a standard dictionary edited by lexicographers and the author of the definition cited by the Board lacks lexicographic expertise. He argues that the Board did not consider his current line of products, which he provided to the examining attorney, but instead relied on a random collection of outdated products collected from Google Images. He argues the Board should not have considered these images because they lack foundation, are inadmissible hearsay, and are irrelevant to the current perception of the mark in the marketplace. He also argues that the majority of the marked products contain no sexual imagery.
Mr. Brunetti’s arguments have no merit. For ex parte proceedings, the Board permits the examining attorney to consider materials from the Internet, having adopted a “somewhat more permissive stance with respect to the admissibility and probative value of evidence.” TBMP §§ 1208, 1208.03. The pedigree of the author of a definition may affect the weight that evidence is given but does not render the definition irrelevant. Similarly, the ages of the images collected by the examining attorney may affect evidentiary weight, not relevance.
Mr. Brunetti also argues that the Board ignored probative evidence that the mark is not vulgar. He argues that both he and the owner of a high-end clothing store declared that the mark was not vulgar. He argues that the meaning of the term “fuct” is ambiguous, but that to the extent it has any meaning, it is “Friends yoU Can’t Trust.” See Mavety, 33 F.3d at 1374 (“commending] the practice” of erring on the side of publication when marks are not clearly scandalous). He claims that in over twenty years of operation, he received only a single complaint about his brand name and the brand is mass-distributed by “high-end national retailers” like Urban Outfitters. Appellant’s Br. 8. Finally, he argues that two of Urban Dictionary’s seven definitions of the term “fuct” refer to his brand name, while only one definition is vulgar.
Mr. Brunetti’s proffered evidence does not change our conclusion that substantial evidence supports the Board’s findings. The Board explicitly considered Mr. Brun-etti’s declaration and found it “stretche[d] credulity” that “fuct” was chosen as an invented or coined term for “Friends yoU Can’t Trust,” given the contradictory record evidence. J.A. 9-10. Mr. Brunetti’s unverifiable claim about the number of customer complaints may demonstrate that the mark is not offensive to a certain segment of the market. That does not satisfy his burden on appeal, however, to establish that the Board lacked substantial evidence for its determination that a “substantial composite” of the American public would find the mark vulgar. And the fact that the Board could have relied on one of the other five definitions of the term “fuct” on Urban Dictionary—a website to which anyone can anonymously submit definitions—does not demonstrate that the Board’s reliance on that website is not substantial evidence. The Board reasonably focused on the highest rated definition, suggesting that it is more common or accurate than the alternative, non-vulgar definitions.
Mr. Brunetti argues that even if FUCT is vulgar, § 2(a) does not prohibit the registration of vulgar marks—only “immoral” or “scandalous” marks. He argues that to be immoral or scandalous, a mark must be more than merely vulgar. He argues that extending § 2(a) to vulgar marks is contrary to the plain language of the statute.
We do not agree. We have previously held “the PTO may prove scandalousness by establishing that a mark is ‘vulgar.’” Fox, 702 F.3d at 635; see also Boulevard Entm't 334 F.3d at 1340 (“A showing that a mark is vulgar is sufficient to establish that it ‘consists of or comprises immoral ... or scandalous matter’ within the meaning of section 1052(a).”). We are bound by these holdings. ;
Even if we could overrule our prior holding that a showing of vulgarity is sufficient to establish that a mark “consists of or comprises immoral ... or scandalous matter,” we see no justification for doing so in light of the evidence of record. At the time of the passage of the Lanham Act, dictionaries defined “scandalous” as “shocking to the sense of truth, decency, or propriety,” “[g]iving offense to the conscience or moral feelings,” or “calling out [for] condemnation.” McGinley, 660 F.2d at 485-86 (citing Webster’s New International Dictionary (2d ed. 1942); Funk & Wagnalls New Standard Dictionary (1945)). Other definitions characterize scandalous as “disgraceful,” “offensive,” or “disreputable.” Id. (citing Webster’s New International Dictionary (2d ed. 1942); Funk & Wagnalls New Standard Dictionary (1945)). We see no definition of scandalous that, in light of the PTO’s fact findings, would exempt Mr. Brunetti’s mark.
We see no merit in Mr. Brunetti’s arguments relating to whether the mark is scandalous and therefore prohibited registration under § 2(a). Substantial evidence supports the Board’s finding the mark FUGT is vulgar and- therefore the Board did not err in concluding the mark is not registrable under § 2(a).
II. Section 2(a)’s Bar on Immoral or Scandalous Marks is Unconstitutional Under the First Amendment
When Mr. Brunetti filed his appeal, his constitutional argument was foreclosed by binding precedent. In McGinley, our predecessor court held the refusal to register a •■mark under § 2(a) does not bar the applicant from using the mark, and therefore does not implicate the First Amendment. 660 F.2d at ' 484. Commentators heavily criticized McGinley and our continued reliance on it, particularly in light of the mapy changes to .First Amendment jurisprudence over the last thirty years. In re Tam, 808 F.3d 1321, 1333-34 & n.4 (Fed. Cir. 2016) (en banc). We reconsidered McGinley en banc in Tam, Which held the disparagement provision of § 2(a) unconstitutional under the First Amendment because it discriminated on the basis of content, message, and viewpoint. Id. at 1334-37, 1358. We held that, although trademarks serve a commercial purpose as source identifiers in the marketplace, the disparagement provision of § 2(a) related to the expressive character of marks, not their commercial purpose. Id. at 1337-39. As either a content-based or viewpoint-based regulation of expressive speech, the disparagement provision was subject to strict scrutiny.- Id. at 1339. It was undisputed that the measure did not survive such scrutiny. Id.
We -rejected the government’s arguments that § 2(a) did not implicate ‘ the First Amendment, holding instead that the PTO’s denial of marks had a, chilling effect on speech. Id. at 1339-45. We also-rejected the government’s arguments that-. trademark registration was government speech, id. at 1345-48, and that trademark registration was a federal subsidy, id. at 1348-55. Finally, we held the disparagement provision did not survive even the lesser scrutiny afforded to commercial speech under Central Hudson Gas & Electric. Corp. v. Public. Service Commission, 447 U.S. 557, 563, 100 S.Ct. 2343, 65 L.Ed.2d 341 (1980), because the government had put forth no substantial interests justifying the regulation of speech. Tam, 808 F.3d at 1355-58.
The en banc court noted that § 2(a) contains a hodgepodge of differing prohibitions on registration, and as such, the holding in Tam was limited to § 2(a)’s disparagement provision. Id. at 1330; see also id. at 1330 n.l. However, the court left open whether other portions of §, 2 may also be unconstitutional, and held that McGinley was overruled insofar as it could prevent a future panel from reconsidering the constitutionality of other portions of § 2. Id. at 1330 n.l.
Following the issuance of our en banc decision in Tam, we requested additional briefing from both parties in this case on “the impact of the Tam decision on Mr. Brunetti’s case, and in particular whether there is any basis for treating immoral and scandalous marks differently than disparaging marks.” In re Brunetti, No. 15-1109, Docket No. 51 (Fed. Cir. Dec. 22, 2015). Both parties filed letter briefs. The government stated that “given the breadth of the Court’s Tam decision and in view of the totality' of the Court’s reasoning,” there is no reasonable basis for treating immoral or scandalous marks differently than disparaging marks. Gov’t Letter Br. 2, In re Brunetti, No. 15-1109, Docket No. 52 (Fed. Cir. Jan. 21, 2016). It maintained, however; that if the 'Solicitor General sought Supreme Court review of our en banc decision in Tam, “the government may argue that, under reasoning less sweeping than that adopted in Tam, the bar on registration of scandalous and immoral marks would survive even if the bar on registration of disparaging .marks were held invalid.” Id. at 4. The Supreme Court subsequently granted certiorari. Lee v. Tam, — U.S. -, 137 S.Ct. 80, 195 L.Ed.2d 902 (2016).
