CONNECTION DISTRIBUTING CO.; Rondee Kamins; Jane Doe; John Doe, Plaintiffs-Appellants, v. Eric H. HOLDER, Jr., Attorney General, Defendant-Appellee, 557 F.3d 321

SUTTON, J., delivered the opinion of the court, in which BOGGS, C. J., BATCHELDER, DAUGHTREY, GILMAN, GIBBONS, ROGERS, COOK, McKEAGUE, GRIFFIN, and KETHLEDGE, JJ., join. KENNEDY, J. (pp. 343-61), delivered a separate dissenting opinion in which MARTIN, MOORE, COLE, CLAY, and WHITE, JJ., joined. MOORE, J. (pp. 361-67), and CLAY, J. (pp. 367-69), delivered separate dissenting opinions, in which MARTIN and COLE, JJ., joined. WHITE, J. (pp. 369-72), also delivered a separate dissenting opinion. SUTTON, Circuit Judge.

OPINION At issue in this case is whether a provision of the Child Protection and Obscenity Enforcement Act of 1988, Pub.L. No. 100-690, § 7513, 102 Stat. 4485, 4487 (codified as amended at 18 U.S.C. § 2257), violates (1) the First Amendment’s free-speech guarantee, either as applied to the plaintiffs or on its face, or (2) the Fifth Amendment’s privilege against self-incrimination. I. A. Prior to 1988, Congress attempted to prevent the exploitation of children through pornography in at least two ways. It banned all obscene pornography, whether involving children or not.

See Protection of Children Against Sexual Exploitation Act of 1977, Pub.L. No. 95-225, 92 Stat. 7 (1978) (codified as amended at 18 U.S.C. §§ 2251-2252, 2256). And it banned all other pornography involving children under the age of 18. See Child Protection Act of 1984, Pub.L. No. 98-292, 98 Stat. 204 (codified as amended at 18 U.S.C. §§ 2251-2254); Child Sexual Abuse and Pornography Act of 1986, Pub.L. No. 99-628, 100 Stat. 3510 (codified as amended in various sections of 18 U.S.C.).

No one in this case questions Congress’s authority to prevent the exploitation of children in these ways, and indeed the Supreme Court has made it clear that these kinds of regulations represent a permissible means of addressing the problem. See New York v. Ferber, 458 U.S. 747, 756-66, 102 S.Ct. 3348, 73 L.Ed.2d 1113 (1982); United States v. 12 200-Foot Reels of Super 8mm. Film, 413 U.S. 123, 126, 129-130, 93 S.Ct. 2665, 37 L.Ed.2d 500 (1973); cf. Ashcroft v. Free Speech Coal., 535 U.S. 234, 239, 122 S.Ct. 1389, 152 L.Ed.2d 403 (2002).

In 1986, the Attorney General’s Commission on Pornography determined that, although efforts to eradicate child pornography had “drastically curtailed its public presence,” they “ha[d] not ended the problem.” Final Report of the Attorney General’s Commission on Pornography 134 (1986). While “[slexual exploitation of children has retreated to the shadows,” the Commission observed, “no evidence ... suggests that children are any less at risk than before.” Id.

One lingering problem, the Commission found, was that the pornography industry’s proclivity for using youthful-looking models often made it difficult to discern whether underage models were being used in various publications and movies. Id. at 138-39. Prompted by the Commission’s report and recommendations, Congress in 1988 enacted the Child Protection and Obscenity Enforcement Act. See Am. Library Ass’n v. Barr (ALA I), 956 F.2d 1178, 1182 (D.C.Cir.1992).

Section 7513(

a) of the Act, known by its codified

section number as § 2257, attempted to address this problem by adding a reporting and verification requirement to the existing laws designed to prevent child pornography. Under § 2257 of Title 18, those who create materials depicting “actual sexually explicit conduct” must maintain records of their models’ ages and identities.

The Act defines “actual sexually explicit conduct,” 18 U.S.C. § 2257(a)(1), as “sexual intercourse, including genital-genital, oral-genital, anal-genital, or oral-anal, whether between persons of the same or opposite sex,” id. § 2256(2)(A)(i), as well as bestiality, masturbation, sadistic or masochistic abuse, and the “lascivious exhibition of the genitals or pubic area of any person,” id. § 2256(2)(A)(ii)-(v); see id. § 2257(h)(1). Under the Act’s reporting requirements, a regulated producer must examine, and retain a copy, of, each model’s or performer’s photo identification.

See id. § 2257(b); 28 C.F.R. § 75.2. It must make these records available for inspection by the government upon request. See 18 U.S.C. § 2257(c); see also 28 C.F.R. §§ 75.4-.5. And it must include a statement in its publications noting where the relevant records are kept and who maintains them. See 18 U.S.C. § 2257(e); see also 28 C.F.R. § 75.6. The requirements of the Act together with the implementing regulations apply to “primary” and “secondary” “producers” of sexually explicit images.

Primary producers are those who create a visual representation of actual sexually explicit conduct through videotapes, photographs or computer manipulations. 18 U.S.C. § 2257(h)(2)(A)(i); 28 C.F.R. § 75.1(c)(1). Secondary producers are (1) those who use such images for “assembling, manufacturing, publishing, duplicating, reproducing, or reissuing” any material containing regulated images, 18 U.S.C. § 2257(h)(2)(A)(ii), and (2) those who upload such images to a website or otherwise manage the content of the website, id. § 2257(h)(2)(A)(iii); 28 C.F.R. § 75.1(c)(2).

Primary producers must “create and maintain” records relating to all of the visual depictions they produce, indexed by performer and publication, while a secondary producer may meet its burden by obtaining a copy of the primary producer’s records. 28 C.F.R. § 75.2(a)-(b). In addition, no one may knowingly sell, transfer or offer for sale in interstate commerce materials containing covered images unless they contain the required labels. 18 U.S.C. § 2257(f)(4). A regulated entity that fails to follow these requirements is subject to criminal penalties. Id. § 2257(f).

The Act makes it a felony not to comply with these requirements, and a producer convicted of violating the Act may be fined and subject to as many as five years in prison. Id. § 2257(i). Since 1988, Congress has amended § 2257 several times: in 1990, 2003 and 2006. See Child Protection Restoration and Penalties Enhancement Act of 1990, Pub.L. No. 101-647, §§ 301(b), 311, 104 Stat. 4816, 4816-17; Prosecutorial Remedies and Other Tools to End the Exploitation of Children Today (PROTECT) Act, Pub.L. No. 108-21, § 511(a), 117 Stat. 650, 684-85 (2003); Adam Walsh Child Protection and Safety Act of 2006, Pub.L.

No. 109-248, § 502(a), 120 Stat. 587, 625-26. In addition, the Attorney General has issued implementing regulations on three occasions. See 57 Fed.Reg. 15,017 (Apr. 24, 1992); 70 Fed.Reg. 29,607 (May 24, 2005); 73 Fed.Reg. 77,432 (Dec. 18, 2008). With one exception, the main provisions of the Act challenged here — the record-keeping and labeling requirements that apply to depictions of actual sexually explicit conduct, 18 U.S.C. § 2257(a)-(c), (e) — have not materially changed since 1988.

The exception is a provision added to the law through the 2003 amendments, which allows law-enforcement officials to use records required by the statute in prosecuting certain other crimes. See Pub.L. No. 108-21, § 511(a)(1), 117 Stat. at 684. The other significant changes to the law, not directly challenged here, include: expanding the statute’s scope to cover computer-based images, see id. § 511(a)(2), 117 Stat. at 685; see also Pub.L.

