United States v. Jay Liestman

97 F.4th 1054

Opinion

In the

United States Court of Appeals For the Seventh Circuit ____________________ No. 21-3225 UNITED STATES OF AMERICA, Plaintiff-Appellee, v.

JAY A. LIESTMAN, Defendant-Appellant. ____________________

Appeal from the United States District Court for the Western District of Wisconsin. No. 3:20-cr-00006-jdp-1 — James D. Peterson, Chief Judge. ____________________

ARGUED OCTOBER 27, 2023 — DECIDED APRIL 8, 2024 ____________________

Before SYKES, Chief Judge, and EASTERBROOK, ROVNER, WOOD, BRENNAN, SCUDDER, ST. EVE, KIRSCH, JACKSON- AKIWUMI, LEE, and PRYOR, Circuit Judges.1 SCUDDER, Circuit Judge. Before us is Jay Liestman’s chal- lenge to the federal sentence he received for transporting child pornography in violation of 18 U.S.C. § 2252(a)(1). The

1 Circuit Judge Kolar did not participate in the consideration or deci-

sion of this case. 2 No. 21-3225

district court imposed an enhanced mandatory minimum sentence of 15 years’ imprisonment under § 2252(b)(1) be- cause Liestman had been convicted seven years earlier of pos- sessing child pornography in violation of Wisconsin law. The question presented is whether that state conviction qualifies as a predicate conviction under § 2252(b)(1), which prescribes enhanced penalties for certain recidivist child sex offenders. Aligning with the approach of a majority of the circuits, we hold that the answer is yes and affirm Liestman’s sentence. I In October 2019 Jay Liestman took to the Kik messenger app and divulged his sexual interest in underage boys to an undercover FBI agent. In ensuing discussions, Liestman sent the agent a link to 561 videos depicting sexual assaults of children. A federal prosecution followed, and Liestman pleaded guilty to a single count of transporting child pornography. See 18 U.S.C. § 2252(a)(1). This was not his first child sex offense. Several years earlier, Liestman’s efforts to meet a fourteen-year-old boy for sex culminated in two felony convictions under Wisconsin law, one for attempted child enticement, see Wis. Stat. § 948.07, and a second for the possession of child pornography, see id. at § 948.12(1m). At the federal sentencing, the government contended that Liestman’s prior conviction for possessing child pornography triggered 18 U.S.C. § 2252(b)(1)’s enhancement for repeat sex offenders, which increases the mandatory minimum term of imprisonment from 5 to 15 years if the defendant has a prior conviction “under the laws of any State relating to … the pro- duction, possession, receipt, mailing, sale, distribution, ship- ment, or transportation of child pornography.” Liestman in- sisted that the enhancement did not apply because Wis. Stat. No. 21-3225 3

§ 948.12(1m) reached offense conduct that Congress did not expressly enumerate in the text of § 2252(b)(1). Relying on our decision in United States v. Kaufmann, 940 F.3d 377 (7th Cir. 2019), the district court disagreed and sentenced Liestman to the enhanced mandatory minimum of 15 years. The parties renew their positions on appeal, and we chose to convene the full court to decide whether Liestman’s prior offense of conviction for possessing child pornography under Wis. Stat. § 948.12(1m) can serve as a predicate offense under § 2252(b)(1). Doing so requires application of the categorical approach. II A For all the consternation it tends to elicit, the categorical approach serves an essential need. Throughout the United States Code, Congress has attached adverse consequences to the fact that a person has been convicted of a certain kind of prior offense. See, e.g., 18 U.S.C. § 924(e)(1) (providing en- hanced sentence for felon-in-possession defendants convicted of three prior violent felonies or serious drug offenses); 8 U.S.C. § 1227(a)(2)(A)(iii) (providing for the removal of non- citizens convicted of aggravated felonies); 5 U.S.C. § 8902a(b)(1) (providing for the debarment of health care ser- vice providers convicted of offenses “relating to fraud, cor- ruption, breach of fiduciary responsibility, or other financial misconduct in connection with the delivery of a health care service or supply”). When Congress does so, it ordinarily describes the range of qualifying offenses in general terms to account for the sheer variety of state and federal laws on the books. See Taylor v. 4 No. 21-3225

United States, 495 U.S. 575, 590–91 (1990) (surveying a range of state burglary offenses); Diaz-Rodriguez v. Garland, 55 F.4th 697, 720–22 (9th Cir. 2022) (canvassing the “wide variety of approaches” states have taken “to labeling, categorizing, and defining crimes against children”). Because of this, determin- ing whether a particular prior offense triggers an adverse con- sequence can be challenging. The categorical approach emerged to address that challenge. Its cornerstone—rooted in both practical and Sixth Amendment concerns—is its insistence that we look only to the formal definition of the prior offense, cutting real-world facts out of the equation. Mathis v. United States, 579 U.S. 500, 504 (2016); see also Kawashima v. Holder, 565 U.S. 478, 483 (2012) (emphasizing that the categorical approach looks to “the statute defining the crime of conviction, rather than the specific facts underlying the crime”). Under the categorical approach, a prior offense can trigger a statutory consequence only if its statutory elements are defined in such a way that all possible violations of the statute, however committed, would fall within Congress’s chosen federal benchmark. If so, then an offense is one that categorically—meaning in all cases— triggers the federal statutory consequence. The Supreme Court first interpreted a statute to require categorical analysis in Taylor v. United States, 495 U.S. 575 (1990). There the Court addressed whether Arthur Taylor’s prior convictions for second-degree burglary under Missouri law qualified as “violent felon[ies]” that could trigger an en- hanced sentence under § 924(e) of the Armed Career Criminal Act. Id. at 578–79. That Act defines the term “violent felony” to include, among other offenses, any crime that “is bur- glary.” 18 U.S.C. § 924(e)(2)(B) & (B)(ii). No. 21-3225 5

Focusing on the text, structure, and history of the enhance- ment, the Court concluded that the word “burglary” in § 924(e) referred to “the generic sense in which the term [was then] used in the criminal codes of most States.” 495 U.S. at 598. It then devised a generic definition of “burglary” cover- ing any offense that has “the basic elements of unlawful or unprivileged entry into, or remaining in, a building or struc- ture, with intent to commit a crime.” Id. at 599. To determine whether Taylor’s convictions met that generic understanding of “burglary,” the Court looked to the elements of Taylor’s state convictions alone, without regard to how he actually committed those crimes. Section 924(e), the Court explained, “mandates a formal categorical approach” that “look[s] only to the statutory definitions of the prior offenses, and not to the particular facts underlying those convictions.” Id. at 600. Today Taylor stands as the prime example of the so-called generic strand of categorical analysis. Its rationale is straight- forward. When Congress hinges the applicability of a statu- tory consequence on whether a defendant’s prior convictions qualify as a certain kind of offense—like burglary—we as- sume that Congress intended to give that term a uniform, fed- eral “definition independent of the labels used by the various States’ criminal codes.” Id. at 575. And courts can give effect to Congress’s “unadorned reference” to an offense only by “com[ing] up with a ‘generic’ version of the crime” against which the elements of state offenses can be compared. Shular v. United States, 140 S.Ct. 779, 783 (2020). In the years since Taylor, the generic approach has played an important role in our categorical approach case law. See, e.g., United States v. Hatley, 61 F.4th 536, 539 (7th Cir. 2023) (ge- neric extortion); United States v. Misleveck, 735 F.3d 983, 988 6 No. 21-3225