On June 19, 2017, the Supreme Court unanimously affirmed our en banc .decision in Tam. Matal v. Tam, — U.S. —, 137 S.Ct. 1744, 198 L.Ed.2d 366 (2017). The Court held that trademarks are private, not government, speech. Id. at 1757-61. Pursuant to two opinions authored by Justice Alito and Justice Kennedy, it concluded that § 2(a)’s bar on the registration of disparaging marks discriminated based on viewpoint. Id. at 1763 (Alito, J.); id. at 1765 (Kennedy, J.). The Court explained the disparagement provision “offends a bedrock First Amendment principle: Speech may not be banned on the ground that it expresses ideas that offend.” Id. at 1751 (Alito, J.); accord id. at 1766 (Kennedy, J.). The plurality opinion, authored by Justice Alito and joined by Chief Justice Roberts, Justice Thomas, and Justice Breyer, further concluded that the constitutionality of the disparagement provision could not be sustained by analyzing trademark registration as either a federal subsidy or a federal •program. Id. at 1760-63 (Alito, • J.). The remaining four participating' Justices opined, in a concurring opinion authored by Justice Kennedy, that “the viewpoint discrimination rationale renders unnecessary any extended treatment-of other questions raised by the parties.” Id. at 1765 (Kennedy, J.).
Both opinions held the disparagement provision unconstitutionally restricted free speech, left open was “the question of whether Central Hudson provides the appropriate test for deciding free speech challenges to provisions of the Lanham Act.” Id. at 1764 n.1.7 (Alito, J.); see also id. at 1767 (Kennedy, J,). Justice Alito’s opinion concluded the disparagement provision failed even the intermediate test under Central Hudson because the prohibition was not narrowly drawn to a substantial government interest. Id. at 1764-65 (Alito, J.). Justice Kennedy’s opinion concluded that, because the disparagement provision discriminates based op. viewpoint, -it was subject to heightened scrutiny, which it did not withstand. Id. at 1767-68 (Kennedy, J.). Neither opinion reached the constitutionality of other provisions of § 2 of. the Lanham Act. See, e.g., id. at 1768 (Kennedy, J.).
Following the issuance of the Supreme Court’s decision in Tarn, we requested additional briefing from the' parties regarding the impact of the Supreme Court’s decision on Mr. Brunetti’s case. In re Brunetti, No. 15-1109, Docket No. 58 (Fed. Cir. June 20, 2017). Both parties submitted -letter briefs and we heard oral argument on August 29, 2017. The government contends Tam does not resolve the constitutionality of § 2(a)’s bar on registering immoral or scandalous marks because the disparagement provision implicates viewpoint discrimination, whereas the immoral or scandalous provision is viewpoint neutral. Gov’t Letter Br. 6-9, In re Brunetti, No. 15-1109, Docket No. 60 (Fed. Cir. July 20,2017).
■ While we question the viewpoint neutrality of. the immoral or scandalous provision, we need, not . resolve that issue. Independent of whether the immoral or scandalous provision is viewpoint discriminatory, we conclude the provision, impermissibly discriminates based on content in violation of the First Amendment.
A. Section 2(a)’s Bar on Registering Immoral or Scandalous Marks is an . Unconstitutional Content-Based Restriction on Speech
The government restricts speech based on content when “a law applies to particular speech because of the topic discussed or the- idea or message expressed.” Reed v. Town of Gilbert, — U.S. -, 135 S.Ct. 2218, 2227, 192 L.Ed.2d 236 (2015). Content-based statutes are presumptively invalid. RAV v. City of St. Paul, Minn., 505 U.S. 377, 382, 112 S.Ct. 2538, 120 L.Ed.2d 305 (1992). To survive, such statutes must withstand strict scrutiny review, which requires the government to “prove that the restriction furthers a compelling interest and is narrowly tailored to achieve that interest.” Reed, 135 S.Ct. at 2231 (quoting Ariz. Free Enter. Club’s Freedom Club PAC v. Bennett, 564 U.S. 721, 734, 131 S.Ct. 2806, 180 L.Ed.2d 664 (2011)); United States v. Playboy Entm’t Grp., Inc., 529 U.S. 803, 813, 120 S.Ct. 1878, 146 L.Ed.2d 865 (2000) (“If a statute regulates speech based on its content, it must be narrowly tailored to promote a compelling Government interest. If a less restrictive alternative would serve the Government’s purpose, the legislature must use that alternative.”). Strict scrutiny applies whether a government statute bans or merely burdens protected speech. See Playboy, 529 U.S. at 812, 120 S.Ct. 1878 (“The- Government’s content-based burdens must satisfy the same rigorous scrutiny as its content-based bans.”).
The government concedes that § 2(a)’s bar on registering immoral or scandalous marks is -a content-based restriction on speech. Oral Arg. at 11:57-12:05. And the government does not assert that the immoral or scandalous provision survives strict scrutiny review. Instead, the government contends § 2(a)!s content-based bar on registering immoral or scandalous marks does not implicate the First Amendment because trademark registration' is either a government subsidy program or limited public forum. Gov’t Letter Br. 14, In re Brunetti, No. 15-1109, Docket No. 60 (Fed. Cir. July 20, 2017); Oral Arg. at 12:06-21, 18:15-39. Alternatively, the government argues trademarks are commercial speech implicating‘only the intermediate level of scrutiny set forth in Central Hudson. Gov’t Letter Br. 15, In re Brunetti, No. 15-1109, Docket No. 60 (Fed. Cir. July 20, 2017); Oral Arg. at 35:05-17. Under a less exacting degree of scrutiny, the government argues the immoral or scandalous provision is an appropriate content-based restriction tailored to substantial government interests. We consider these arguments in turn.
1. Trademark Registration is Not a Government Subsidy Program
The Spending Clause of the U.S. Constitution “provides Congress broad discretion to tax and spend for the ‘general Welfare,’ including by funding particular state or private programs or activities.” Agency for Int’l Dev. v. All. for Open Soc’y Int’l, Inc., 570 U.S. 205, 133 S.Ct. 2321, 2327-28, 186 L.Ed.2d 398 (2013); U.S. Const, art. I, § 8, cl. 1. Within this discretion is the authority to attach certain conditions to the use of its funds “to ensure they are used in the manner Congress intends.” Id. at 2328; see also Rust v. Sullivan, 500 U.S. 173, 198, 111 S.Ct. 1759, 114 L.Ed.2d 233 (1991) (“The condition that federal funds will be used only to further the purposes of a grant does not violate constitutional. rights.”). Other government-imposed conditions may impermissibly impinge the First Amendment rights of fund recipients. Pursuant to the long-established unconstitutional conditions doctrine, the government may not restrict a recipient’s speech simply because the government provides him a benefit:
[E]ven though a person has no “right” to a valuable governmental benefit and even though the government may deny him the benefit for any number of reasons, there are some reasons upon which the government may not rely. It may not deny a benefit to-a person on a basis that infringes his constitutionally protected interests—especially, his interest in freedom of speech.
Perry v. Sindermann, 408 U.S. 593, 597 (1972); accord Bd. of Cty. Comm’rs v. Umbehr, 518 U.S. 668, 674, 116 S.Ct. 2342, 135 L.Ed.2d 843 (1996) (“[T]he threat of the loss of [a valuable financial benefit] in retaliation for speech may chill speech on matters of public concern .... ”). Conditions attached to government programs may unconstitutionally restrict First Amendment rights even if the program involves Congress’ authority to direct spending under the Spending Clause. See, e.g., Agency for Int’l Dev., 133 S.Ct. at 2330-31 (holding Congress could not restrict appropriations aimed at combating the spread of HIV/AIDS to only organizations that affirmatively opposed prostitution and sex trafficking); FCC v. League of Women Voters, 468 U.S. 364, 399-400, 104 S.Ct. 3106, 82 L.Ed.2d 278 (1984) (rejecting the government’s argument that Congress’ spending power justified conditioning funding to public broadcasters on their refraining from editorializing). The constitutional line, while “hardly clear,” rests between “conditions that define the limits of the government spending program— those that specify the activities Congress wants to subsidize—and conditions that seek to leverage funding to regulate speech outside the contours of the program itself.” Agency for Int’l Dev., 133 S.Ct. at 2328.