No. 109-248, § 502(a)(1), 120 Stat. at 625, and clarifying that the record-keeping and labeling requirements do not reach those who merely provide services related to the images (such as film-processing, distribution and internet-access services), see Pub.L. No. 109-248, § 502(a)(4), 120 Stat. at 625-26. B.

Founded in 1976, Connection Distributing publishes several magazines devoted to “[sjwinging,” a lifestyle “philosophy that holds that monogamy is incompatible with human nature and that the freedom to share sexual experiences with other like minded couples strengthens the bond of a couple’s relationship.” Br. at 14-15. Connection’s magazines facilitate swinging by providing a venue for like-minded individuals to share their sexual interests, preferences and availability. See Connection Distrib. Co. v. Reno (Connection I), 154 F.3d 281, 285 (6th Cir.1998).

Although the magazines contain editorials and feature stories, they principally consist of advertisements — some in text alone but most in text accompanied by photographs. The advertisements serve as a form of uninhibited self-promotion, as they depict the featured individuals in graphic detail, exhibit and discuss the individuals’ preferred sexual practices and invite readers to share similar experiences with them. Id.; see also Br. at 18. Sometimes the photo advertisements depict individuals’ full bodies, including their faces, but 85-90% of the advertisers do not reveal their faces. JA 393.

A typical photograph thus portrays either just a featured body part or the full body with the face cropped or blocked out. Id.; see, e.g., JA 1021-23, 1027, 1029-30, 1036, 1052, 1075, 1089-90. Individuals do not mention their full names in the advertisements but instead share their names, addresses and phone numbers with Connection. JA 383, 393-94, 397.

Some advertisements mention individuals’ mailing addresses, see, e.g., JA 1020-23, and others mention access codes that allow readers to respond to the advertisements through Connection, see, e.g., JA 1019, 1163-65, 1171-72, which acts as something of a matchmaker by forwarding written responses to individual advertisements for a fee and by allowing subscribers to use its 900-number service to contact individual advertisers, see Connection I, 154 F.3d at 285. C. In 1995, Connection filed a complaint challenging the validity of § 2257 and its implementing regulations on First Amendment grounds.

Seeking declaratory and injunctive relief, it claimed that the statute was unconstitutional (1) as applied to Connection and its advertisers and (2) on its face. The district court denied Connection’s motion for a preliminary injunction, and a panel of this court affirmed. Focusing on the likelihood-of-success inquiry, the panel held that the claimants had little prospect of establishing that the statute violated the First Amendment as applied to Connection because the law amounted to a content-neutral regulation and survived intermediate scrutiny. Connection I, 154 F.3d at 284, 288-94, 296.

The panel did not address the merits of Connection’s facial challenge. On remand, the district court granted

summary judgment against Connection. A panel of this court reversed and remanded, directing the district court to “allow the parties additional discovery” and to “reconsider the matter in light of [four] recent Supreme Court precedent^].” Connection Distrib. Co. v. Reno (Connection II), 46 Fed.Appx. 837, 837 (6th Cir.2002). In asking the district court to reconsider its decision in the light cast by these precedents, the Connection II panel held that these intervening decisions did not affect its prior holding that intermediate scrutiny applied to this challenge. Id. at 837.

After the Connection II panel remanded the case and after Congress amended the statute in 2003, Connection filed an amended complaint and added three new plaintiffs: Rondee Kamins, the publisher of Connection; and Jane and John Doe, two anonymous adults who “wish to publish” in Connection’s publications but have “refrain[ed] from doing so for fear of having [their] identities] revealed to the government.” JA 32-33. Connection again sought a preliminary injunction, and the government again sought

summary judgment. The district court denied the preliminary injunction and granted

summary judgment to the government. II. Since Congress enacted § 2257 in 1988, two federal appellate decisions have addressed First Amendment as-applied challenges to the law — one by the D.C. Circuit, Am. Library Ass’n v. Reno (ALA II), 33 F.3d 78, 87-90 (D.C.Cir.1994), the other by this court, Connection I, 154 F.3d at 288-94. Today’s dispute presents a renewal of Connection’s as-applied challenge to the law’s record-keeping and disclosure provisions and a facial challenge to them.

The “usual judicial practice” is to address an as-applied challenge before a facial challenge because it generally will be more “effieien[t],” because this sequencing decreases the odds that facial attacks will be addressed “unnecessarily” and because this approach avoids encouraging “gratuitous wholesale attacks upon state and federal laws.” Bd. of Trs. of the State Univ. of N.Y. v. Fox, 492 U.S. 469, 484-85, 109 S.Ct. 3028, 106 L.Ed.2d 388 (1989). We thus start by addressing the as-applied challenge. A.

In renewing its as-applied attack, Connection argues that the Act’s record-keeping and disclosure provisions, 18 U.S.C. § 2257(a)-(c), (e), suppress the free expression of Connection and any subscribers who wish to place sexually explicit personal advertisements in its magazines. We disagree — for many of the reasons that Connection I denied the company’s request for a preliminary injunction in 1998. Intermediate scrutiny applies.

In attempting to address the problem of underage pornography, Congress did not ban all images of sexually explicit conduct (on the theory that some images would involve minors) or ban all images of sexually explicit conduct that appeared to involve minors (on the same theory). Instead of suppressing these categories of expression, Congress chose to regulate the records of those creating and distributing sexually explicit images.

Still, as Connection correctly points out, § 2257’s record-keeping requirements do not entirely ignore the content of the producers’ images because the requirements apply only to materials that visually depict certain listed acts. Id. § 2257(a); see also id. § 2256(2)(A). But that reality does not make the law a presumptively invalid content-based regulation of speech so long as the requirements are “justified without reference to the content of the regulated speech.” Ward v. Rock Against Racism, 491 U.S. 781, 791, 109 S.Ct. 2746, 105 L.Ed.2d 661 (1989) (internal quotation marks omitted); see City of Renton v.

Playtime Theatres, Inc., 475 U.S. 41, 47-48, 106 S.Ct. 925, 89 L.Ed.2d 29 (1986); cf. City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 434-35, 440-41, 122 S.Ct. 1728, 152 L.Ed.2d 670 (2002) (plurality); id. at 448-49, 122 S.Ct. 1728 (Kennedy, J., concurring in the judgment). So long in other words as the law addresses the collateral or “secondary effects” of the expression, not the effect the expression itself will have on others, it will be treated as content neutral. See City of Renton, 475 U.S. at 47, 106 S.Ct. 925; Connection I, 154 F.3d at 291.

By contrast, statutes that single out speech for special treatment because of the effect its content will have on its audience amount to content-based restrictions subject to strict scrutiny. See United States v. Playboy Entm’t Group, Inc., 529 U.S. 803, 811-12, 120 S.Ct. 1878, 146 L.Ed.2d 865 (2000). Gauged by this benchmark, § 2257 is content neutral. Connection I, 154 F.3d at 290-91. As Connection concedes, Congress’s “unanimous concern” in enacting the provision was to deter the production and distribution of child pornography. Br. at 6.

Congress singled out these types of pornography for regulation not because of their effect on audiences but because doing so was the only way to ensure that its existing ban on child pornography could be meaningfully enforced. That objective not only is independent of the content of the regulated speech, but it also is a concern of the highest order, one that relates to a category of speech that the government may regulate, indeed completely suppress, based on its content. See Ferber, 458 U.S. at 756-66, 102 S.Ct. 3348; cf. R.A. V. v. City of St. Paul, 505 U.S. 377, 383-86, 112 S.Ct. 2538, 120 L.Ed.2d 305 (1992).

If Congress may suppress child pornography in its entirety due to its scarring impact on the children exploited in its production, surely it may facilitate the enforcement of laws devoted to that end by imposing a proof-of-age requirement on the producers and distributors of images of sexually explicit conduct — without triggering the most rigorous scrutiny known to constitutional law.