(7th Cir. 2013) (generic arson). But sometimes the categorical approach must proceed in a different way. As the Supreme Court recently explained in Shular, Congress has drafted many federal sentencing enhancements in ways that make the generic approach a poor fit. Instead of prompting courts to ask whether prior offenses qualify as discrete crimes like “burglary,” “arson,” or “extortion,” many enhancements turn instead on whether a defendant’s prior offense has some other attribute. See 140 S.Ct. at 783 (explaining that many statutes “ask the court to determine not whether the prior conviction was for a certain offense, but whether the conviction meets some other criterion”). The question these statutes ask is not whether a prior conviction is a particular kind of offense, but rather whether something else is true of its statutory ele- ments. Consider, for example, the Armed Career Criminal Act’s definition of “serious drug offense”: any “offense under State law[] involving manufacturing, distributing, or possessing with intent to manufacture or distribute, a controlled sub- stance.” 18 U.S.C. § 924(e)(2)(A)(ii). In Shular, the Court did not view the various categories of conduct listed by this pro- vision as offenses in need of generic definition. See 140 S.Ct. at 784–85. Looking to “statutory text and context,” and in par- ticular to Congress’s use of the word “involving” rather than “is,” the Court concluded that Congress intended to reach any state offense whose elements “necessarily entail one of the types of conduct identified in § 924(e)(2)(A)(ii).” See id. (cleaned up). Taylor and Shular illustrate that the categorical approach is not a one-size-fits-all formula. Rather, the proper categorical analysis can take different forms depending on the language No. 21-3225 7

Congress uses to frame the federal benchmark against which courts must compare prior offenses. Although categorical analysis always focuses on the elements of prior offenses, the precise mechanics of deciding whether those elements trigger a statutory consequence turn on how Congress articulates the applicable federal benchmark. B With these principles in mind, we return to the question before us. We start from the common point of agreement be- tween Liestman and the government that 18 U.S.C. § 2252(b)(1) calls for a categorical analysis of one sort or an- other. Our task is to determine what Congress intended that analysis to look like. To put things another way, we must in- terpret the language used to frame § 2252(b)(1)’s federal benchmark and decide what kind of categorical comparison it calls for, keeping in mind that Congress is free—subject only to constitutional constraints—to frame its sentencing en- hancements to require whatever inquiry it thinks most pru- dent. By its terms, § 2252(b)(1) states that a defendant like Jay Liestman who violates § 2252(a)(1) “shall” serve a default sen- tence of “not less than 5 years and not more than 20 years.” The mandatory statutory range increases to 15 to 40 years, however, if “such person has a prior conviction” under the laws of any State relating to aggra- vated sexual abuse, sexual abuse, or abusive sexual conduct involving a minor or ward, or the production, possession, receipt, mailing, sale, distribution, shipment, or transportation of 8 No. 21-3225

child pornography, or sex trafficking of chil- dren. Id. No one contends that Liestman’s prior conviction for pos- sessing child pornography relates to sex trafficking or sexual abuse as § 2252(b)(1) uses these terms. So we are left to deter- mine whether Wis. Stat. § 948.12(1m) is categorically an of- fense “relating to … the production, possession, receipt, mail- ing, sale, distribution, shipment, or transportation of child pornography.” That observation takes us to the text of Wis. Stat. § 948.12(1m) as it stood at the time Liestman violated that statute in 2013. See Portee v. United States, 941 F.3d 263, 266 (7th Cir. 2019) (“We consider the version of the State’s crimi- nal statute in effect at the time of the offense.”). Because the statute and its implementing definitions remain the same to- day as they did at the time of Liestman’s offense conduct, we refer to the current version of the statute for ease of readabil- ity. Wis. Stat. § 948.12(1m) provides that [w]hoever possesses, or accesses in any way with the intent to view, any undeveloped film, photographic negative, photograph, motion picture, videotape, or other recording of a child engaged in sexually explicit conduct under all of the following circumstances [commits a Class D felony]: (a) The person knows that he or she possesses or has accessed the material. (b) The person knows, or reasonably should know, that the material that is possessed or No. 21-3225 9

accessed contains depictions of sexually ex- plicit conduct. (c) The person knows or reasonably should know that the child depicted in the material who is engaged in sexually explicit conduct has not attained the age of 18 years. Wis. Stat. § 948.12(1m). With the pertinent statutory language on the table, we agree with Liestman that Wis. Stat. § 948.12(1m) is broader than § 2252(b)(1) in two respects. First, unlike § 2252(b)(1), § 948.12(1m) prohibits “access[ing]” child pornography in ad- dition to possessing it. Second, Wisconsin law considers a wider range of material to be child pornography. Both Wis- consin and federal law criminalize the possession of material depicting minors engaging in “sexually explicit conduct.” See 18 U.S.C. § 2256(8); Wis. Stat. § 948.12(1m). But while Wiscon- sin defines this term as any “[l]ewd exhibition of intimate parts”—to include “the breast, buttock, anus, groin, scrotum, penis, vagina or pubic mound”—federal law defines it to cover only the “lascivious exhibition of the anus, genitals, or pubic area.” Compare 18 U.S.C. § 2256(2)(A)(v) with Wis. Stat. §§ 939.22(19); 948.01(7)(e). So a defendant can be con- victed in Wisconsin for possessing material that would not support a federal prosecution. See United States v. Gleich, 397 F.3d 608, 614 (8th Cir. 2005) (holding that an image of “partial buttocks” was “not of genitals or of a pubic area” and there- fore did not meet the federal “definition of sexually explicit conduct”). Resisting this conclusion, the government contends that the Wisconsin Supreme Court’s decision in State v. Petrone, 10 No. 21-3225

468 N.W.2d 676 (1991), eliminates any possibility that a de- fendant could be convicted under Wis. Stat. § 948.12(1m) for possessing images of a minor’s bare breast or buttock. On the government’s reading, Petrone held that in order for an image to be “lewd” it must expose a “child’s genitals or pubic area.” 468 N.W.2d at 688. In the government’s view, then, Petrone forecloses the possibility that an image of a minor’s bare breast or buttock could qualify as “lewd” within the meaning of Wis. Stat. § 948.01(7)(e)—even though Wis. Stat. § 939.22(19) expressly identifies both as “intimate parts.” We are not persuaded. Foremost, it makes little sense to conclude that the Wisconsin legislature, in adding “breast” and “buttock” to § 939.22(19), did not intend to prohibit the possession of images displaying those parts of a child’s body. More, Petrone itself interpreted a different offense, Wis. Stat. § 940.203(2) (1988), that incorporated a narrower definition of “sexually explicit conduct” covering only the “lewd exhibi- tion of the genitals or pubic area of any person.” Id. § 940.203(6)(b) (1988). This makes clear that Petrone rooted the limitations it placed on the scope of § 940.203(2) in § 940.203(6)(b)’s definition of “sexually explicit conduct,” not in the meaning of “lewd.” We have little trouble concluding, then, that a defendant could be prosecuted under § 948.12(1m) for possessing mate- rial that would not be considered child pornography under federal law. We therefore agree with Liestman that § 948.12(1m) is broader than § 2252(b)(1) in this way as well. C But that conclusion does not resolve this appeal. After all, § 2252(b)(1) requires only that a prior state offense “relat[e] to” No. 21-3225 11