The government argues, pursuant to the government subsidy framework articulated in Agency for International Development, that § 2(a)’s bar on registering immoral or scandalous marks is simply a reasonable exercise of its spending power, in which the bar on registration is a constitutional condition defining the limits of trademark registration. Our court rejected the applicability of this analysis to trademark registration, 9-3, in our en banc decision in Tam.1 808 F.3d at 1348-55. The four Justices who reached the issue in Tam likewise held the government subsidy framewprk does not apply to trademark registration! 137 S.Ct. at 1761 (Alito, J.). Justice Alito explained in his plurality opinion that while the constitutional framework articulated in Agency for International Development “‘is not always self-evident,’ no difficult question is presented here.” Id. (quoting 133 S.Ct. at 2330 (alterations omitted)).
Unlike trademark registration, the programs at issue in the Supreme Court’s cases upholding the constitutionality of conditions under the Spending Clause necessarily and directly implicate Congress’ power to spend or control government property. For example, Rust addressed a condition on the distribution of federal funds for family planning services. 500 U.S. at 177, 111 S.Ct. 1759. The' Supreme Court’s plurality opinion in United States v. American Library Association, Inc. upheld a condition on federal funding for Internet access to public libraries. 539 U.S. 194, 212, 123 S.Ct. 2297, 156 L.Ed.2d 221. (2003). While Regan v. Taxation with Representation of Washington concerned tax exemptions and deductions, the Supreme Court specified, “[b]oth tax exemptions and tax-deductibility are a form of subsidy that is administered through' the tax' system.” 461 U.S. 640, 544, 103 S.Ct. 1997, 76 L.Ed.2d 129 (1983). “The federal registration of a trademark is nothing like the programs at issue in these cases;” Tam, 137 S.Ct. at 1761 (Alito, J.).
Trademark registration does not implicate Congress’ power to spend funds. An applicant does not receive federal funds upon the PTO’s consideration of, or grant of, a trademark. The only exchange of funds flows from the applicant to the PTO. The applicant pays the applicable trademark process and service fees set forth in 37 C.F.R. § 2.6(a)(1), which are then made “available to the Director to carry out the activities of the [PTO].” 35 U.S.C. § 42(c)(1). As explained in our en banc opinion in Tam, since 1991, trademark registration fees—not appropriations from taxpayers—have entirely funded the direct operating expenses associated with trademark registration. 808 F.3d at 1353 (citing, e.g., Figueroa v. United States, 466 F.3d 1023, 1028 (Fed. Cir. 2006)). Congress’ authority to direct funds is thus not implicated by either the operating expenses necessary to examine a. proposed mark or the PTO’s ultimate grant of trademark registration. Of course, trademark registration does not persist entirely independent ’ of federal funds. The government must expend certain federal funds, including but not limited to the cost of PTO employee benefits and costs associated with trademark enforcement, in connection with trademark registration. See id, (citing Figueroa, 466 F.3d at 1028). But to the extent government resources are tangentially, involved with trademark registration, “just about every -government service requires the expenditure of ■' government funds.” Tam, 137 S.Ct. at 1761 (listing, for example, police and fire .protection, and copyright and motor vehicle registrations) (Alito, J.); Tam, 808 F.3d at 1353 (“Trademark registration does not implicate the Spending Clause merely because of this attenuated spending, else every benefit or regulatory program provided by the government would implicate the Spending Clause.”). The government’s involvement in processing and issuing trademarks does not transform trademark registration into a government subsidy.
Nor is the grant of trademark registration a subsidy equivalent. “Registration is significant. The Lanham Act confers important legal rights and benefits on trademark owners who register their marks.” B&B Hardware, Inc. v. Hargis Indus., Inc., — U.S.-, 135 S.Ct. 1293, 1300, 191 L.Ed.2d 222 (2015) (quotation marks omitted). These benefits are numerous and include the “right to exclusive nationwide use of that mark where there was no prior use by others,” Tam, 808 F.3d at 1328, a presumption of validity, 15 U.S.C. § 1057(b), incontestability in certain situations, id. § 1065, the right to sue in federal court, id. § 1121, the right to recover treble damages for willful infringement, id. § 1117, a complete- defense to. state or common law claims of trademark dilution, id. § 1125(c)(6), the assistance of U.S. Customs and Border Protection in restricting importation of infringing or counterfeit goods, id. § 1124; 19 U.S.C. § 1526, the right to prevent “cybersquatters” from misappropriating a domain name, 15 U.S.C. § 1125(d), and qualification for a simplified process for obtaining recognition and protection of a mark in countries that have signed the Paris Convention; see id. § 1141b (Madrid Protocol); Paris Convention for. the Protection of Industrial Property art. 6 'quinquies, July 14, 1967, 21 U.S.T. 1583, 828 U.N.T.S. 305. While these benefits are valuable, they are not analogous to Congress’ grant of federal funds. The benefits of trademark registration arise from the statutory framework of the Lanham Act, and the Lanham Act in turn derives from the Commerce Clause.
Our sister courts confirm that when government registration does not implicate Congress’ authority under the Spending Clause, the government subsidy line of case law does not govern the constitutionality of § 2(a)’s bar on registering immoral or scandalous marks. See, e.g., Dep’t of Tex., Veterans of Foreign Wars v, Tex. Lottery Comm’n, 760 F.3d 427, 436 (6th Cir. 2014) (en banc) (holding a bingo program that was “merely licensed and regulated by the state” was “wholly distinguishable from the subsidies in Taxation with Representation and Rust simply because no public monies or ‘spending1 by the state are involved”); Bullfrog Films, Inc. v. Wick, 847 F.2d 602, 503, 509 (9th Cir. 1988) (explaining that a treaty under which certain “educational,, scientific and cultural audio-visual- materials” were granted various benefits, but no federal funds, was “fundamentally different” from government subsidy programs). As the D.C, Circuit noted, “[t]he Supreme Court has never extended the subsidy doctrine to situations not involving financial benefits.” Autor v. Pritzker, 740 F.3d 176, 182-83 (D.C. Cir. 2014) (declining to apply the subsidy doctrine to a presidential directive that impacted committee members who were unpaid). We can see no reason to treat trademark registration differently.
If the government is correct that a registration program, such as this, gives the government the authority to regulate the content of speech, then' every government registration program would, provide the government with similar censorship authority. For example, there is no principled basis to distinguish between the registration of trademarks and the registration of copyrights under the government program rationale. The subsidy line of case law cannot justify the government’s content-based bar on registering immoral or scandalous marks. -
2,- Trademark Registration is Not a Limited Public Forum
The constitutionality of speech restrictions on government property are analyzed under the Supreme Court’s “forum analysis,” which “determine[s] when a governmental entity, in regulating property in its charge, may place limitations on speech.” See, e.g., Christian Legal Soc’y of the Univ. of Cal. Hastings Coll. of the Law v. Martinez, 561 U.S. 661, 669, 130 S.Ct. 2971, 177 L.Ed.2d 838 (2010). The forum analysis is driven by the principle that “the government need not permit all forms of speech on property that it owns and controls.” Int’l Soc. for Krishna Consciousness, Inc. v. Lee, 505 U.S. 672, 678, 112 S.Ct. 2701, 120 L.Ed.2d 541 (1992); see also Lehman v. City of Shaker Heights, 418 U.S. 298, 303, 94 S.Ct. 2714, 41 L.Ed.2d 770 (1974) (“In much the same way that a newspaper or periodical, or even a radio or television station, need not accept every proffer of advertising from the general public, a city transit system has discretion to develop and make reasonable choices concerning the type of advertising that may be displayed in its vehicles.”). The government may not, however, restrict all private speech on its property solely because it is the owner. To determine the constitutional bounds of speech restrictions on government property, the forum analysis instructs us to first’ classify the government’s property as one of three forums.