Because the proof-of-age requirement exists in spite of, not because of, the effect Connection’s speech will have on its audience and because the required records allow the enforcement of constitutionally sound laws aimed at eradicating the production of child pornography, § 2257 does not amount to a presumptively invalid content-based regulation of speech. See Ferber, 458 U.S. at 756-66, 102 S.Ct. 3348.

Nor does the law implicate the central risk of a content-based regulation of speech: that the government has imper-missibly interfered with the free exchange of ideas by imposing trade barriers on certain viewpoints but not on others. See Ward, 491 U.S. at 791, 109 S.Ct. 2746 (“The principal inquiry in determining content neutrality ... is whether the government has adopted a regulation of speech because of disagreement with the message it conveys.”); R.A.V., 505 U.S. at 387, 112 S.Ct. 2538; Clark v. Cmty. for Creative Non-Violence, 468 U.S. 288, 295, 104 S.Ct. 3065, 82 L.Ed.2d 221 (1984).

No doubt, § 2257 favors a particular viewpoint on this issue: Congress is against child pornography and is using this law to prevent it. Although that kind of viewpoint discrimination normally would be fatal to a law, that is not true here because the Constitution allows the government to embrace this viewpoint and to act on it by imposing a complete trade barrier on the production and trafficking of this kind of speech. See Ferber, 458 U.S. at 756-66, 102 S.Ct. 3348.

What we have, then, is a valid speech-related end — eliminating child pornography — followed by a means of achieving that end, a proof-of-age requirement that refers to the content of the speech (specifically defined images) not because of its effect on the audience but because it is the kind of speech that implicates the government’s ban on child pornography.

That kind of sensible reference to the content of speech — how else would the government impose a proof-of-age requirement designed to address child pornography? — does not rise to the level of a presumptively impermissible content-based regulation of speech. In addition to Connection I, two other federal courts have addressed the level of scrutiny applicable to a free-speech challenge to § 2257. Both share our conclusion that intermediate scrutiny applies. See ALA II, 33 F.3d at 85-87; Free Speech Coal. v. Gonzales, 406 F.Supp.2d 1196, 1205-06 (D.Colo.2005).

Section 2257 satisfies intermediate scrutiny. A law satisfies mid-level scrutiny if it advances a “substantial” government interest, if the measure does not “burden substantially more speech than is necessary” and if the measure leaves open “ample alternative channels for communication.” Ward, 491 U.S. at 791, 799-800, 109 S.Ct. 2746 (internal quotation marks omitted); see also Turner Broad. Sys., Inc. v. FCC, 512 U.S. 622, 662, 114 S.Ct. 2445, 129 L.Ed.2d 497 (1994).

No one disputes that the government’s interest in protecting children is “substantial.” And a universal age-verification requirement advances that interest in a reasonably tailored way for several reasons: It ensures that primary producers of pornography confirm that performers are of age before filming them; it permits secondary producers (who rarely will know the performers) to ensure that the individuals depicted in their publications are of age; it prevents children from attempting to pass themselves off as adults; and it creates a compliance system in which law-enforcement officers not only can identify the performers depicted in magazines and movies and verify their ages but also can eliminate subjective disputes with producers over whether a model’s apparent age should have triggered an age-verification check.

See ALA II, 33 F.3d at 86, 88-90; see also Connection I, 154 F.3d at 291-92; Final Report of the Attorney General’s Commission on Pornography 138-39. In objecting to this conclusion, Connection argues that the record-keeping requirements place undue barriers on the advertisers’ interests in engaging in anonymous speech. Yet Connection is not a particularly credible advocate for anonymous speech, as it does not permit advertisers to submit photos or other information without identifying who they are. Nothing in the statute, moreover, makes the required records available to the public. Cf.

Watchtower Bible & Tract Soc’y of N.Y., Inc. v. Vill. of Stratton, 536 U.S. 150, 166-67, 122 S.Ct. 2080, 153 L.Ed.2d 205 (2002). The advertisers must give the records only to Connection, to whom each advertiser already will have given material that most people would consider deeply personal — sexually explicit pictures of themselves with identifying names and addresses. Having entrusted Connection with these pictures and having already been required by Connection to disclose their identities, what individuals would shy away from producing verification of their ages as well?

None, it seems to us, or at least none who would have a reasonable basis for doing so. But, Connection adds, § 2257 also makes the information available to the government upon request. To the extent the advertisers are concerned that the law gives the government access to their names, addresses and other identifying information, they have no more to complain about than every taxpayer in the country.

To the extent their concern is that the government somehow plans to use this information for a purpose for which it was not intended, say, to target swingers for mistreatment, they offer nothing more than two unverified anecdotes to support the point, and both anecdotes have nothing to do with the improper use of these records by government agents. That leaves one other risk: that § 2257 undermines a central benefit of Connection’s magazines — its creation of a forum for individuals to present sexually explicit pictures of themselves to the world without disclosing who they are.

Under § 2257, it is true, the government has access to these images and the identities of the individuals for the limited purpose of ensuring they are of age. But under Connection’s business model, so potentially does the rest of the world. The essence of the company’s service is to facilitate non-platonic connections between anonymous advertisers and anonymous subscribers, something it does by giving advertisers the opportunity to respond to inquiries they have solicited and by giving advertisers the opportunity to lift the veil of anonymity, to say nothing of other veils, to these unknown inquirers.

It may be that there are advertisers in Connection’s magazines who have greater privacy concerns about revealing their identities to law-enforcement officers for the limited purpose of confirming their age than about revealing their identities to unknown inquirers for the purpose of facilitating a liaison. But the question is whether such individuals would have a cog-nizably reasonable basis for suppressing their communications in this setting, and that is something Connection’s affidavits and evidence do not support.

Although Connection points out that its paper subscriptions have decreased since the passage of the Act in 1988, it fails to account for the possibility, acknowledged by its own expert, that the rise of Connection’s internet service, internet-based chat rooms and the like explain the decline. According to Connection’s editor, the company’s own internet-based service went from non-existent in 1995 to doing “fairly well” by 2000. JA590.

Connection also contends that a universal age-verification requirement is over-inclusive because it requires Connection to maintain records of individuals who are “in their 30s, 40s, 50s and beyond.” Br. at 34. A rule requiring photo identification only where the depicted individual appears to be under some threshold age, it argues, would accomplish Congress’s goal without burdening free expression that has little likelihood of implicating the underlying concerns of the statute.

But in enacting a content-neutral proof-of-age requirement, Congress need not employ “the least speech-restrictive means of advancing the Government’s interests” but must show only that the government’s “interest ... would be achieved less effectively absent the regulation” and that the measure “do[es] not burden substantially more speech than is necessary.” Turner, 512 U.S. at 662, 114 S.Ct. 2445 (internal quotation marks omitted). No doubt requiring identification only where the individuals appear to be below a threshold age “would lead to accurate determinations in many cases.” Connection I, 154 F.3d at 292.

But it could not do so without injecting “an ineffectual subjectivity” into the proof-of-age requirement and without effectively delegating enforcement of this critical issue to the industry being regulated- — -two of the problems Congress permissibly sought to correct. Id.; ALA II, 33 F.3d at 90 (“The entire point of the Act is to prevent subjective determinations of age by implementing a uniform procedure that applies to all performers.”). The record in this case illustrates the problem.

Although Connection maintains that “a simple look at the photos” in its magazines “makes ... clear” that the persons pictured are obviously “not minors,” Br. at 34, the record proves otherwise.