the conduct it enumerates to trigger an enhanced sentence. Whether § 948.12(1m) qualifies depends on the effect the phrase “relating to” has on the nexus required between a prior offense’s elements and the conduct enumerated by the federal enhancement. Does it, as Liestman argues, disqualify as a predicate any offense broader than the conduct specified in the enhancement? Or does it signal an intent by Congress to adopt a less exacting standard? This presents a pure ques- tion of statutory interpretation. The proper beginning point is the phrase “relating to” it- self. Congress left the term undefined, requiring us to give it its ordinary meaning absent countervailing evidence of a con- trary intent in the text or structure of § 2252(b)(1). See Perrin v. United States, 444 U.S. 37, 42 (1979) (“A fundamental canon of statutory construction is that, unless otherwise defined, words will be interpreted as taking their ordinary, contempo- rary, and common meaning.”); Smith v. United States, 508 U.S. 223, 228–29 (1993) (giving the word “use” in 18 U.S.C. § 924(c)(1) its ordinary meaning). This presumption is an im- portant one, because the ordinary meaning of “relating to” is broad. As the Supreme Court explained in Morales v. Trans World Airlines, Inc., 504 U.S. 374 (1992), the phrase means “to stand in some relation; to have bearing or concern; to pertain; refer; to bring into association with or connection with.” Id. at 384 (quoting Black’s Law Dictionary 1158 (5th ed. 1979)). Alt- hough Morales concerned federal preemption, the Court has applied this definition in the categorical approach context as well. See Pugin v. Garland, 599 U.S. 600, 607 (2023) (observing that Congress’s use of “relating to” in 8 U.S.C. § 1101(a)(43)(S) “ensures that [the] statute covers offenses that have ‘a connec- tion with’ obstruction of justice”). 12 No. 21-3225

Were “relating to” in § 2252(b)(1) understood in this way—as reaching all state offenses that bear a connection with the enumerated conduct that follows—it would be no stretch to conclude that § 948.12(1m) could trigger an en- hanced federal sentence notwithstanding its overbreadth. That “relating to” should receive its ordinary meaning is only a presumption, however, and like any other presumption it can be overcome. The challenge for Liestman is that other clues in the text, structure, and history of § 2252(b)(1), as well as its place in the overall statutory scheme, only reinforce that Congress intended to use “relating to” in its broad, ordinary sense. First, by giving “relating to” its broad and ordinary mean- ing, we avoid treating that language as synonymous with nar- rower connecting language Congress has used to frame other sentencing enhancements. Consider once more 18 U.S.C. § 924(e)(2)(A)(ii), which defines as a “serious drug offense” any “offense under State law … involving manufacturing, distributing, or possessing with intent to manufacture or dis- tribute, a controlled substance.” In Shular, the Supreme Court held that an offense is one “involving” § 924(e)(2)(A)(ii)’s listed conduct only if it “necessarily entail[s]” it. 140 S.Ct. at 783; see also Kawashima, 565 U.S. at 484 (giving “involving” the same interpretation in 8 U.S.C. § 1101(a)(43)(M)(i)). Inter- preting “relating to” in § 2252(b)(1) in a manner that disqual- ifies as a predicate any state offense that sweeps more broadly than the conduct that “relating to” introduces would be no different from holding that prior offenses must “necessarily entail” (or “involve”) the enumerated conduct. It is, of course, possible for Congress to use different words to convey the same meaning. But the presumption usually runs in the other direction, and we are hesitant to adopt an interpretation that No. 21-3225 13

attaches no significance to Congress’s choice of the broad “re- lating to” language in § 2252(b)(1). Second, in the very statute that added the “relating to” lan- guage to § 2252(b)(1), Congress amended another sentencing enhancement—in 18 U.S.C. § 2241(c)—to expressly require the kind of relationship Liestman reads into § 2252(b)(1). See Child Pornography Prevention Act of 1996 (CPPA), Pub. L. No. 104-208, § 121, 110 Stat. 3009, 3009-30. With respect to the § 2241(c) enhancement, Congress made clear that a state of- fense would trigger enhanced penalties only if it “would have been an offense under” § 2241(a) or (b) “had the offense oc- curred in a Federal prison.” Id. at 110 Stat. 3009-31. What this shows is that the same Congress that enacted § 2252(b)(1) knew full well how to condition the applicability of a sentenc- ing enhancement on a prior offense’s congruence with federal law. See also United States v. Portanova, 961 F.3d 252, 257 & n.29 (3d Cir. 2020) (providing similar examples). We find it hard to believe, then, that Congress used “relating to”—the broadest of connecting language—to achieve the same end in § 2252(b)(1), particularly in light of the availability of nar- rower phrases like “involving” or even “is” that Congress has used to frame other statutes. Third, at the time Congress added the “relating to” lan- guage to § 2252(b)(1) in 1996, only a fraction of states defined child pornography in a manner congruent with or narrower than the federal definition. By the government’s accounting, which Liestman does not meaningfully contest, 40 states crim- inalized the possession of some material—like an image of a minor’s breast or buttock—that federal law does not reach. See Addendum to Supplemental Brief of United States. If Liestman’s cramped interpretation of “relating to” is right, it 14 No. 21-3225

would mean that on the day of its enactment, a large swath of § 2252(b)(1) could apply to only a handful of states scattered across the country. Remember that the enhancement’s enu- merated list of acts—including possession, receipt, mailing, and sale—all take as their object the defined term “child por- nography.” If “relating to” is not broad enough to permit overbreadth in state definitions of child pornography, this portion of the enhancement had no effect in the vast majority of states at the time Congress expanded § 2252(b)(1) to cover state offenses. Given the purpose of the enhancement to “ad- dress high recidivism rates among child sex offenders,” United States v. Kraemer, 933 F.3d 675, 683 (7th Cir. 2019)—no doubt a persistent and grave problem—it defies belief that Congress intended for the enhancement to have such limited effect. The Supreme Court has relied on this kind of backdrop evidence in rejecting interpretations of other sentencing enhancements. Returning to Taylor, the Court there rejected a narrow, common-law definition of burglary in part because “construing ‘burglary’ to mean common-law burglary would come close to nullifying that term’s effect in the statute, because few of the crimes now generally recognized as burglaries would fall within the common-law definition.” 495 U.S. at 594. More recent examples abound, and each comes directly to us from the Supreme Court. See, e.g., Pugin, 599 U.S. at 607 (admonishing that courts “should not lightly conclude that Congress enacted a self-defeating statute”) (internal quotation marks omitted); Quarles v. United States, 139 S.Ct. 1872, 1879 (2019) (avoiding an interpretation that would “eliminate[]” “many States’ burglary statutes … as predicate offenses under § 924(e)”); Stokeling v. United States, 139 S.Ct. 544, 551–53 (2019) (rejecting an interpretation under No. 21-3225 15