The first two forums are traditional public forums and designated public forums. Traditional public forums are places such as “streets and parks which have immemorially been held in trust for the use of the public and, time out of mind, have been used for purposes of assembly, communicating thoughts between citizens, and discussing public questions,” Perry Educ. Ass’n v. Perry Local Educators’ Ass’n, 460 U.S. 37, 45, 103 S.Ct. 948, 74 L.Ed.2d 794 (1983) (quotation marks omitted). Designated public forums are created when “government property that has not traditionally been regarded as a public forum is intentionally opened up for that purpose.” Pleasant Grove City, Utah v. Summum, 555 U.S. 460, 469, 129 S.Ct. 1125, 172 L.Ed.2d 853 (2009). In these foruips, “the government’s ability to permissibly restrict expressive conduct is very limited.” United States v. Grace, 461 U.S. 171, 177, 103 S.Ct. 1702, 75 L.Ed.2d 736 (1983). Content-based restrictions on speech “must be narrowly tailored to serve a compelling government interest, and restrictions based on viewpoint are prohibited.” Pleasant Grove, 555 U.S. at 469, 129 S.Ct. 1125 (internal citation omitted).
The remaining forum category is the limited public forum, at times referred to as a non-public forum. Limited public forums are places the government has “limited to use by certain groups or dedicated solely to the discussion of certain subjects.” Id. at 470, 129 S.Ct. 1125. As with traditional and designated public forums, regulations that discriminate based on viewpoint in limited public forums are presumed unconstitutional. Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 830, 115 S.Ct. 2510, 132 L.Ed.2d 700 (1995). Content-based restrictions on speech are subject to a lesser degree of scrutiny and remain constitutional “so long as the distinctions drawn are reasonable in light of the purpose served by the forum.” Cornelius v. NAACP Legal Def & Educ. Fund, Inc., 473 U.S. 788, 806, 105 S.Ct. 3439, 87 L.Ed.2d 567 (1985). Thus, where the government has opened its property for a limited purpose, it can constitutionally restrict speech consistent with that purpose as long as “the regulation on speech is reasonable and not an effort to suppress expression merely because officials oppose the speaker’s view.” Perry, 460 U.S. at 46, 103 S.Ct. 948.
The government argues that the federal trademark registration program is a limited public forum, subjecting § 2(a)’s content-based restriction on marks comprising immoral or scandalous subject matter to a less demanding degree of scrutiny. Gov’t Letter Br. 14-15, In re Brunetti, No. 15-1109, Docket No. 60 (Fed. Cir. July 20, 2017). Without articulating why the federal trademark registration program is a limited public forum, the government’s letter brief analogizes trademark registration to city buses and a military cemetery. Id. at 14. At oral argument, the government identified the principal register as the limited public forum, which it contended is a metaphysical forum much like the forum at issue in Rpsenberger. Oral Arg. at 28:40-58.2 :
The Supreme Court has found the existence of a limited public forum only when the government restricts speech on its own property. At one end of that spectrum are venues that are owned and controlled by government entities. See, e.g., Greer v. Spock, 424 U.S. 828, 838, 96 S.Ct. 1211, 47 L.Ed.2d 505 (1976) (military base); Jones v. N.C. Prisoners’ Labor Union, Inc., 433 U.S. 119, 134, 97 S.Ct. 2532, 53 L.Ed.2d 629 (1977) (prison facilities); Int'l Soc. for Krishna Consciousness, 505 U.S. at 680-83, 112 S.Ct. 2701 (Port Authority airport terminal). These cases unquestionably concern “a governmental entity, ... regulating property in its charge.” See Christian Legal, 561 U.S. at 669, 130 S.Ct. 2971. Other cases involve property that is clearly government owned, although present in public locations. See, e.g., United States v. Kokinda, 497 U.S. 720, 727-30, 110 S.Ct. 3115, 111 L.Ed.2d 571 (1990) (sidewalk outside of Postal Service); Members of City Council of L.A. v. Taxpayers for Vincent, 466 U.S. 789, 814, 104 S.Ct. 2118, 80 L.Ed.2d 772 (1984) (public utility poles). Several of the Court’s remaining limited public forum cases involve speech restrictions that occur on public school property. See, e.g., Christian Legal, 561 U.S. at 679 n.12, 130 S.Ct. 2971 (registered student organization); Good News Club v. Milford Cent. Sch., 533 U.S. 98, 106, 121 S.Ct. 2093, 150 L.Ed.2d 151 (2001) (public school opened for instruction and recreation); Lamb’s Chapel v. Ctr. Moriches Union Free Sch. Dist., 508 U.S. 384, 390-92, 113 S.Ct. 2141, 124 L.Ed.2d 352 (1993) (public school opened for social, civic, and recreational uses); Perry, 460 U.S. at 46-47, 103 S.Ct. 948 (public school mail facilities).
While some of the Supreme Court’s limited public forum cases have involved forums that exist “more in a metaphysical than in a spatial or geographic sense,” these forums have nonetheless been tethered to government properties. See Rosenberger, 515 U.S. at 830, 115 S.Ct. 2510. In Rosenberger, the Supreme Court considered a University’s distribution of funds through a Student Activities Fund (“SAF”) intended to “support a broad range of extracurricular student activities that ‘are related to the educational purpose of the University.’” Id. at 824, 115 S.Ct. 2510. The Court concluded the SAF was a limited public forum that “effects a sweeping restriction on student thought and student inquiry in the context of University sponsored publications.” Id. at 829-30, 836, 115 S.Ct. 2510. Although the SAF was “metaphysical”—in that it concerned use of the University’s funds rather than the University’s facilities, id. at 830, 115 S.Ct. 2510— the effect of its restrictions on speech were felt on the government’s property, the University. See id. at 836, 115 S.Ct. 2510 (explaining the SAF “risks the suppression of free speech and creative inquiry in one of the vital centers for the Nation’s intellectual life, its college and university cam-' puses”). The forum at issue in Cornelius likewise involved a more abstract forum— a charity drive—but that drive was “conducted in the federal workplace during working hours.” 473 U.S. at 790, 105 S.Ct. 3439. And while the Supreme Court has applied the forum analysis to broadcasting, it did so in the context of a state-owned broadcaster’s sponsorship of a particular debate at its facilities. See Ark. Educ. Television Comm’n v. Forbes, 523 U.S. 666, 669, 672, 118 S.Ct. 1633, 140 L.Ed.2d 875 (1998).
Because trademarks are by definition used in commerce, the trademark registration program bears no resemblance to these limited public forums. The speech that flows from trademark registration is not tethered to a public school, federal workplace, or any other government property. A principal feature of trademarks is that they help “consumers identify goods and services that they wish to purchase, as well as those they want to avoid.” Tam, 137 S.Ct. at 1751. “These marks make up part of the expression of everyday life, as with the names of entertainment groups, broadcast networks, designer clothing, newspapers, automobiles, candy bars, toys, and so on.” Id. at 1768 (Kennedy, J.). By their very purpose, trademarks exist to convey messages throughout commerce. It is difficult to analogize the Nike swoosh or the Nike JUST DO IT mark located on a Nike shirt in a Nike store as somehow a government created limited public forum. The registration and use of registered trademarks simply does not fit within the rubric of public or limited public forum cases. “[T]he forum analysis requires consideration not only of whether government property has been opened to the public, but also of the nature and purpose of the property at issue.” Preminger, 517 F.3d 1299 (internal citations omitted); see also Cornelius, 473 U.S. at 805, 106 S.Ct. 3439 (examining the nature of the government property involved).
A snapshot of marks recently rejected under the immoral or scandalous provision reveals the breadth of goods and services impacted by § 2(a)’s bar on such marks, including speech occurring on clothing, books, websites, beverages, mechanical Contraptions, and live entertainment. These refusals chill speech anywhere from the Internet to the grocery store. And none of them involve government property over which the government can assert a right to.“legally preserve the property under its control for the use to which it is dedicated.” Lamb’s Chapel, 508 U.S. at 390, 113 S.Ct. 2141.