A brief glance at one of the issues of the magazine included in the record reveals many images (particularly the frequent depiction of mere body parts) from which no lay observer could readily discern the individuals’ ages, see, e.g., JA 1020, 1022-23, 1025-30, 1034, 1036-38, 1041, 1045, 1052, 1060, 1068-69, 1075-76, 1085, 1089-90, 1092, 1098-1105, as well as a number of images that appear (and in some cases purport) to portray youthful individuals, see, e.g., JA 1027,1032, 1041, 1052, 1060, 1068-89, 1075, 1089-90.

Connection counters that these images tell just part of the story: Many of the pictures it receives from would-be advertisers, when combined with other information submitted, including the advertisement text and payment details, provide enough information for Connection’s editors to estimate the advertisers’ ages, and the pictures are simply cropped before they appear in one of Connection’s magazines. Connection’s editor asserts in her affidavit that she has encountered only two cases where the age of the individual depicted was in doubt.

Even if we accept all of this as true, it does nothing to diminish the subjectivity inherent in ad hoc, appearance-based judgments based on amateur photographs. Nor does it address the risks of delegating enforcement of this critical issue to the subjects of the regulation or of human error in evaluating submitted advertisements. The editors who screen submitted advertisements, notably, have no particular training in age identification, and the record reflects some instances where the editors approved advertisements that expressly describe persons below Connection’s alleged self-imposed age cutoff of 21.

See, e.g., JA 380-82, 415-16, 422. Most importantly, however, these arguments do nothing to solve the problem of applying an appearance-based standard to advertisements that contain only isolated body parts. So long as Connection continues to be an advertising forum in which individuals may submit advertising text along with photographs of body parts, it offers no reasonable basis for ensuring that even the most earnest and law-abiding peddler of pornography can verify that its images depict adults rather than minors. The testimony of the government’s expert, Dr. Francis Biro, does not overcome this problem.

True, Dr. Biro acknowledged that the “vast majority” of the photographs he examined showed individuals over 21 and “many” in their thirties, forties and fifties. JA 479. True also, it appears that Dr. Biro attempted to estimate the ages of some models based on images of body parts alone, concluding that some models were under 21 but most were not.

Yet he stressed that “there’s no way of establishing an exact age by looking at an individual” depicted in such images, JA 487, and he noted that determining whether a pictured model is above or below the age of majority is not an “exact science,” JA 475, particularly when the picture does not capture enough relevant details. Whatever Dr. Biro can or cannot do in this respect, moreover, matters little to the resolution of this case.

Even if it is true that, at least up to certain stages in a person’s development, an expert sometimes may be able to distinguish between individuals of different ages based on an examination of mere body parts, that does not help Connection. It offers no evidence that its own staff has the capacity to make the same nuanced determinations, and its editor admitted they had no training in the subject at all. The statute also leaves “ample alternative channels” of communication for Connection’s advertisers.

Although § 2257 closes one narrow door for Connection advertisers — by prohibiting them from submitting sexually explicit images to its magazines without providing photo identification — it leaves open other doors of communication. Keep in mind that, even before Congress passed the Act, Connection itself prevented its advertisers from making anonymous submissions. It required advertisers then, and it does not mind requiring advertisers now, to provide their names and addresses in connection with all submissions, and it has always said that only adults may make submissions.

By requiring Connection to demand a verifiable identification with these submissions, the Act merely ensures that the advertisers are who they say they are — in terms of name, address and age. In this sense, “the Act, by its terms, bans no form of expression.” ALA II, 33 F.3d at 88; see Connection I, 154 F.3d at 294.

It simply adds a requirement — a photo identification — that is consistent with Connection’s existing identification requirements and that therefore is unlikely to affect many adult advertisers who otherwise wish to share their photographs with Connection and who otherwise are willing to comply with Connection’s existing identification and age requirements. Similar record-keeping requirements, indeed, are routinely required to assist the enforcement of tax, employment and immigration laws. See ALA II, 33 F.3d at 91.

Once these requirements are satisfied, individual advertisers may publish their pictures anonymously to their hearts’ content. Even then, moreover, another alternative remains: Connection’s internet service provides an independent channel of communication, and the government does not argue that these parts of the Act and regulations reach this service. Recent Supreme Court cases do not undermine the holding of Connection I.

Connection insists that much has changed since our court first addressed these issues — that four recent Supreme Court decisions support its position, and that changes to the statute (and the implementing regulations) have impermissibly expanded the statute’s reach. We disagree. In Free Speech Coalition, the Court invalidated statutory provisions that criminalized the possession of any image that “appears to be[] of a minor engaging in sexually explicit conduct,” 18 U.S.C. § 2256(8)(B) (2000); see 535 U.S. at 258, 122 S.Ct. 1389.

Yet these provisions could not be justified as remedying harms that flow from the production of child pornography because the images at issue were “produced without using any real children.” 535 U.S. at 239, 122 S.Ct. 1389. The government instead defended the law on the ground that the speech it singled out was unprotected by the First Amendment precisely because of its content. Id. at 249, 257, 122 S.Ct. 1389.

Its asserted rationales — that pedophiles might use such “virtual child pornography” to entice children to participate or might “whet their own sexual appetites with the pornographic images” — each targeted a “harm [that] flows from the content of the images, not from the means of their production.” Id. at 241-42, 122 S.Ct. 1389 (internal quotation marks omitted). In invalidating the provisions as facially overbroad because they impermissibly abridged a “substantial” amount of protected speech, id. at 256, 258, 122 S.Ct. 1389, the Court did not expressly say it was applying strict scrutiny to the law.

But what it did can only be described as employing the tools of this most skeptical level of review, as it invalidated the provisions because they did not hew closely enough to any of the government’s asserted interests in enacting the law. See id. at 251-58, 122 S.Ct.

Section 2257, by contrast, is content neutral and subject only to intermediate scrutiny. Free Speech Coalition did not answer (because it did not confront) the question raised by a content-neutral record-keeping requirement. The proper analogy to this case thus is not a law that criminalizes the possession of images that appear to be child pornography, but a law that requires the producers of apparent child pornography to keep their production records to allow law-enforcement officers to ensure that actual child pornography was not involved.

Nothing in Free Speech Coalition suggests that such a law would be invalid. At issue in Watchtower was an ordinance that required door-to-door canvassers and pamphleteers to register with the government in advance of their activities. See 536 U.S. at 154-58, 122 S.Ct. 2080. In concluding that the law failed intermediate scrutiny, the Court reasoned that it affected a broad spectrum of speech, hindered an historically significant mode of communication and destroyed anonymous and spontaneous advocacy by making the registration records open to the public at large. See id. at 162, 165-69, 122 S.Ct.

Section 2257, however, does none of these things: It affects only a narrow category of speech and does so for the limited purpose of preventing speech (child pornography) that the First Amendment does not protect; it does not condition speech on announcing to the public at large what the speaker plans to say or why he plans to say it; and it does so in a setting in which the publisher of these advertisements already required the individuals to disclose their identities, see ALA II, 33 F.3d at 94.

Alameda Books and Playboy Entertainment support the application of intermediate scrutiny to this case and support the distinction that Connection I drew (and we draw today) between restrictions that are justified without reference to the content of the regulated speech and those that are not. Compare Alameda Books, 535 U.S. at 440-41, 122 S.Ct. 1728 (plurality), and id. at 448-49, 122 S.Ct. 1728 (Kennedy, J., concurring in the judgment), with Playboy Entm't 529 U.S. at 811-12, 120 S.Ct. 1878.