which “many States’ robbery statutes would not qualify as ACCA predicates”); United States v. Stitt, 139 S.Ct. 399, 403, 406 (2018) (interpreting “burglary” in § 924(e) to include the burglary “of a structure or vehicle that has been adapted or is customarily used for overnight accommodation” in part because “a majority of state burglary statutes” covered such places at the time Congress enacted the enhancement into law); Esquivel-Quintana v. Sessions, 581 U.S. 385, 395 (2017) (declining to adopt interpretation that “would categorically exclude the statutory rape laws of most States”). Together, these points combine to reinforce the starting presumption that Congress used “relating to” in § 2252(b)(1) in its broad ordinary sense. Congress’s use of narrower lan- guage in other statutes, its proven ability to use different words to require congruence between prior offenses and fed- eral law, and the effect a narrower construction would have on the scope of the enhancement all suggest an intent to heighten sentences for defendants convicted of state offenses that bear a connection with any of § 2252(b)(1)’s enumerated list of child-pornography-related acts, even if those offenses sweep more broadly than § 2252(b)(1) in some respects. Albeit in more abbreviated reasoning, we reached this pre- cise conclusion in United States v. Kaufmann. In no uncertain terms, we rejected the view that § 2252(b)(1) “require[s] the state statute of conviction to be the same as or narrower than … analogous federal law,” 940 F.3d at 378, and instead held that Darin Kaufmann’s prior Indiana convictions for pos- sessing child pornography triggered the recidivism enhance- ment even though Indiana, like Wisconsin, defines child por- nography more broadly than does federal law. See id. at 380– 81. 16 No. 21-3225

Kaufmann was not an aberration. Then, and now, a major- ity of circuits to have interpreted “relating to” in § 2252(b)(1) and materially identical enhancements elsewhere in Chapter 110 of Title 18 of the U.S. Code, see 18 U.S.C. §§ 2251(e); 2252(b)(2); 2252A(b)(1); 2252A(b)(2), have given that phrase its broad, ordinary meaning and permitted state offenses to serve as predicates despite some amount of overbreadth. See United States v. Mayokok, 854 F.3d 987, 993 & n.2 (8th Cir. 2017) (holding that Congress used “relating to” in § 2252(b)(1) to “subject a wider range of prior convictions to the § 2252(b)(1) enhancement” and that the defendant’s Minnesota conviction for possessing child pornography could serve as a predicate even though Minnesota defines child pornography more broadly than federal law); Portanova, 961 F.3d at 257–58, 262 (same); United States v. Bennett, 823 F.3d 1316, 1322–25 (10th Cir. 2016) (holding that Colorado conviction was one “relat- ing to” the possession of child pornography within the mean- ing of § 2252A(b)(1) even though it “punish[es] the possession of [some] visual depictions that fall outside the federal defini- tion of child pornography”); but see United States v. McGrat- tan, 504 F.3d 608 (6th Cir. 2007); United States v. Reinhart, 893 F.3d 606 (9th Cir. 2018). D Represented by very able counsel, Liestman urges us to depart from the majority position and give § 2252(b)(1) a nar- rower construction, principally on the basis of cases like Mellouli v. Lynch, 575 U.S. 798 (2015), and United States v. Ruth, 966 F.3d 642 (7th Cir. 2020), that have interpreted “relating to” more narrowly in other sentencing enhancements. From these cases, he presses a general rule that introductory language like “relating to” should not be understood to capture No. 21-3225 17

offenses that sweep more broadly than the language it intro- duces, particularly when that language is defined. At times, Liestman appears to take this position as a matter of statutory construction. At other times, he seems to see it as a general limitation on how the categorical approach operates when a sentencing enhancement lists conduct rather than offenses. Both views are mistaken. On the latter point, the Supreme Court has never inti- mated that non-generic categorical analysis—which Liestman fairly calls the conduct-based approach—must follow me- chanical rules applicable to all sentencing enhancements without regard to differences in text, structure, and purpose. To the contrary, the unifying principle that ties the Supreme Court’s categorical approach cases together is the recognition that whether the categorical approach applies at all—and, if so, what form it takes—are fundamentally questions of statu- tory interpretation. See, e.g., United States v. Davis, 139 S.Ct. 2319, 2327 (2019) (observing that whether the categorical ap- proach applies can be determined only through examination of statute’s text, context, and history); Esquivel-Quintana, 581 U.S. at 391 (explaining that the meaning of “sexual abuse of a minor” must be determined “using the normal tools of statu- tory interpretation”). Liestman’s rule also stands at odds with the Supreme Court’s recent decision in Pugin v. Garland, which held that, as used in 8 U.S.C. § 1101(a)(43)(S), the phrase “relating to” carries its broad, ordinary meaning. See 599 U.S. at 607. That provision, in conjunction with 8 U.S.C. § 1227(a)(2)(A)(iii), makes removable any non-citizen convicted of a felony “relating to obstruction of justice.” The question in Pugin was whether a state conviction can trigger removal where the state 18 No. 21-3225

offense does not require obstruction of a pending investigation or legal proceeding. See id. at 602–03. In holding that it can, the Court relied most heavily on dictionary definitions and state and federal law demonstrating that the backdrop understanding of obstruction of justice at the time Congress added § 1101(a)(43)(S) to the Immigration and Nationality Act encompassed pre-investigatory methods of obstruction. But the Court was quick to underscore that Congress’s use of the phrase “‘relating to’ … resolved” “any doubt [that] remain[ed]” about the scope of the statute. Id. at 607. The choice of the words “relating to,” the Court explained, “ensures that [the] statute covers [any] offense[] that ha[s] ‘a connection with’ obstruction of justice.” Id. Pugin shows that nothing inherent in the categorical ap- proach precludes courts from giving “relating to” its ordinary meaning. Indeed, it is difficult to imagine where any such lim- itation would come from other than the Constitution, which Liestman does not invoke in this appeal. Cf. Portanova, 961 F.3d at 262–63 (considering and rejecting an as-applied void- for-vagueness challenge to § 2252(b)(1)). When Congress frames a sentencing enhancement, it is generally free to pre- scribe whatever approach it thinks will best achieve its policy aims. We therefore see no reason why—as a matter of statu- tory construction—the phrase “relating to” cannot be under- stood broadly in § 2252(b)(1). Or stated another way, we see no reason why Congress could not have intended to use the phrase “relating to” to capture state offenses that, although broader than § 2252(b)(1)’s enumerated list of conduct in some respects, bear the sort of connection to that conduct that the phrase “relating to” typically captures. And, indeed, the No. 21-3225 19