That registered, marks also appear on the government’s principal register does not transform trademark registration into a limited public forum. The government does not open the principal register to any exchange of ideas—it is ancillary to trademark registration. The principal register is simply a database identifying the marks approyed for use in commerce. Oral Arg. at, 29:28-41.- Apart from its function as a database, the government has been unable to define exactly what the principal register is, or where it is located. Id, at 29:34-64. If the government can constitutionally restrain the expression of private speech in commerce because such speech is identified in a government database, so too could the government restrain speech occurring on private land or in connection with privately-owned vehicles, simply because those private properties are listed in a database. Cf. Tam, 137 S.Ct. at 1760 (“For if the registration of trademarks constituted government speech, other systems of government registration could easily be characterized in the same way”). As the government recognized, such a suppression of speech would raise serious concerns under the unconstitutional conditions doctrine. See Oral Arg. at 29:56-30:34 ,(“[T]he key difference there is the application of the unconstitutional conditions doctrine, which is the significant constraint on the government’s ability to abuse its power over something like a land registry to influence speech outside a program.”). The government fails to articulate a reason why the government’s listing of registered trademarks in a database creates a limited public forum. And. if it did then every government registration program including titles to land, registration of cars, registration of wills or estates, copyrights, even marriage licenses could similarly implicate a limited public forum. We thus conclude- that government registration of trademarks does not create a limited public forum in which the government can more freely restrict speech.
3. The Prohibition on the Registration of Immoral or Scandalous Trademarks Targets the Expressive Content of Speech and Therefore Strict Scrutiny Should Be Applied
Commercial speech is speech which does “no more than propose a commercial transaction.” Va. State Bd. of Pharmacy v. Va. Citizens Consumer Council, Inc., 425 U.S. 748, 762 (1976) (citation omitted). Trademarks certainly convey a commercial message, but not exclusively so. There is no doubt that trademarks “identify the source of a product or service, and therefore play a role in the ‘dissemination of information as to who is producing and selling what product, for what reason, and at what price.’” Tam, 808 F.3d at 1338 (quoting Va. State Bd. of Pharmacy, 425 U.S. at 765). However, trademarks—including immoral or scandalous trademarks—also “often have an expressive content.” Tam, 137 S. Ct. at 1760. For immoral or scandalous marks, this message is often uncouth. But it can espouse a powerful cause. See, e.g., FUCK HEROIN, Appl. No. 86,361,326; FUCK CANCER, Appl. No. 86,290,011; FUCK RACISM, Appl. No. 85,608,559. It can put forth a political view, see DEMOCRAT.BS, Appl. No. 77,042,069, or REPUBLICAN.BS, Appl. No. 77,042,071. While the speech expressed in trademarks is brief, “powerful messages can sometimes be-conveyed in just a few words.” Tam, 137 S. Ct. at 1760.
The test used by the PTO to prohibit immoral or scandalous marks is'whether a “substantial composite of the general public” 3 would find the mark “shocking to the sense of truth, decency, or propriety; disgraceful; offensive; disreputable; ... giving offense to the conscience or moral feelings; ... or calling but for condemnation.” Fox, 702 F.3d at 665. There can be no question that the immoral or scandalous prohibition targets the expressive components of the speech. As in this case, the agency often justifies its’ rejection of marks on the grounds that they convey offensive ideas. J.A. 8-9 (explaining that Mr. Brunetti’s use of his trademark' is scandalous because his mark “objectifies women and offers degrading examples of extreme misogyny” and contains a theme “of extreme nihilism” with “anti-social imagery” and is “lacking in taste”). These are each value judgments about the expressive message behind the trademark. Whether marks comprise ¡immoral or scandalous subject matter hinges on the expressive, not source-identifying, nature of trademarks.
While different provisions of the Lan-ham Act may appropriately be classified as targeting a mark’s source-identifying information—for example, § 2(e)’s bar on registering marks that are “merely descriptive” or “geographically descriptive"—the immoral or scandalous provision targets a mark’s expressive message, which is separate and distinct from the commercial purpose of a mark as a source identifier. Justice Kennedy explained in his concurrence: “The central purpose of trademark registration is to facilitate source identification.... Whether a mark is disparaging bears no plausible relation to that goal.” 137 S.Ct. at 1768 (Kennedy, J.). We find the same logic applies to the immoral or scandalous prohibition. As in the case, of disparaging marks, - the PTO’s rejections under § 2(a)’s bar on immoral or scandalous marks are necessarily based in the government’s belief that the rejected mark conveys an expressive message— namely, a message that is scandalous or offensive to a substantial composite of the general population. See Tam, 808 F.3d at 1338. Section 2(a) regulates the expressive components of speech, not the commercial components of speech, and as such it should be subject-to strict scrutiny. See Sorrell v. IMS Health Inc., 564 U.S. 552, 565, 131 S.Ct. 2653, 180 L.Ed.2d 544 (2011). There is no dispute that § 2(a)’s bar on the registration of immoral’or scandalous marks is unconstitutional if strict scrutiny applies.
4. Section 2(a)’s Bar on Immoral or Scandalous Marks Does Not Survive Intermediate Scrutiny
Section 2(a)’s bar on the registration of immoral or scandalous marks is unconstitutional even if treated as a regulation of purely commercial speech reviewed according to the intermediate scrutiny framework established in Central Hudson, 447 U.S. at 566, 100 S.Ct. 2343. Intermediate scrutiny requires that “the State must show at least that the statute directly advances a substantial governmental interest and that the measure is drawn to achieve that interest.” Sorrell, 564 U.S. at 572, 131 S.Ct. 2653.
Commercial speech is subject to a four-part test which asks whether (1) the speech concerns lawful activity and is not misleading; (2) the asserted government interest is substantial; (3) the regulation directly advances that government interest; and (4) whether the regulation is “not more extensive than necessary to serve that interest.” Central Hudson, 447 U.S. at 566, 100 S.Ct. 2343; see also Bd. of Tr. of State Univ. of N.Y. v. Fox, 492 U.S. 469, 479-80, 109 S.Ct. 3028, 106 L.Ed.2d 388 (1989) (explaining the fourth prong of Central Hudson requires “not necessarily the least restrictive means but ... a means narrowly tailored to achieve the desired objective”). “Under a commercial speech inquiry, it is the State’s burden to justify its content-based law as consistent with the First Amendment.” Sorrell, 564 U.S. at 565, 131 S.Ct. 2653.
The immoral or scandalous provision clearly meets the first prong of the Central Hudson test, which requires we first confirm the speech “concern lawful activity and not be misleading.” 447 U.S. at 566, 100 S.Ct. 2343. Section 2(a)’s provision barring immoral or scandalous marks, like the disparagement provision, does not address misleading, deceptive, or unlawful marks. Rather it is concerned with whether a mark is offensive, scandalous, or vulgar to a substantial composite of the general public.