Alameda Books also supports Connection I for a separate reason: It emphasizes the deference that courts owe to legislative judgments about the collateral effects of speech. See 535 U.S. at 437-40, 122 S.Ct. 1728 (plurality); id. at 449, 122 S.Ct. 1728 (Kennedy, J., concurring in the judgment). Nor do any of the changes to § 2257 or its implementing regulations enhance Connection’s as-applied challenge. Although Connection does not directly challenge any of the 2006 amendments to the law, it does challenge (or at least rely upon) two of the 2003 amendments.

One of those amendments enlarged the list of offenses for which the government may use the records required by § 2257 as evidence, most notably by allowing the records to be used in prosecuting child-pornography, sexual-exploitation-of-children and obscenity offenses. See Pub.L. No. 108-21, § 511(a)(1), 117 Stat. at 684 (amending 18 U.S.C. § 2257(d)(2)). While this amendment to the law supplies the basis for plaintiffs’ Fifth Amendment challenge (more on that below), it has no real bearing on Connection’s First Amendment claim.

If “criminal penalties for obscenity offenses are consistent with the First Amendment,” even though they may deter some amount of non-obscene expression, Alexander v. United States, 509 U.S. 544, 557, 113 S.Ct. 2766, 125 L.Ed.2d 441 (1993), individuals may not prevent the government from using records of actually obscene expression in an otherwise-lawful obscenity prosecution. The other statutory change made by the 2003 amendments extends § 2257 to internet and other computer-based images. See Pub.L. No. 108-21, § 511(a)(2), 117 Stat. at 685 (amending 18 U.S.C. § 2257(h)(3), recodified at id. § 2257(a)).

But that alteration, too, makes no difference here. Connection itself maintains that this change to § 2257 is “not self-executing” and is effective only to the extent implemented by the revised regulations. Reply Br. at 10 n. 2. And as it acknowledges, the district court held that Connection’s online activities are exempt under the regulations themselves, a conclusion the government has declined to challenge on appeal, see Br. at 19 n. 6. In dissent, Judge Moore maintains that strict scrutiny should govern this as-applied challenge.

Yet at no point in its panel brief or in its supplemental en banc brief did Connection urge us to apply strict scrutiny to this case, and, with respect, the reasons given in Connection I by a panel of this court, in ALA II by the D.C. Circuit and in today’s opinion justify continuing to apply mid-level scrutiny to this dispute.

Judge Moore also maintains that, even if intermediate scrutiny applies, § 2257 should be invalidated, and in doing so she makes a convincing case why the law would have difficulty withstanding an as-applied attack by a mature-adults-only magazine that included photographs only of readily identifiable mature adults. But, with respect, that is not this case, and it is not Connection’s publications.

By allowing photographs of individuals who appear to be, and in some cases purport to be, youthful and by allowing photographs of body parts alone, Connection simply is not a standard-bearer for the mature-adults-only publication. It thus cannot be the beneficiary of the First Amendment difficulties such a claim would present. B. That brings us to Connection’s facial challenge to the Act. A facial challenge to a law is no small matter. At stake is not an attempt to invalidate the law in a discrete setting but an effort “to leave nothing standing,” Warshak v.

United States, 532 F.3d 521, 528 (6th Cir.2008) (en banc), to invalidate the law in each of its applications, to take the law off the books completely. That, to be sure, is the fate some laws deserve — either because the defect in the law infects all or virtually all of its applications (say, a race-based classification or a law serving an unconstitutional purpose) or because the constitutional problems cannot meaningfully be severed. See, e.g., Edwards v. Aguillard, 482 U.S. 578, 585-594, 107 S.Ct. 2573, 96 L.Ed.2d 510 (1987); City of Houston v.

Hill, 482 U.S. 451, 468-69, 107 S.Ct. 2502, 96 L.Ed.2d 398 (1987); see also Michael C. Dorf, Facial Challenges to State and Federal Statutes, 46 Stan. L.Rev. 235, 279-82 (1994). But before the courts will announce such a judgment, they generally insist that the claimant show one of two things: (1) that there truly are “no” or at least few “circumstances” in “which the Act would be valid,” United States v. Salerno, 481 U.S. 739, 745, 107 S.Ct. 2095, 95 L.Ed.2d 697 (1987); see also Wash. State Grange v. Wash.

State Republican Party, — . u.S. -, 128 S.Ct. 1184, 1190, 170 L.Ed.2d 151 (2008); or (2) that a court cannot sever the unconstitutional textual provisions of the law or enjoin its unconstitutional applications. To do otherwise would amount to a judicial trespass — a court’s striking of a law in all of its applications even though the legislature has the prerogative and presumed objective to regulate some of them. This rule normally would make short work of the plaintiffs’ facial challenge. Our court’s rejection of the as-applied challenges to § 2257 in Connection I and today, to say nothing of the D.C.

Circuit’s rejection of a more far-reaching challenge to the law in ALA II, demonstrate that the law has numerous constitutional applications — a conclusion that normally would end the matter. In conventional constitutional litigation, it rarely (if ever) will be the ease that a court, having upheld the constitutionality of a law in the context of the as-applied challenge before it, will proceed to strike the law in all of its applications based on hypothetical applications of the law to hypothetical individuals not before the court.

But the courts rightly lighten this load in the context of free-speech challenges to the facial validity of a law. Although “[l]itigation by hypothetical” generally is frowned upon, if not barred, in other areas of constitutional litigation, see Warshak, 532 F.3d at 529, it is sometimes required in free-speech cases. See Broadrick v. Oklahoma, 413 U.S. 601, 612-13, 93 S.Ct. 2908, 37 L.Ed.2d 830 (1973); Los Angeles Police Dep’t v. United Reporting Publ’g Corp., 528 U.S. 32, 38-39, 120 S.Ct. 483, 145 L.Ed.2d 451 (1999); cf. Fox, 492 U.S. at 483-84, 109 S.Ct. 3028.

Here, for example, even though § 2257 may be applied constitutionally to Connection and the individual plaintiffs, the whole point of a facial challenge, or what the courts in the First Amendment context have come to call an overbreadth challenge, is to permit the claimant to strike the law in its entirety based on its application to other individuals not before the court.

The overbreadth doctrine thus changes the customary rules of constitutional litigation: It relaxes the general prohibition against vicarious litigation by allowing claimants to assert the rights of third parties, and it permits a court to strike a law in its entirety even though it legitimately may be enforced in some other settings. Broadrick, 413 U.S. at 612-13, 93 S.Ct. 2908.

Due to the risk that “enforcement of an overbroad law” may “deter[ ] people from engaging in constitutionally protected speech” and may “inhibit[] the free exchange of ideas,” the courts will strike a law on its face “if it prohibits a substantial amount of protected speech” both “in an absolute sense” and “relative to the statute’s plainly legitimate sweep.” United States v. Williams, — U.S. -, 128 S.Ct. 1830, 1838, 170 L.Ed.2d 650 (2008); see also Broadrick, 413 U.S. at 615, 93 S.Ct. 2908.

Even in free-speech cases, however, facial invalidation of a statute remains “strong medicine that is not to be casually employed.” Williams, 128 S.Ct. at 1838 (internal quotation marks omitted). And even in this setting, facial challenges remain “disfavored” because they frequently require courts to “anticipate a question of constitutional law in advance of the necessity of deciding it” or to “formulate a rule of constitutional law broader than is required by the precise facts to which it is to be applied.” Wash.

State Grange, 128 S.Ct. at 1191 (internal quotation marks omitted); see also id. (“[Fia-da! challenges threaten to short circuit the democratic process by preventing laws embodying the will of the people from being implemented in a manner consistent with the Constitution.”). The Supreme Court therefore has “vigorously enforced the requirement that a statute’s overbreadth be substantial ... relative to the statute’s plainly legitimate sweep,” Williams, 128 S.Ct. at 1838, and has placed “the burden of demonstrating ... substantial over-breadth” on the claimant, Virginia v.