statutory features discussed above lead us to believe that this was Congress’s intent with § 2252(b)(1). Neither Mellouli nor Ruth calls this conclusion into ques- tion. Today we apply the categorical approach while giving the phrase “relating to” its broad ordinary meaning in the context of § 2252(b)(1). But we do not suggest that this mean- ing extends equally to all instances where “relating to” ap- pears in the U.S. Code. As with any statutory term, we must take each appearance as it comes, discerning meaning from a holistic analysis of text, context, purpose, and history. See Gundy v. United States, 139 S.Ct. 2116, 2126 (2019) (describing statutory interpretation as a “holistic endeavor, which deter- mines meaning by looking not to isolated words, but to text in context, along with purpose and history”). Mellouli and Ruth involved different statutory contexts. Take Mellouli first. The Supreme Court there considered whether Moones Mellouli’s Kansas conviction for possessing drug paraphernalia triggered removal under the Immigration and Nationality Act. See 575 U.S. at 802. The operative provision, 8 U.S.C. § 1227(a)(2)(B)(i), makes removable any non-citizen convicted of violating “any law or regulation of a State, the United States, or a foreign country relating to a controlled substance (as defined in section 802 of Title 21).” (Emphasis added.) The parenthetical cross reference to § 802 proved important, because § 802(6) defines “controlled substance” to include any “drug or other substance, or immediate precursor” listed on the federal drug schedules. 21 U.S.C. § 802(6). The Court declined to adopt a broad interpretation of the term “relating to” in § 1227(a)(2)(B)(i). See id. at 811. Instead, it held that a prior offense triggers removal under the statute 20 No. 21-3225

only if it necessarily involves a federally scheduled drug. See id. In reaching that conclusion, however, the Court recognized that the phrase “relating to” is ordinarily “broad.” Id. In de- parting from that ordinary meaning, the Supreme Court stressed that a number of textual and historical clues specific to § 1227(a)(2)(B)(i) “tug[ged] … in favor of a narrower read- ing.” Id. at 812 (internal quotation marks omitted). The statute’s history stood front and center in the Court’s analysis. Earlier versions of § 1227(a)(2)(B)(i), the Court em- phasized, enumerated specific, federally scheduled drugs like “opium, coca leaves, [and] cocaine.” Id. at 806–07 (internal quotation marks omitted). But “[o]ver time, Congress amended the statute to include additional … drugs.” Id. at 807. This “increasingly long list” of individually identified drugs became unwieldy, leading Congress in 1986 to replace the specific enumeration of qualifying controlled substances with a simple cross-reference to the federal drug schedules through 21 U.S.C. § 802. See id. All of this made evident that Congress’s use of the phrase “relating to” in § 1227(a)(2)(B)(i) was not intended to authorize removal based on convictions involving non-federally scheduled drugs and, therefore, that Mellouli’s Kansas conviction for possessing drug parapher- nalia was not a categorical match. Ruth is much the same—a holding rooted in the history and context of the statute under review. In Ruth, we ad- dressed whether an Illinois cocaine conviction qualified as a “felony drug offense” that would trigger a sentencing en- hancement under 21 U.S.C. § 841(b)(1)(C). See 966 F.3d at 645– 46. Federal law defines “felony drug offense” to include any offense “that prohibits or restricts conduct relating to narcotic drugs.” 21 U.S.C. § 802(44) (emphasis added). The term No. 21-3225 21

“narcotic drugs,” in turn, is defined in exactingly technical de- tail, down to the specific kinds of chemical isomers of co- caine—optical and geometric, but not positional—that qual- ify. See 21 U.S.C. § 802(17). Invoking Shular, we held that the Illinois conviction could satisfy that standard only if it “nec- essarily entail[s] the conduct identified in § 802(44).” Id. at 647. Our insistence on a strict one-for-one match was driven by the highly technical nature of the statutes at issue, the sheer particularity of which suggested that Congress’s definition was intended to be exhaustive of the kinds of drugs that can trigger an enhancement under § 841(b)(1)(C). Our overarching point is that neither Mellouli nor Ruth supports Liestman’s key contention—that the phrase “relat- ing to” cannot receive its broad, ordinary meaning when in- troducing sentence-enhancing conduct. Instead, those cases turned on statutory and historical features that have limited relevance to the proper interpretation of § 2255(b)(1), a provi- sion whose meaning must be discerned in the light of its own statutory context. Because that context supports a broad un- derstanding of “relating to,” we hold—in alignment with our decision in Kaufmann and with the views of a majority of the circuits to have considered the issue—that “relating to” in § 2252(b)(1) brings within the ambit of the enhancement any prior offense that categorically bears a connection with (or, put in statutory terms, “relates to”) “the production, posses- sion, receipt, mailing, sale, distribution, shipment, or trans- portation of child pornography,” regardless of whether it sweeps more broadly than that enumerated conduct in some respects. 22 No. 21-3225

III All that remains is to apply our holding to Jay Liestman’s case. The question is whether Wis. Stat. § 948.12(1m) categori- cally—meaning in all cases—prohibits conduct “relating to” the conduct enumerated by 18 U.S.C. § 2252(b)(1). As in any other categorical approach case, we must assume that Liest- man was convicted for “the least serious conduct” covered by § 948.12(1m), Borden v. United States, 593 U.S. 420, 441 (2021), which would be for accessing images lewdly exhibiting a mi- nor’s breast or buttock. We must then ask whether that con- duct bears the necessary connection to the conduct enumer- ated in § 2252(b)(1). We have little trouble concluding that ac- cessing images lewdly exhibiting a minor’s breasts or but- tocks bears the necessary connection and, therefore, that § 948.12(1m) is a categorical match. The line between “accessing” and “possessing” child por- nography is razor thin, if indeed one exists at all. Accessing was added to § 948.12(1m) to plug a possible hole in the statute made apparent by cases like State v. Mercer, 782 N.W.2d 125 (Wis. 2010), in which defendants began to argue that their ef- forts to view child pornography over the internet did not qualify as “possession” as that term has historically been un- derstood. But those arguments never succeeded. So when the Wisconsin legislature added “accessing” to § 948.12(1m), any practical difference between accessing and possessing re- mained theoretical. It appears to remain so today. Liestman has not identified a single case—nor has our independent research uncovered one—in which a defendant has been convicted of accessing child pornography under circumstances that would not qual- ify as possession under Wisconsin law. Indeed, under Mercer, No. 21-3225 23

“an individual knowingly possesses child pornography when that individual affirmatively pulls up images of child pornog- raphy on the Internet and views those images knowing that they contain child pornography.” Id. at 136. We find it diffi- cult to imagine, let alone see with any clarity, daylight be- tween accessing and possessing child pornography under § 948.12(1m). Even if such a difference could be posited in the abstract, accessing child pornography still clearly bears the requisite connection to the possession or receipt of child pornography. Statutes that criminalize accessing and possessing child por- nography “address[] the same harm—sexual exploitation of minors—that [§ 2252(b)(1)] targets.” Kaufmann, 940 F.3d at 380. Both seek to penalize participation in the market for sex- ually abusive images of minors. Though they may do so in different ways, the core purpose is the same. Nothing about the broader scope of Wisconsin’s child por- nography laws changes this. While we agree that Wisconsin’s definition of child pornography reaches anatomy—the breasts and buttocks—that federal law does not, images of those parts of the body qualify as “sexually explicit conduct” under § 948.01(7)(e) only if they are “lewd.” At the time Liest- man violated § 948.12(1m), an image of a minor’s breast or buttock could qualify as “lewd” only if the minor was por- trayed as a “sex object.” United States v. Griesbach, 540 F.3d 654, 655 (7th Cir. 2008) (internal quotation marks omitted). There can be little doubt that accessing such images “relate[s] to” the conduct described by § 2252(b)(1)—namely, the traf- ficking (in all manner of ways) of sexually abusive imagery. Wis. Stat. § 948.12(1m) is therefore a categorical match. For these reasons, we AFFIRM. 24 No. 21-3225