Central Hudson’s, second prong, requiring a substantial government interest, is not met. The only government interest related to the immoral or scandalous provision that we can discern from the government’s briefing is its interest in “protecting public order and morality.” Gov’t Letter Br. 15 & n.6, In re Brunetti, No. 15-1109, Docket No. 60 (Fed. Cir. July 20, 2017).4 At oral argument, the government struggled to identify the substantial interest in barring registration of trademarks comprising immoral or scandalous subject matter. The government framed its interest based on the government’s own perception of proposed marks, including what .types of marks the government would “want to promote” or “has deemed to be most suitable.” Oral Arg. at 22:35-41, 22:56-23:00. At another point, the government indicated its interest is to shield its examiners from immoral or scandalous marks: “whether or not its examiners are forced to decide whether one drawing of genitalia is confusingly similar to another drawing of genitalia.” Id. at 21:51-22:12. Ultimately, the government stated, “Congress’ primary interest is the promotion of the use of non-scandalous marks in commerce.” Id. at 23:33-42; see also id. at 25:21-32 (“Promoting commerce that doesn’t include the use of source identifiers that are graphic sexual images or profanities that are going to be off-putting to a substantial composite of the public.”). Whichever articulation of the government’s interest we choose, the government has failed to identify a substantial interest justifying its suppression of immoral or scandalous trademarks.5 -
First, the government does not have a substantial interest in promoting certain trademarks over others. The Supreme Court rejected the government’s claim that trademarks are government speech. Tam, 137 S.Ct. at 1757-61. Our conclusion that trademark registration is neither a government subsidy nor a limited public forum forecloses any remaining interest 'the government may have in approving only marks it “has deemed to be most suitable.” Oral Arg, at 22:56-23:00; see also Tam, 137 S.Ct. at 1760-63 (plurality rejecting the government subsidy argument) (Alito, J.),
Second, Supreme Court precedent makes clear that the government’s general interest in protecting the public from marks it deems “off-putting,” whether to protect the general public or the government itself, is not a substantial interest justifying broad suppression of speech. “[T]he fact that society may find speech offensive is not a sufficient reason for suppressing it.” Hustler Magazine, Inc. v. Falwell, 485 U.S. 46, 55, 108 S.Ct. 876, 99 L.Ed.2d 41 (1988); Bolger v. Youngs Drug Prods. Corp., 463 U.S. 60, 71, 103 S.Ct. 2875, 77 L.Ed.2d 469 (1983) (“At least where obscenity is not involved, we have consistently held that the fact that protected speech may be offensive to some does not justify its suppression.” (citation omitted)); Cohen v. California, 403 U.S. 15, 21, 91 S.Ct. 1780, 29 L.Ed.2d 284 (1971) (“[T]he mere presumed presence of unwitting listeners or viewers does not serve automatically to justify curtailing all speech capable of giving offense.”); Cox, 379 U.S. at 551 (“[Constitutional rights may not be denied simply because of hostility to their assertion or exercise.”). “Where the designed benefit of a content-based speech restriction is to shield the sensibilities of listeners, the general rule is that the right of expression prevails, even where no less restrictive alternative exists.” Playboy, 529 U.S. at 813, 120 S.Ct. 1878.
The Supreme Court’s decision in Tam supports our conclusion that the government’s interest in protecting the public from off-putting marks is an inadequate government interest for First Amendment purposes. See, e.g., 137 S.Ct. at 1764 (applying Central Hudson and rejecting the government’s ■ “interest in preventing speech expressing ideas that offend” because “that idea strikes at the heart of the First Amendment”) (Alito, J.). In Tam, the Court acknowledged that it is a “bedrock First Amendment principle” that “Speech may not be banned on the ground that it expresses ideas that offend.” Tam, 137 S.Ct. at 1751 (Alito, J.); see also id. at 1767 (“[T]he Court’s eases have long prohibited the government from justifying a First Amendment burden by pointing to the offensiveness of the speech to be suppressed.”) (Kennédy, J.). Both Justice Alito’s and Justice Kennedy’s opinions support their conclusions that the disparagement provision is unconstitutional citing cases holding “the public expression of ideas may not be prohibited merely because the ideas are themselves offensive to some of their hearers.” Id, at 1763 (collecting cases) (quoting Street v. New York, 394 U.S. 576, 592, 89 S.Ct. 1354, 22 L.Ed.2d 572 (1969)) (Alito, J.); id, at 1767 (citing Justice Ali-to’s opinion at 1763-64) (Kennedy, J.); see also Texas v. Johnson, 491 U.S. 397, 414, 109 S.Ct. 2533, 105 L.Ed.2d 342 (1989) (“If there is a bedrock principle underlying the First Amendment, it is that the government may not prohibit the expression of an idea simply because society finds the idea itself offensive or disagreeable.”); Coates v. Cincinnati, 402 U.S. 611, 615, 91 S.Ct. 1686, 29 L.Ed.2d 214 (1971) (“[M]ere public intolerance or .animosity cannot be the basis for abridgment of these constitutional freedoms,”). The government’s interest in suppressing speech because it is ..off-putting is unavailing.
While the government’s interest in Tam related to a viewpoint-based restriction on speech, we note the cases on which the Supreme Court relied are not so limited. The cases cited in Tam are directed to speech that may be offensive, but not all involve speech that is disparaging or viewpoint discriminatory. Many involve speech that, rather than, disparaging others, involved peaceful, demonstrations. See, e.g., Bachellar v. Maryland, 397 U.S. 564, 566-67, 90 S.Ct. 1312, 25 L.Ed.2d 570 (1970) (peaceful Vietnam war protest carrying signs such as “Make . Love not War”); Tinker v. Des Moines Indep. Cmty. Sch. Dist., 393 U.S. 503, 509-14, 89 S.Ct. 733, 21 L.Ed.2d 731 (1969) (wearing black armbands to protest Vietnam war);. Cox v. Louisiana, 379 U.S. 536, 545, 550-51, 85 S.Ct. 453, 13 L.Ed.2d 471 (1965) (protesting segregation and discrimination); De Jonge v. Oregon, 299 U.S. 353, 365, 57 S.Ct. 255, 81 L.Ed. 278 (1937) (peaceful political meeting), Several other cases do not appear to involve viewpoint discrimination at all. For example, Hustler Magazine concerned a parody interview of Jerry Falwell in which the actor playing him stated his “‘first time’ was during a drunken incestuous rendezvous with his mother in an outhouse.” 485 U.S. at 48,108 S.Ct. 876. While such a parody interview is offensive, its function as a parody does not clearly involve the expression of beliefs, ideas, or perspectives. Similarly, the. ordinance at issue in Coates was not limited to restricting disparaging speech or certain viewpoints, but prohibited any conduct perceived as “annoying to persons. passing by.” 402 U.S. at 611, 91 S.Ct. 1686. The Supreme Court’s narrative that the government cannot justify restricting speech because it offends, together with its reliance on cases involving a variety of different speech restrictions, reinforce our conclusion that the government’s interest in protecting the public from off-putting marks is not substantial.
Finally, the government does not have a substantial interest in protecting the public from scandalousness and profanities. The government attempts to justify this interest by pointing to the Supreme Court’s decision,in FCC v. Pacifica Foundation, 438 U.S. 726, 98 S.Ct. 3026, 57 L.Ed.2d 1073 (1978). In Pacifica, the Supreme Court upheld the constitutionality of the FCC’s declaratory order determining that an afternoon radio broadcast of George Carlin’s “Filthy Words” monologue was indecent and potentially sanctionable. Id. at 730-32, 98 S.Ct. 3026. The Court explained “references to excretory and sexual material .., surely lie at the periphery of First Amendment concern.” Id. at 742, 98 S.Ct. 3026. The Court justified the FCC’s order, however, because radio broadcasting has “a uniquely pervasive presence in the lives of all Americans” and is “uniquely accessible to. children, even those too young to read,” confronting Americans “in the privacy of the home, where the individual’s right to be left alone plainly outweighs the First Amendment rights of an intruder.” Id. at 749, 98 S.Ct. 3026. The Court stressed: “It is appropriate to emphasize the narrowness of our holding.” Id. at 750, 98 S.Ct. 3026. Subsequent precedent explained that other mediums of communication, such as dial-in-; services or the Internet, are “manifestly different from a situation in which a listener does not want the received message.” Sable Commc’ns of Cal., Inc. v. FCC, 492 U.S. 115, 128, 109 S.Ct. 2829, 106 L.Ed.2d 93 (1989) (“Unlike an unexpected outburst on a radio broadcast, the message received by one who places a call to a dial-a-pom service is not so invasive or surprising that it prevents an unwilling listener from avoiding exposure to it.”); Reno v. Am. Civil Liberties Union, 521 U.S. 844, 868-69, 117 S.Ct. 2329, 138 L.Ed.2d 874 (1997) (explaining Pacifica does not control because “the Internet is not as ‘invasive’ as radio or television”).