Hicks, 539 U.S. 113, 122, 123 S.Ct. 2191, 156 L.Ed.2d 148 (2003); see also N.Y. State Club Ass’n v. City of New York, 487 U.S. 1, 14, 108 S.Ct. 2225, 101 L.Ed.2d 1 (1988) (“To succeed in its [facial-over-breadth] challenge, [the plaintiff] must demonstrate from the text of [the statute] and from actual fact that a substantial number of instances exist in which the [l]aw cannot be applied constitutionally.”). In attempting to strike § 2257 in its entirety on overbreadth grounds, Connection argues that the law would be unconstitutional as applied to a magazine that depicted only “mature adult models,” Supp.

Br. at 3, who “are clearly and visibly not minors,” Br. at 44. That may well be true, particularly if the magazine not only confined itself to self-evidently mature models but also did not permit the depiction of isolated body parts. The D.C. Circuit reached a similar conclusion, “agreeing] with [plaintiffs’] suggestions that certain applications of the record-keeping requirements may well exceed constitutional bounds,” noting that “an illustrated sex manual for the elderly” would be “an obvious example.” ALA II, 33 F.3d at 90. Even so, this argument does not supply a basis for invalidating § 2257.

Connection has not pointed us to any such magazine or book and has not introduced any evidence showing that this third-party situation even exists. That alone is reason enough to give us pause. But even if we accepted Connection’s submission, even if we assumed in other words that such magazines and books exist and that § 2257 could not validly be applied to them, that would not satisfy the company’s burden for dispensing the “strong medicine” of over-breadth. At this point in the case, there is little basis for dispute that § 2257 complies with the First Amendment in most settings.

As we have shown, it is constitutional as applied to Connection and the individual plaintiffs, and Connection does not dispute, and indeed all but concedes, that the law would be constitutional in most other settings. In its panel brief, Connection notes that “the congressional record arguably supports legislation” that targets “sexually explicit depictions of young looking persons,” Br. at 33, and it has suggested that the law would be constitutional if it applied only to people who appear to be under 26 or under 30, Connection Distrib. v.

Keisler (Connection III), 505 F.3d 545, 571-72 & n. 6 (6th Cir.2007) (Moore, J., concurring). Nor does Connection dispute that this “legitimate sweep” of the law represents the vast majority of its applications. The report of the Attorney General’s Commission on Pornography notes that “[pjerhaps the single most common feature of models is their relative, and in the vast majority of cases, absolute youth.” Final Report of the Attorney General’s Commission on Pornography 229.

At a Senate Judiciary Committee hearing on § 2257 and related legislation, an administrator from the Department of Justice National Obscenity Enforcement Unit testified that “[o]ne who is 25 to 30 years of age is virtually never seen in pornographic videos or magazines.” JA 120. Nothing in the record contradicts these statements or the general notion that, when people buy or share pornography, they typically do so with respect to publications or movies involving the young.

Far from contradicting these aspects of the legislative record, Connection elaborates on them, explaining that “the crux of the problem that Congress sought to address was rooted in the fact that commercial producers of sexually explicit films used youthful looking actors and actresses as young as eighteen years old and nearly always younger than twenty five years old.” Br. at 8 (emphasis added).

A central theme in Connection’s fourteen-year attack on this law, indeed, is that its magazine and subscribers are unconventional— that they are “unorthodox” and “controversial” conveyors and purveyors of pornography, that swingers generally are middle-aged individuals and that their advertisements accordingly represent a “minority” of the models and performers generally featured in the materials produced by the pornography industry. Supp. Br. at 7-8, 24; Br. at 16-17.

On this record and in the face of these concessions, we have no basis for reaching any conclusion other than this: § 2257 most conspicuously applies to publications involving youthful-looking models and performers, which is the setting in which it is easiest to accept the constitutionality of these proof-of-age requirements and which at any rate is the setting in which the plaintiffs do not challenge the law’s validity. Connection at most has identified a discrete application of the statute that may be problematic.

Yet the question is not whether the claimant can imagine some “overbreadth”; it is whether the claimant can show “substantial overbreadth.” At the panel stage of this case, the judges on their own initiative raised a second overbreadth problem, one not raised in Connection’s amended complaint, in its briefs before the district court or in its briefs before the panel.

By its terms, the panel observed, the statute seems to apply to a couple who produced, but never distributed, a home video or photograph of themselves engaging in sexually explicit conduct, because the record-keeping requirements apply to anyone who produces sexually explicit images, see 18 U.S.C. § 2257(a)-(b), regardless of whether the images are sold, traded, or otherwise distributed, see id. § 2257(h)(2)(A)(i)-(iii); Connection III, 505 F.3d at 552.

Invoking the constitutional-avoidance doctrine and the rule of lenity, the government points to language in the statute suggesting that it does not cover this situation. Supp. Br. at 20 (noting the statute’s references to a producer’s “business premises,” to the involvement of multiple performers and to contractual or similar relationships between producers and performers). The avoidance doctrine no doubt often goes a long way in defusing potential conflicts between Congress’s enactments and the Constitution.

Whether the doctrine allows us to read the statutory text in the way the government proposes, however, is far from clear. Viewed in isolation, the Act’s reference to “business premises” might suggest that the statute draws a line between commercial and noncommercial pornography. Yet subsequent amendments to the law, apparently in response to a Tenth Circuit decision addressing a related point, Sundance Assocs., Inc. v. Reno, 139 F.3d 804 (10th Cir.1998), make it clear that the law covers commercial and non-commercial pornography. See 18 U.S.C. § 2257(h)(2)(A), (h)(2)(B)(iii).

Peddlers of child pornography, it turns out, are as apt to do so for non-pecuniary purposes as for other motives, which is why Congress extended the law to this setting. Final Report of the Attorney General’s Commission on Pornography 134-35. Once we eliminate the possibility of a commercial/non-commercial line of coverage, that leaves at least two other possibilities that would exclude coverage in this setting. One is that the statute applies only to pornography created for sale or trade, an

interpretation that would not apply to pornography created by an adult couple for home consumption. In his appellate briefs in this case and in the

preamble to a recently promulgated rule amending the regulations implementing § 2257, the Attorney General has embraced this view, construing the statute as “limited to pornography intended for sale or trade,” 73 Fed.Reg. at 77,456. But in view of the terms of the statute, see 18 U.S.C. § 2257(a), (b), (h)(2)(A), (h)(2)(B)(iii), the existence of a textual hook for this

interpretation is open to question. The other possibility is that, even if the law applied to such a couple, it would rarely matter because most (if not all) of the identifying and record-keeping information required by the Act necessarily would lie within the four corners of the couple’s home, which is where the law requires it to be kept. See Id. § 2257(c). But in view of the labeling requirements of the implementing regulations, 28 C.F.R. § 75.6(b), it remains unclear whether such a couple would satisfy these requirements without knowing they were doing so.

Given these complexities and given the absence thus far of any such application of the statute, we see no need to resolve the point one way or another, and thus we do not take a stand on the issue. Let us instead assume for the sake of argument that the panel was right — that the law’s record-keeping and disclosure requirements would apply to sexually explicit images produced by such a couple. And let us assume for the sake of argument that the panel was right in concluding that this application of the law would be unconstitutional.

Does it follow that the panel was also correct in holding that this as-applied defect requires the invalidation of § 2257 in its entirety? Not in our view. First, we have no record, and therefore no context, for assessing the substantiality of this overbreadth problem. Because the plaintiffs did not raise this theory of unconstitutionality in their complaint or in the district court, the record is utterly barren about whether some, many, indeed any, American couples are affected by this proposed application of the statute — and, if so, in what ways.