WOOD, Circuit Judge, joined by ROVNER, JACKSON- AKIWUMI, LEE, and PRYOR, Circuit Judges, dissenting. When the Framers of the Constitution “split the atom of sover- eignty,” as Justice Kennedy put it in U.S. Term Limits, Inc. v. Thornton, 514 U.S. 779, 838 (1995) (Kennedy, J., concurring), they set in motion complex forces with which we are still dealing today. On the plus side, our federal system pre- serves local accountability and choice, while at the same time it harnesses the power of the country as a whole for matters of national or international concern. But on the mi- nus side, we have learned that federalism isn’t always easy, and it isn’t always neat. The question now occupying the en banc court’s attention is a case in point. Federal criminal law intersects with its counterparts in the 50 states, the District of Columbia, and the territories in myriad ways. One of those ways—the one involved here—relates to the use of state convictions to enhance a federal sentence. In an ideal world, it would be easy to identify which state convictions should be used in conjunction with federal sentencing: whichever ones Congress specifies in the rele- vant recidivism statute. But applying that rule turns out to be easier said than done. Enter the Supreme Court: In order to foster uniformity across the country in the face of the countless variations in state statutes, the Court has inter- preted the federal recidivism statutes as normally taking a “categorical” approach to the task of comparing a state con- viction with a federal counterpart. (Obviously, when Con- gress instructs otherwise, it has the last word.) But even the categorical approach has at least two, and perhaps more, forms, depending on the language of the statute that is be- ing applied. The majority in the present case has concluded that certain state convictions broadly relate to Jay Liest- No. 21-3225 25

man’s federal offense, and thus a recidivism enhancement to his sentence was properly assessed. With respect, I do not read the governing statute in the same way, and so I dissent. I Because the majority has provided the key background facts about Liestman’s conviction, I move straight to the legal question before the en banc court: whether the punishment for Liestman’s federal crime of transporting child pornography in violation of 18 U.S.C. § 2252(a)(1) was subject to being enhanced. If so, he faced not a five-year minimum and 20-year maximum term of imprisonment, but instead a 15-year minimum and 40-year maximum based on his earlier Wisconsin conviction for possessing child pornography in violation of Wis. Stat. § 948.12(1m). See 18 U.S.C. § 2252(b). Like the majority, I understand this to be a question of law—one that turns in the first instance on the proper interpretation of section 2252(b), but that also depends on the pertinent state law. But before turning to that issue, it is useful to review what the “categorical” approach is and how the Supreme Court has applied it. The case usually credited for adopting this method of reconciling state convictions with federal law is Taylor v. United States, 495 U.S. 575 (1990). The question before the Court in that proceeding concerned the meaning of the word “burglary” as it was used in 18 U.S.C. § 924(e), which “provides a sentence enhancement for a defendant who is convicted under 18 U.S.C. § 922(g) … and who has three prior convictions for specified types of offenses, including ‘burglary.’” 495 U.S. at 577–78. The problem arose because there was (and still is) wide variation among the states over the conduct that qualifies as “burglary.” After examining 26 No. 21-3225

the legislative history of the enhancement provision, the Court concluded that it “always has embodied a categorical approach to the designation of predicate offenses.” Id. at 588. In other words, Congress was trying to identify predi- cate crimes that have “certain specified elements,” not crimes that happen to bear the label “burglary.” Id. at 588– 89. Lest there was any doubt, the Court underscored that it found “implausible” the idea “that Congress intended the meaning of ‘burglary’ for purposes of § 924(e) to depend on the definition adopted by the State of conviction.” Id. at 590. Rather than relying on state definitions, the Taylor Court settled on a generic definition of burglary. But that was not Taylor’s only contribution to this area. Equally important was the way in which the Court applied that generic definition of burglary to the case before it. That required the resolution of another general issue: “whether the sentencing court in applying § 924(e) must look only to the statutory definitions of the prior offenses, or whether the court may consider other evidence concerning the de- fendant’s prior crimes.” Id. at 600. It opted for what it called a “formal” categorical approach, under which the sentenc- ing court may consider only the statutory definitions and elements of a prior crime, not the particular facts underly- ing those convictions. Id. In the years following Taylor, the Court has returned fre- quently to the categorical approach. See, e.g., Begay v. United States, 553 U.S. 137 (2008); Kawashima v. Holder, 565 U.S. 478 (2012); Moncrieffe v. Holder, 569 U.S. 184 (2013); Descamps v. United States, 570 U.S. 254 (2013); Mathis v. United States, 579 U.S. 500 (2016); Esquivel-Quintana v. Sessions, 581 U.S. 385 (2017); United States v. Stitt, 586 U.S. __, 139 S. Ct. 399 (2018); No. 21-3225 27

Stokeling v. United States, 586 U.S. __, 139 S. Ct. 544 (2019); Shular v. United States, 589 U.S. 154 (2020); Pereida v. Wil- kinson, 592 U.S. 224 (2021); Borden v. United States, 593 U.S. 420 (2021); United States v. Taylor, 596 U.S. 845 (2022); Pugin v. Garland, 599 U.S. 600 (2023). And this issue is a staple of the lower courts’ diet. For present purposes, it is not neces- sary to dissect every one of these Supreme Court decisions. It is enough briefly to discuss Shular, and then to consult the other decisions as needed when I turn to the present appeal. The setting of Shular was a familiar one: it dealt with the proper way to apply the enhancement mandated by the Armed Career Criminal Act, 18 U.S.C. § 924(e), to a defend- ant with prior convictions for a “serious drug offense.” 589 U.S. at 156. This was a trickier problem than the one ad- dressed in Taylor, where the Court was dealing with an old common-law crime, burglary, and it had only to decide how that crime should be defined for purposes of section 924(e). With the straightforward Taylor solution off the table, the parties in Shular presented two options to the Court: the first, urged by the government, compared the elements of the prior state offense to the conduct identified in section 924(e) (that is, the “manufacturing, distributing, or pos- sessing with intent to manufacture or distribute, a con- trolled substance,” id.); the second, advanced by Shular, re- garded each of the activities listed in the statute as separate offenses, next deduced what those generic offenses were, and finally compared the state elements to those generic crimes. The Court opted for the government’s approach. In so doing, it added a layer of complexity to Taylor’s categorical approach. While the analysis is still applied in a categorical 28 No. 21-3225