The government’s interest in protecting the public from profane and scandalous marks is not akin to the government’s interest in protecting children and other unsuspecting listeners from a barrage of swear words over the radio in Pacifica. A trademark is not foisted upon listeners by virtue of its being registered. Nor does registration make a scandalous mark more accessible to children. Absent any concerns that trademark registration invades a substantial privacy interest in an intolerable manner, the government’s interest amounts to protecting everyone, including adults, from scandalous content. But even when “many adults themselves would find the material highly offensive,” adults have a First Amendment right to view and hear speech that is profane and scandalous. Playboy, 529 U.S. at 811, 120 S.Ct. 1878 (First Amendment right to view “sexually explicit adult programming or other programming that is indecent”); Sable, 492 U.S. at 115, 109 S.Ct. 2829 (“Sexual expression which is indecent but not obscene is protected by the First Amendment.”). In crafting a substantial government interest, “the government may not ‘reduce the adult population ... to ... only what is fit for children.’ ” Bolger, 463 U.S. at 73, 103 S.Ct. 2875 (citation omitted); cf. Sable, 492 U.S. at 131, 109 S.Ct. 2829 (“[T]he statute’s denial of adult access to telephone messages which are indecent but not obscene far exceeds that which is necessary to limit the access of minors to such messages .,..”).
Even if we were to hold that the government has a substantial interest in protecting the public from scandalous or immoral marks, the government could not meet the third prong of Central Hudson, which requires the regulation directly' advance the government’s asserted interest. 447 U.S. at 566, 100 S.Ct. 2343. As the government has repeatedly exhorted, § 2(a) does not directly prevent applicants from using their marks. Regardless of whether a trademark is federally registered, an applicant can still brand clothing with his mark, advertise with it on the television or radio, or place it on billboards along the highway. In'this electronic/Internet age, to the extent that the government seeks to protect the general population from scandalous material, with all due respect, it has completely failed.
Finally, no matter the government’s interest, it cannot meet the fourth prong of Central Hudson, The PTO’s inconsistent application of the immoral or scandalous provision creates an “uncertainty [that] undermines the likelihood that the [provision] has been carefully tailored.” See Reno, 521 U.S. at 871, 117 S.Ct. 2329. Nearly -identical marks have been approved by one examining attorney and rejected as scandalous or immoral by another. The PTO registered the mark FUGLY for use on clothing, but refused registration for use on alcoholic beverages. Compare Reg. No. 5,135,615, with Appl. No. 78,866,347. See also COCAINE, Appl. No. 78,829,207 (rejected), COCAINE, Reg. No. 1,340,874 (accepted). The PTO registered NO BS! BRASS, Reg. No. Reg. No. 5,053,-827, for entertainment services but rejected NO BS ZONE, Appl. No. 76,626,390, for internet training. NO $#! + , Appl. No. 85,855,449, was rejected, but $#*! MY DAD SAYS, Reg. No. 4,142,745, was allowed. See also ROLL TURD, Appl. No. 86,448,988 (rejected), TURD HERDERS, Reg. No. 5,180,286 (registered). Although the language in these marks is offensive, we cannot discern any pattern indicating when the incorporation of an offensive term into a mark will serve as a bar to registration and when it will not.
One commentator has written that, of the forty marks containing the acronym MILF for which written records were available as of 2011, twenty marks received an office action refusing registration based on § 2(a), while twenty did not. Anne Gilson LaLonde & Jerome Gilson, Trademarks Laid Bare: Marks That May Be Scandalous or Immoral, 101 Trademark Rep. 1476, 1478-82 (2011). It is difficult |» understand what distinguished the refused marks, which included GOT MILF (clothing), MILF MANIA (adult online services), MILF SEEKER (adult entertainment services), and FROM SOCCER MOM TO MILF (self-help books for women), from the marks which were registered including DIARY OF A MILF (adult online services), BACKROOM MILF (adult online services), FAT MILF (sandwich), and MILF NEXT DOOR (adult online services). Id. Another empirical study identified words that served as the basis of a § 2(a) refusal in some marks but were material components of other marks approved by the PTO. The authors found that to the extent there are general trends in the PTO’s treatment of the offensive terms, “those general trends are apparently inconsistent with one another.” Meghan M. Carpenter & Mary Garner, NSFW: An Empirical Study of Scandalous Trademarks, 33 Cardozo Arts & Ent. L.J. 321, 356-61 (2015). Even marks that reference the indisputably vulgar term “fuck,” like the mark at issue here, are not always rejected as a matter of course. The PTO registered the mark FCUK, but rejected the marks FUCT and F**K PROJECT as scandalous. It allowed the registration of MUTHA EFFIN BINGO, Reg. No. 4,183,272, and IF WE TOUCH IT, IT’S FN GOLDEN, Reg. No. 4,100,978, but not F ALL F’S APPAREL FOR THE F’N ANGRY, Appl. No. 78,420,315.6
The Trademark Trial and Appeal Board has itself noted the vague and subjective nature of the scandalous inquiry. In re In Over Our Heads, Inc., 1990 WL 354546 at *1 (“[T]he guidelines for determining whether a mark is scandalous or disparaging are somewhat vague and the determination of whether a mark is scandalous or disparaging is necessarily a highly subjective one.”). It can no doubt be a difficult task to determine public perceptions of a trademark’s morality or immorality, -offensiveness, or even vulgarity. As the Supreme Court has explained, “it is largely because governmental officials cannot make principled distinctions in this area that the Constitution leaves matters of taste and style so largely to the individual.” Cohen, 403 U.S. at 25, 91 S.Ct. 1780. To be sure, there are other trademark’s whose offensiveness cannot be reasonably questioned; the government attached an appendix with examples of such marks which it has rejected to this court. But the subjectivity in the determination of what is immoral or scandalous and the disparate and unpredictable application of these principles cause us to conclude that the prohibition at issue in this case would also fail the fourth prong of the Central Hudson analysis.
We conclude that the government has not presented us with a substantial government interest justifying the § 2(a) bar on immoral or scandalous marks. As we concluded in Tam, “All of the government’s proffered interests boil down to permitting the government to burden speech it finds offensive.” Tam, 808 F.3d at 1357. We also conclude that the government has failed to demonstrate that its restriction will advance the interests it asserts and that it is narrowly tailored to achieve that objective. Section 2(a)’s bar on immoral or scandalous marks does not survive intermediate scrutiny under Central Hudson.
5. There Is No Reasonable Definition of the Statutory Terms Scandalous and Immoral Which Would Preserve Their Constitutionality
We construe statutes narrowly to preserve their constitutionality, when possible. See Schneider v. Smith, 390 U.S. 17, 26, 88 S.Ct. 682, 19 L.Ed.2d 799 (1968). However, “[t]he infringement of First Amendment rights will not be cured if the narrowing construction is so unforeseeable that men of common intelligence could not have realized the law’s limited scope at the only relevant time, when their acts were committed, or if the law remains excessively sweeping even as narrowed.” Gregory v. City of Chi., 394 U.S. 111, 121, 89 S.Ct. 946, 22 L.Ed.2d 134 (1969) (citations omitted). Our duty to avoid constitutional questions “is not a license for the judiciary to rewrite language enacted by the legislature. Any other conclusion, while purporting to be an exercise in judicial restraint, would trench upon the legislative powers vested in Congress.” United States v. Albertini, 472 U.S. 675, 680, 105 S.Ct. 2897, 86 L.Ed.2d 536 (1985) (citations omitted). It is thus permissible to construe a statute in a manner that preserves its constitutionality only where the construction is reasonable.
The concurrence agrees that the scandalous and immoral prohibitions as construed by the government, this court, and our predecessor court are unconstitutional. This court and its predecessor have consistently defined “scandalous” as “shocking to the sense of truth, decency, or propriety; disgraceful; offensive; disreputable; giving offense to the conscience or moral feelings; or calling out for condemnation.” Fox, 702 F.3d at 635; accord McGinley, 660 F.2d at 485; Riverbank Canning Co., 95 F.3d at 328. The concurrence proposes that we “narrow the immoral-scandalous provision’s ' scope to obscene marks in order to preserve its constitutionality.” Cone. Op. 5-6. While the legislature could rewrite the statute to adopt such a standard, we cannot.