That contextual vacuum by itself counsels in favor of choosing discretion over valor in dealing with this over-breadth challenge. See N.Y.

State Club Ass’n, 487 U.S. at 14, 108 S.Ct. 2225 (rejecting First Amendment overbreadth challenge to local antidiscrimination law, which the plaintiffs asserted could not constitutionally apply to purely private clubs, where “[n]o record was made” concerning those clubs, the Court was “not informed of the characteristics of any particular clubs” and thus it “[could] not conclude that the [l]aw threatens to undermine the associational or expressive purposes of any club, let alone a substantial number of them”). But that is just half of the problem.

The record not only presents a contextual vacuum; it also presents a law-enforcement vacuum, making this debate all the more abstract and all the more vulnerable to inaccurate rather than accurate judicial decision-making. The government has informed us that, during the twenty years that § 2257 has been in existence, it has never been enforced in this setting.

It has informed us that it has no intention of enforcing the law in this setting — as proved by the fact that the Attorney General, a party to this ease and the sole defendant in it, has taken the position that the statute “does not apply to images that an adult couple produces of its own intimate activity for the couple’s private enjoyment at home.” Supp. Br. at 20. And it has informed us that, in connection with the promulgation of a final rule amending the regulations implementing § 2257, the Attorney General has stated in the

preamble to the new regulations that “[t]he statute ... is limited to pornography intended for sale or trade.” 73 Fed.Reg. at 77,456. Connection offers no evidence of a contrary enforcement record, and we are not aware of any case law, regulations, even news reports, mentioning the possibility of a different application of the statute — at least until the panel raised the idea during the third appeal of this case. Sure enough, there may be a first time for everything.

And we do not mean to suggest that a couple potentially affected by this hypothetical application of the law could not bring a declaratory-judgment action or an as-applied challenge to the law today, whether in their own names or as an anonymous John and Jane Doe.

But that does not mean litigation by proxy makes sense in this setting, one that has yet to come to pass, one that may never come to pass and one that presents three layers of abstraction: (1) no record of any kind about this form of middle-aged sexual expression; (2) no record of its prevalence; and (3) no idea how a government that for twenty years has not applied the law in this setting, that indeed disclaims the authority to apply the law in this setting, ultimately would choose to apply the law if it ever changed its mind.

Overbreadth plays several essential roles in protecting free speech, but this simply is not one of them. Second, this hypothetical application of the statute, even when it is considered on this thin record, makes no difference to the outcome of this case. Even if we assume that the statute covers such a video and even if we assume that the First Amendment would not permit Congress to impose its record-keeping requirements in this unusual setting, that would not materially advance Connection’s efforts to show substantial overbreadth.

What makes this hypothetical seemingly helpful to Connection — the extension of the statute to a setting that is far removed from the underlying purposes of the Act, that makes little sense and that raises constitutional red flags — undermines much of its significance. Connection offers no argument, much less proof, that there are a meaningful number of individuals who would be adversely affected by this construction of the law.

Which takes us back to the central point: The question in the context of a facial challenge is not whether a court can conceive of one or more unconstitutional applications of a statute; it is whether the alleged unconstitutional sweep of the statute is “substantial ... relative to the statute’s plainly legitimate sweep.” Williams, 128 S.Ct. at 1838. No such showing has been made here.

The record and the plaintiffs’ concessions establish that the overwhelming majority of applications of § 2257 do not offend the free-speech guarantees of the Constitution, and a “vigorous” enforcement of the “substantial overbreadth” requirement prohibits a party from leveraging a few alleged unconstitutional applications of the statute into a ruling invalidating the law in all of its applications. Because the burden rests with Connection to establish substantial overbreadth, see Hicks, 539 U.S. at 122, 123 S.Ct. 2191; N.Y.

State Club Ass’n, 487 U.S. at 14, 108 S.Ct. 2225, and because Connection has produced no evidence on this score, it simply has not supplied a basis for knocking § 2257 out in its entirety- The concept of “substantial over-breadth,” we acknowledge, has some elusive qualities, and it likely is the key source of our disagreements in this case.

A first run at applying the Supreme Court’s cases in this area might suggest a more concrete approach than we have offered — of placing, say, the number of overall applications of the statute in the denominator and the number of unconstitutional applications of the statute in the numerator. With this ratio in hand, we could identify a certain threshold of unconstitutional applications — of, say 10%, 25%, 50% or more — and label that threshold as the turning point for a finding of substantial overbreadth. But the Supreme Court has never gone down this road — and with good reason.

Substantial overbreadth involves not just an inquiry into the legitimate and illegitimate sweep of a statute; it also involves an inquiry into the “absolute” nature of a law’s suppression of speech. Together, these questions require as much in the way of judgment as they do a comparison between the constitutional and unconstitutional applications of a law. Ultimately, the critical question is this: Under what circumstances is it appropriate to invalidate a law in all of its applications when its invalidity can be shown (or assumed) in just some of its applications?

When we think about the problem that way, it is hard to understand who is being hurt by resisting the plaintiffs’ call to invalidate the statute on its face. The middle-aged couple is not likely to be chilled by the statute. Over twenty years and numerous administrations, the statute has never been enforced in this setting, and the Attorney General has publicly taken the position that he will not enforce the statute in this setting.

But even if this track record does not suffice to give the hypothetical couple peace of mind, they have a remedy — a John and Jane Doe as-applied challenge to the law, together with attorney fees if they win. What, then, of the hypothetical pornography magazine or sex manual that involves only the middle-aged and the elderly? There, too, we have not been told of any enforcement efforts in this area, and ALA II and today’s case offer ample indications that such an application of the law would run into serious First Amendment problems.

On the other side of the equation, we are being asked to invalidate a law in its entirety based on a worst-case scenario that, to our knowledge, has never occurred, that may never come to pass and that has not been shown to involve a materially significant number of people. How strange, moreover, that we would impose such a remedy after the government has withstood every as-applied First Amendment challenge to the law by the real people and businesses to whom it most naturally has been applied over the last twenty years.

Judge Kennedy’s dissent notes, correctly, that there are costs to requiring case-by-case adjudication: Some individuals who are wrongfully chilled from speaking may decline to seek redress because litigation is time-consuming and, if they lose, it can be expensive. See Hicks, 539 U.S. at 119, 123 S.Ct. 2191. But there are serious costs to the alternative, too. Because courts are ill-equipped to “resolve questions of constitutionality” in “every conceivable situation which might possibly arise,” Gonzales v.

Carhart, 550 U.S. 124, 127 S.Ct. 1610, 1639, 167 L.Ed.2d 480 (2007), we risk deciding wrongly if we answer abstract questions without a proper factual record and with only our judicial imagination to guide us. See United States v. Raines, 362 U.S. 17, 22, 80 S.Ct. 519, 4 L.Ed.2d 524 (1960).

And sustaining even a meritorious facial challenge to “a law that in some of its applications is perfectly constitutional ... has obvious harmful effects” because it throws out the bad with the good, including what we and all of the parties agree is a perfectly legitimate effort to prevent child pornography when applied to publications and films involving youthful-looking models. Williams, 128 S.Ct. at 1838. These costs help to explain why “[a]s-applied challenges are the basic building blocks of constitutional adjudication.” Carhart, 127 S.Ct. at 1639 (quotation marks omitted) (alteration in original).

So long as overbreadth remains “strong medicine that is not to be casually employed,” Williams, 128 S.Ct. at 1838 (internal quotation marks omitted), and remains a measure only of “last resort,” Broadrick, 413 U.S. at 613, 93 S.Ct. 2908, it has no application here. Judge Kennedy’s dissent notes, correctly again, that the absence of a prior application of the law to private couples who create and keep sexually explicit images in their homes does not by itself doom this facial-overbreadth challenge.