fashion—that is, in a way that does not depend on the facts of the particular case—Shular establishes that the categorical approach is a methodology that can be used in different ways. Specifically, it identifies two versions of the categori- cal approach. The first and more familiar one “requires the court to come up with a ‘generic’ version of a crime—that is, the elements of ‘the offense as commonly understood.’” Id. at 158 (citation omitted). The second (previously unrecog- nized) variant asks the court “to determine not whether the prior conviction was for a certain offense, but whether the conviction meets some other criterion.” Id. One such criteri- on focuses on the elements of an offense: the court must de- termine whether the proposed state predicate offense has the designated elements that Congress highlighted. Id. The latter approach, the Court held, was the appropriate one for Shular’s case. It explained that the operative terms describing a “serious drug offense” logically referred to conduct, not to any recognizable crime such as burglary, ar- son, or extortion. The Court rejected the idea that something “involving” the designated activities necessarily described separate crimes. Wrapping up the opinion, the Court con- cluded that Shular’s prior Florida conviction for selling co- caine and possessing that drug with the intent to sell it “in- volved” precisely the conduct covered by the federal stat- ute, and thus the enhancement was proper. With this general background about the categorical ap- proach, I now turn to the particular statutes involved in Liestman’s case. No. 21-3225 29

II A My first step is to look carefully at the statute of convic- tion, the statute governing sentencing, and the predicate state-law offense. For convenience, I set them out here, so that the reader will not need to flip back to the majority opinion to find them. First, we have the statute of convic- tion, 18 U.S.C. § 2252(a)(1): (a) Any person who— (1) knowingly transports or ships using any means or facility of interstate or foreign commerce or in or affecting interstate or foreign commerce by any means including by computer or mails, any visual depiction, if— (A) the producing of such visual depiction involves the use of a minor engaging in sex- ually explicit conduct; and (B) such visual depiction is of such con- duct; …. shall be punished as provided in subsection (b) of this section. Next there is the sentencing provision that applies for section 2252(a) offenses, 18 U.S.C. § 2252(b)(1), which reads as follows in pertinent part: Whoever violates … paragraph (1) … of subsec- tion (a) shall be fined under this title and imprisoned not less than 5 years and not more than 20 years, but 30 No. 21-3225

if such person has a prior conviction … under the laws of any State relating to aggravated sexual abuse, sexual abuse, or abusive sexual conduct involving a minor or ward, or the production, possession, receipt, mail- ing, sale, distribution, shipment, or transportation of child pornography, or sex trafficking of children, such person shall be fined under this title and impris- oned for not less than 15 years nor more than 40 years. Id. (emphasis added). Finally, there is the language of Wis. Stat. § 948.12(1m), the state law underlying Liestman’s potentially qualifying prior conviction: Whoever possesses, or accesses in any way with the intent to view, any undeveloped film, photo- graphic negative, photograph, motion picture, vide- otape, or other recording of a child engaged in sex- ually explicit conduct under all of the following cir- cumstances may be penalized under sub. (3): (a) The person knows that he or she possesses or has accessed the material. (b) The person knows, or reasonably should know, that the material that is possessed or accessed contains depictions of sexually explicit conduct. (c) The person knows or reasonably should know that the child depicted in the material who is en- gaged in sexually explicit conduct has not attained the age of 18 years. Liestman offers two reasons why his prior state convic- tion under section 948.12(1m) does not qualify as a predi- No. 21-3225 31

cate for purposes of section 2252(b)(1). First, he argues that the Wisconsin statute sweeps more broadly than the federal law because it encompasses the lewd exhibition of the breasts and buttocks, see Wis. Stat. §§ 948.01(7), 939.22(19), and the federal law does not, see 18 U.S.C. § 2256(2), (8). Second, he contends that accessing child pornography is dif- ferent from possessing child pornography, and that accessing is not one of the enhancers for section 2252, the relevant federal law. I agree with the majority that there is no merit to the latter argument, and so I do not discuss it further. With respect to the former point, Liestman relies principally on Shular and this court’s decision in United States v. Ruth, 966 F.3d 642 (7th Cir. 2020). The government responds that it is permissible for a state law to cover conduct that goes beyond a federal statute when that statute calls only for ear- lier conduct “relating to” the subject matter. The nub of the question before us is thus how those two words, “relating to,” affect the categorical analysis that normally applies to these cases. The majority, ante at 10–16, sees that phrase as a get-out-of-jail-free card from the cate- gorical approach. The phrase, as they understand it, is so broad and undefined, that even the loosest of connections to the federal crime (here, transporting child pornography) will suffice to qualify the earlier state crime as a proper en- hancer under section 2252(b)(1). In so doing, they rely on several cases from this court, and they then turn to guid- ance from the Supreme Court. My analysis of those materi- als, however, leads me to a different conclusion, as I now explain. 32 No. 21-3225

B In order to provide the legal framework for this statutory-interpretation exercise, I begin with the Supreme Court’s decisions interpreting the phrase “relating to” to see how they apply to the sentencing-enhancement issue here. Two cases are particularly relevant: Morales v. Trans World Airlines, Inc., 504 U.S. 374, 383 (1992), and Mellouli v. Lynch, 575 U.S. 798 (2015). I discuss them in that order. A layperson could be forgiven for being somewhat mys- tified by the notion that Morales has anything to do with Liestman’s sordid activities. As I discuss in more detail in a moment, Morales was about airline pricing practices in a market that only recently had been deregulated. Ascertain- ing the proper prison term for a transporter of child por- nography seems to be a far cry from an effort by state attor- neys general to attack certain airline pricing stratagems as deceptive trade practices. For that matter, child pornogra- phy seems to have little to do with the regulation of pension plans under federal law pursuant to the Employee Retire- ment Income Security Act (ERISA), 29 U.S.C. § 1144(a), an analogy on which the Morales Court leaned heavily. My own examination of Morales (and ERISA, for that matter) leads me to conclude that its understanding of the words “relating to” for purposes of airline pricing provides at most the first step of the analysis for the criminal matter be- fore us. The backdrop for Morales was Congress’s decision to move from a world in which airline prices were regulated by an administrative body (the Civil Aeronautics Board) to a world in which unfettered price competition is the order of the day. The Airline Deregulation Act of 1978, 49 U.S.C. No. 21-3225 33

§ 1301 et seq., carried out that decision. The Act contained two provisions relevant here: first, it included a broad pre- emption section, under which the states were not allowed to enforce “any law, rule, regulation, standard, or other provision having the force and effect of law relating to rates, routes, or services of any air carrier …,” 49 U.S.C. § 1305(a) (emphasis added); second, the Deregulation Act contained a “saving” clause, which as of 1992 read as follows: “Nothing contained in this chapter shall in any way abridge or alter the remedies now existing at common law or by statute, but the provisions of this chapter are in addition to such remedies.” 49 U.S.C. App. § 1506, repealed by Pub. L. 103- 272, § 1(e), July 5, 1994, and replaced by 49 U.S.C. § 40120 (stating “[a] remedy under this part is in addition to any other remedies provided by law”). The job before the Court was to decide whether the state deceptive-practice laws fit within the saving clause, or if in- stead the preemption clause had the effect of overriding them. The Court thought that the latter reading was more faithful to section 1305(a)(1)’s express preemption language. The key phrase, it said, was “relating to.” Consulting the fifth edition of Black’s Law Dictionary, the Court observed that the “ordinary meaning” of those words is broad. Some- thing “relates to” another thing if it “stand[s] in some rela- tion; [has] a bearing or concern; pertain[s]; refer[s]; [or] bring[s] into association with or connection with” that other thing. 504 U.S. at 383. The Court found the same breadth in ERISA’s preemption provision, 29 U.S.C. § 1144(a), which preempts all state laws “insofar as they … relate to any em- ployee benefit plan.” 34 No. 21-3225