It is not reasonable to construe the words immoral and scandalous as confined to obscene material. There is no dispute that an obscene mark would be scandalous or immoral; however, not all scandalous or immoral marks are obscene. All apples are fruit, but not all fruits are apples. As the PTO has explained, “the threshold for objectionable matter is lower for what-can be described as .‘scandalous’ than for ‘obscene.’ ” J.A. 4 (citation omitted); accord McGinley, 660 F.2d at 487 n.9. The PTO' has for a century rejected marks as scandalous or immoral that are clearly not obscene. As set forth above, many of the early cases applying the immoral or scandalous provision;-involved blasphemous marks touching on religion, which were not obscene.7
■ The Supreme Court has made clear that the definition of obscenity for purposes of the First Amendment is “material which deals-with sex in a manner appealing to prurient interest,” ie., “material having a tendency to excite lustful thoughts.” Roth v. United States, 354 U.S. 476, 487 & n.20, 77 S.Ct. 1304, 1 L.Ed.2d 1498 (1957). This “definition does not reflect the precise meaning of ‘obscene’ as traditionally used in the English language,” and instead is limited to “obscene material ‘which deals with sex’," Miller v. California, 413 U.S. 15, 20 n.2, 93 S.Ct. 2607, 37 L.Ed.2d 419 (1973) (emphasis added).
Despite the concurrence’s suggestion to the contrary, none of the dictionary defini- ■ tions cited define “immoral” or “scandalous”- in sexual terms.8 Immoral, Black’s Law Dictionary (1st ed. 1891) (“Contrary to good morals; inconsistent with the rules and principles of morality which regard men as living -in a community, and which are necessary for the public welfare, order, and decency.’’); Immoral, Webster’s Collegiate Dictionary (1898) (“Not moral; inconsistent with good morals; contrary to conscience or the divine law.’’); Scandalous, Id, (“1. Giving offense to the conscience or moral feelings. 2. Disgraceful to ■reputation; opprobrious. 3. Defamatory; libelous.”); .Immorali Webster’s Complete Dictionary (1886) (“Not moral; inconsistent rectitude; contrary to conscience or the divine -law; wicked; unjust; dishonest; vicious”); Scandalous, Id. (“1, Giving offense; exciting reprobation; calling out condemnation; extremely offensive to duty or propriety” “2. Disgraceful to reputation; bringing shame or infamy; opprobrious” “3. Defamatory; libelous”).
Unlike the terms “immoral”, and “scandalous,” the statutory terms at issue in the cases cited in the concurrence are by their' nature limited to material “which deals with sex.” See, e.g., Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 494, 105 S.Ct. 2794, 86 L.Ed.2d 394 (1985) (construing phrase “that, which incites lasciviousness or lust”); Manual Enters., Inc. v. Day, 370 U.S. 478, 482-83, 82 S.Ct. 1432, 8 L.Ed.2d 639 (1962) (opinion of Harlan, J.) (“While in common usage the words have different shades of meaning, the statute since its inception has always been taken as aimed at obnoxiously debasing portrayals of sex.” (footnote omitted)); Swearingen v. United States, 161 U.S. 446, 451, 16 S.Ct. 562, 40 L.Ed. 765 (1896) (“The words ‘obscene,’ ‘lewd,’ and ,‘lascivious,’ as used in the statute, signify that form of immorality which has relation to sexual impurity ....”). We do not see how the words “immoral” and “scandalous” could reasonably be read to be limited to material of a sexual nature. We cannot stand in the shoes of the legislature and rewrite a statute.
Conclusion
The trademark at issue is vulgar. And the government included an appendix in its briefing to the court which contains numerous highly offensive, even shocking, images and words for which individuals have sought trademark registration. Many of the marks rejected under § 2(a)’s bar on immoral or scandalous marks, including the marks discussed in this opinion, are lewd, crass, or even disturbing. We find the use of such marks in commerce discomforting, and aré not eager to see a proliferation of such marks in the marketplace. There are, however, a cadre of similarly offensive images and words that have secured copyright registration by the government. There are countless songs with vulgar lyrics, blasphemous images, scandalous books and paintings, all of which are protected under federal law. No doubt many works registered with the Copyright Office offend a substantial composite of the general public. There are words and images that we do not wish to be confronted with, not as art, nor in the marketplace. The First Amendment, however, protects private expression, even private expression which is offensive to a substantial composite of the general public. The government has offered no substantial government interest for policing offensive speech in the context of a registration program such as the one at issue in this case.
We hold'that the bar in § 2(a) against immoral or scandalous marks is unconstitutional because it violates the First Amendment. We reverse the Board’s holding that Mr. Brunetti’s mark is unregistra-ble under § 2(a).
REVERSED
. The government maintains that our en banc decision in' Tam is not binding on this pahel in light of the Supreme .Court’s decision in Tam. Oral Arg. at 12:23-13:36, 15:43-54. We question the force of this assertion because the Supreme Court did not reverse or otherwise cast doubt on the continuing validity of our government subsidy analysis and other aspects of oúr decision in Tam. See Chen v. Allstate Ins. Co., 819 F.3d 1136, 1138 n.1 (9th Cir. 2016); Balintulo v. Ford Motor Co., 796 F.3d 160, 166 n.28 (2d Cir. 2015). Because we independently reach the same conclusion as the en banc court, we need not decide whether that holding continues to bind future panel decisions in this circuit.
. Apart from the inconsistency of this argument with the government's previous representation in Tam—in which it stated it did not believe the forum analysis applied to trademark registration, and in particular that it did not "regard the register itself as a forum”— this argument fails as a legal matter. See Tam, 808 F.3d 1321, Oral Arg. at 1:14:25-1:14:58; Tam, 808 F.3d at 1353 n.12.
. The PTO justifies its refusals by. “tying censorship to the reaction of the speaker’s audience.” See Tam, 137 S.Ct. at 1766 (Kennedy, J.).
. The government’s brief also made an errant reference to its interest "in the orderly flow of commerce.” Gov’t Letter Br. 15, In re Brunetti, No. 15-1109, Docket No. 60 (Fed. Cir. July 20, 2017). While we do not question the sub-stantiality of this interest, the government has failed to articulate how this interest is in any way advanced by the immoral or scandalous prohibition, or how that provision is narrowly tailored to that interest.
. We note that the government hardly met its burden to identify a government interest at all. To identify this purported interest, the government has done no more than “taken the effect of the statute and posited'that effect as the State’s interest. If accepted, this sort of circular defense can sidestep judicial review of almost any statute, because it makes all statutes look narrowly tailored.” Simon & Schuster, Inc. v. Members of N.Y. State Crime Victims Bd., 502 U.S. 105, 120, 112 S.Ct. 501, 116 L.Ed.2d 476 (1991).
. Tjie PTO's inconsistent rejections under the immoral or scandalous provision also raise concerns about the provision’s vagueness. See Tam, 808 F.3d at 1359 (O'Malley, J., concurring) (opining that § 2(a)'s bar on disparaging marks was unconstitutionally vague and identifying examples "where there is no conceivable difference between the applied-for marks, yet one is approved and the other rejected”). We need not reach whether the immoral or scandalous provision is so vague that it violates the Fifth Amendment.
. With no authority, the concurrence suggests “the central aim of the immoral-scandalous provision ... has been sexual material.” Cone. Op. 1359-60. To the contrary, there is a long history of rejecting numerous categories of non-sexual material under this provision. See generally LaLonde & Gilson, supra, at 1510-14, 1517-33 (discussing the application of the provision to marks related to religion, drug references, violence, disparaging patriotic symbols, mild profanity, and scatological references). Moreover, the.concurrence suggests narrowing the immoral or scandalous provision to obscene material would be consistent with PTO action. The marks cited by the concurrence, like the FUCT mark at issue in this case, would not be properly refused under a prohibition limited to obscenity. See Cone. Op. 1360 n.7.
. The concurrence’s reliance on overlapping dictionary definitions of "immoral,” "scandalous,” and "obscene” ignores this important limitation. The question before us is not whether the, obscene material is "immoral” .and "scandalous,” but rather whether Congress intended the terms "immoral” or "scandalous” to be confined to material that is "obscene” for the purposes of a First Amendment analysis.