A litigant interested in bringing an over-breadth challenge to a law need not await its application to every conceivable fact pattern before filing suit. But that does not mean the government’s track record in this case — of never applying the law in this setting over twenty years and of disclaiming any authority and intention of doing so — has no role to play in the exercise of our judgment about whether to strike this law in its entirety.

That enforcement vacuum together with the absence of any record support for the plaintiffs’ position contribute to the utterly abstract nature of this debate, surely something we may consider in deciding whether to grant over-breadth relief.

And in exercising that judgment, the Supreme Court tells us to consider whether the alleged overbreadth is “substantial” and “real,” Broadrick, 413 U.S. at 615, 93 S.Ct. 2908, the whole point being to determine whether “there [is] a realistic danger that the statute itself will significantly compromise” the First Amendment rights of the parties not before us, such as the hypothesized private couples, Members of the City Council v. Taxpayers for Vincent, 466 U.S. 789, 801, 104 S.Ct. 2118, 80 L.Ed.2d 772 (1984).

Surely one factor to consider in assessing the “realistic danger” of inhibiting speech, but hardly the only factor to consider, is the past practices and future prospects of enforcement. See, e.g., Regan v. Time, Inc., 468 U.S. 641, 651-52 & n. 8, 104 S.Ct. 3262, 82 L.Ed.2d 487 (1984) (plurality opinion); Faustin v. City & County of Denver, 423 F.3d 1192, 1201 (10th Cir.2005); West v. Derby Unified Sch. Dist. No. 260, 206 F.3d 1358, 1368 (10th Cir.2000). One other point deserves mention.

At the panel stage, the court reasoned that, once it had identified one unconstitutional application of the law (here to the hypothetical couple), it must invalidate the entire statute unless it could sever an offending portion of the text from the rest of the statute. That is one option when a constitutional problem has been identified, but it is not the only option. A court may enjoin the unconstitutional applications of the law while preserving the other valid applications of the law. See Brockett v. Spokane Arcades, Inc., 472 U.S. 491, 504-05, 105 S.Ct. 2794, 86 L.Ed.2d 394 (1985); United States v.

Grace, 461 U.S. 171, 180-83, 103 S.Ct. 1702, 75 L.Ed.2d 736 (1983); see also Ayotte v. Planned Parenthood of N. New Eng., 546 U.S. 320, 328-29, 332, 126 S.Ct. 961, 163 L.Ed.2d 812 (2006); cf. Gillian E. Metzger, Facial Challenges and Federalism, 105 Colum. L.Rev. 873, 884-87 (2005). Or a court may assume for the sake of argument that certain applications of the law would be unconstitutional but still reject a facial challenge. See Hicks, 539 U.S. at 122-24, 123 S.Ct. 2191; N.Y.

State Club Ass’n, 487 U.S. at 14-15, 108 S.Ct. 2225; Ferber, 458 U.S. at 773-74, 102 S.Ct. 3348; Broadrick, 413 U.S. at 617-18, 93 S.Ct. 2908; cf. Carhart, 127 S.Ct. at 1639. That last path is the one we have followed here. III. In their amended complaint, the three individual plaintiffs challenge the validity of § 2257 under the Fifth Amendment’s Self-Incrimination Clause. As amended in 2003, the statute allows the government to use the records Connection must maintain as evidence not only of violations of § 2257 but also as evidence to prove violations of other obscenity and pornography-related laws.

See 18 U.S.C. § 2257(d)(2). Because these records could implicate them in crimes, plaintiffs argue that this provision violates their privilege against self-incrimination.

The district court rejected this argument on the merits, concluding that two of the three factors we consider in evaluating a self-incrimination challenge to record-keeping requirements cut against the plaintiffs: The statute’s primary purpose is regulatory — as it encompasses largely lawful activity, not a “highly selective and inherently suspect group of people” — and “merely fulfilling the records requirements does not demonstrate involvement in criminal activity.” JA 60-62 (internal quotation marks omitted); cf. United States v. Alkhafaji, 754 F.2d 641, 643 (6th Cir.1985).

We need not resolve the merits of the plaintiffs’ self-incrimination claim, however, because it is not ripe. The ripeness doctrine, we have recently explained, “is drawn both from

Article III limitations on judicial power and from prudential reasons for refusing to exercise jurisdiction” and “serves to avoid[ ] ... premature adjudication of legal questions,” thus “preventing] courts from entangling themselves in abstract debates that may turn out differently in different settings.” Warshak, 532 F.3d at 525 (internal quotation marks omitted) (omission and first alteration in original). To meet their burden of showing their claim is ripe for review, see Renne v.

Geary, 501 U.S. 312, 316, 111 S.Ct. 2331, 115 L.Ed.2d 288 (1991), the plaintiffs must show (1) that “the claim [is] fit ... for judicial decision in the sense that it arises in a concrete factual context and concerns a dispute that is likely to come to pass” and (2) that “the hardship [to them] of withholding court consideration” outweighs the costs of allowing “litigation by hypothetical,” Warshak, 532 F.3d at 525, 529 (omission in original). The plaintiffs fall short on both fronts.

As for fitness, the Supreme Court has previously held that a pre-enforcement self-incrimination challenge to a reporting requirement is “premature” even when the plaintiff insists that he “intend[s] to engage” in the conduct that triggers the requirement — so long as the plaintiff has yet to assert a privilege claim in response to a government demand for disclosure. Cal. Bankers Ass’n v. Shultz, 416 U.S. 21, 72-74, 94 S.Ct. 1494, 39 L.Ed.2d 812 (1974); see also Trs. for Alaska v.

EPA, 749 F.2d 549, 560 (9th Cir.1984) (dismissing as “unripe” a pre-enforcement self-incrimination challenge to self-monitoring, reporting and record-keeping requirements contained in federal pollutant-discharge permits where no one contended those requirements had been “improperly applied in an actual case”). That is precisely the case here. As the record now stands, we simply “have no idea whether or when” the Attorney General will attempt to inspect any of Connection’s records, let alone refuse to respect a proper claim of privilege. Warshak, 532 F.3d at 526 (internal quotation marks omitted); cf.

Free Speech Coal. v. Gonzales, 483 F.Supp.2d 1069, 1081 (D.Colo.2007) (dismissing on standing grounds the plaintiffs’ self-incrimination attack on § 2257 because they failed to show a concrete injury, as they “ha[d] not produced any evidence that they have ever been subjected to an inspection”). As for hardship, the plaintiffs have not shown that “withholding court consideration” until a concrete conflict arises will prejudice them in any material way. They do not face the kind of dilemma that confronted the petitioners in Albertson v.

Subversive Activities Control Board, 382 U.S. 70, 86 S.Ct. 194, 15 L.Ed.2d 165 (1965). There, after unsuccessfully asserting their self-incrimination privilege to the relevant government agency, the individuals were forced to make a winless choice between complying with the registration requirements “without a [judicial] decision on the merits of their privilege claims” or declining to do so and “risking] onerous and rapidly mounting penalties while awaiting the Government’s pleasure whether to initiate a prosecution against them.” Id. at 75-76, 86 S.Ct. 194.

At least until the Attorney General attempts to obtain § 2257 records from these individuals, they face no greater risk of prospective harm than a claimant concerned that the government will violate his Fourth Amendment rights in future searches. Cf. Warshak, 532 F.3d at 533. At this point, their Fifth Amendment claim simply is not ripe. IV. For these reasons, we affirm the district court’s order granting

summary judgment to the government.