Broad though the term “relating to” may be, however, it is still necessary to tether it to the legal framework of the law in which we find it. After all, as even the Palsgraf court recognized, chains of causation can become so long that lia- bility dissipates. See Palsgraf v. Long Island R. Co., 248 N.Y. 339 (1928). For want of a nail, the kingdom may have been lost, but after a certain point we let that loss lie where it fell. In the law at issue in Morales, however, Congress signaled that it wanted a broad interpretation of laws that “relate to” airline pricing. And following that instruction, the Court held that the state laws before it were close enough to the core of pricing that they were preempted. This was entirely reasonable. We know from antitrust law that there are many ways to compete on value without direct price-fixing: quantities can be manipulated, warranties can be offered, information can be conveyed in advertising, product differ- entiation can occur, loss leaders can lure people in the door, and so on (and on, and on). If the goal of a federal statute is price deregulation, it makes sense to say that state law should not get in the way of creative efforts by the airlines to compete through such measures as advertising, frequent flier programs (a form of discount from the customer’s per- spective), and compensation for disrupted travel. Boiled down to their essentials, these are all forms of competition, and thus they are all logically encompassed by a law that was designed to substitute a competitive regime for a regu- latory one. In short, there is inevitably a zone within which the de- regulatory framework established by the federal Act must be free to reign, and a space beyond that line in which state law is free to continue to operate. No one thinks that the Act confers carte blanche on airline personnel to steal people’s No. 21-3225 35

suitcases, or to skim their credit cards while the customer is paying to check a bag, or to flout a state’s minimum-wage laws, see Hirst v. Skywest, Inc., 910 F.3d 961, 967 (7th Cir. 2018). Those activities are related to the person’s decision to fly somewhere, but not in a way that requires preemption of state law. The same logic explains why the analogy to ERISA fails in the end. ERISA preempts state laws “insofar as they … relate to any employee benefit plan.” 29 U.S.C. § 1144(a). It deals with sensitive actuarial computations that are de- signed to ensure the financial soundness of retirement and benefit plans. Pull one thread out of the tapestry and the whole thing will unravel. The case of Shaw v. Delta Air Lines, Inc., 463 U.S. 85 (1983), nicely illustrates the point. There the Supreme Court had to decide whether ERISA preempted a New York law forbidding employers from discriminating on the basis of pregnancy. The idea was that the New York law “prohibit[ed] employers from structuring their employ- ee benefit plans” in particular ways that predictably might affect the choices available to the plans’ participants. Id. at 97. The Court held that the law “related to” ERISA plans, reasoning that the history of ERISA’s preemption provision “indicated that the section’s preemptive scope was as broad as its language.” Id. In the course of doing so, it offered the following guidance for determining whether ERISA preemption exists: “A law ‘relate[s] to’ an employee benefit plan, in the normal sense of the phrase, if it has a connec- tion with or reference to such a plan.” Id. In order to see whether the state law has that type of connection to an 36 No. 21-3225

ERISA plan, it is necessary to determine what laws Con- gress intended to supplant. Id. at 95–98. Sometimes the proper conclusion is that Congress did intend that displacement, but as California Div. of Labor Stds. Enforcement v. Dillingham Constr., N.A., 519 U.S. 316 (1997), demonstrated in its rejection of ERISA preemption, some- times Congress has no such intent. The state law there was not preempted for several reasons: it did not have a “con- nection with” ERISA plans, and it did not “reference” ERISA plans because its requirements applied to both ap- prenticeship programs covered by ERISA (those where the parties had set up separate funds) and other programs that were not (those where the employer supported the program out of its general assets). What we learn from Morales and the ERISA examples on which it relied is that it is essential to pay attention to legis- lative context. Neither airline deregulation nor the ERISA scheme operates in a vacuum. It was necessary to ascertain the scope of each statute before it was possible to see if a state law “related to” the federal provision for purposes of preemption. In contrast, in Mellouli, the same careful atten- tion to legislative context caused the Court to require a tighter relation between the federal law and the state law. Mellouli concerned the use of state convictions in immi- gration cases. In 2010, Moones Mellouli, a lawful permanent resident of the United States, pleaded guilty to a misde- meanor offense under Kansas law for the possession of drug paraphernalia to store a controlled substance. The “paraphernalia” was a sock, where he had put four Ad- derall pills, which contain a mixture of amphetamine and dextroamphetamine. Adderall is used to treat narcolepsy No. 21-3225 37

and ADHD, among other things. The state court imposed a suspended term of 359 days and 12 months’ probation. After Mellouli successfully completed his probation, he was arrested by agents of Immigration and Customs En- forcement on the ground that the state misdemeanor ren- dered him removable pursuant to 8 U.S.C. § 1227(a)(2)(B)(i). Using words that should by now be familiar to us, that stat- ute authorizes the removal of a foreigner “convicted of a violation of … any law … of a State … relating to a con- trolled substance (as defined in section 802 of Title 21).” Id. The question was whether Mellouli’s Kansas conviction fit that definition. No, was the Court’s answer. It summarized its reason for that holding as follows: We hold that Mellouli’s Kansas conviction for con- cealing unnamed pills in his sock did not trigger re- moval under § 1227(a)(2)(B)(i). The drug- paraphernalia possession law under which he was convicted, Kan. Stat. Ann. § 21–5709(b), by definition, related to a controlled substance: The Kansas statute made it unlawful “to use or possess with intent to use any drug paraphernalia to ... store [or] conceal ... a controlled substance.” But it was immaterial under [the Kansas] law whether the substance was defined in 21 U.S.C. § 802. Nor did the State charge, or seek to prove, that Mellouli possessed a substance on the § 802 schedules. Federal law (§ 1227(a)(2)(B)(i)), therefore, did not authorize Mellouli's removal. 575 U.S. at 801 (emphasis in original). Note that the Court freely recognized in this passage that the law underlying Mellouli’s Kansas conviction “relat- 38 No. 21-3225

ed to” a controlled substance. But that was not enough to support the use of that conviction for removal, for the sim- ple reason that the removal statute included an additional requirement: the only controlled substances that could be used as a reference point were those defined in 21 U.S.C. § 802(6), which in turn refers to schedules I, II, III, IV, and V of part B of that subchapter. Taking the categorical approach, the Court noted that Kansas defines “controlled substances” more broadly than the federal government does; the Kansas definition includes “at least nine substances not included in the federal lists.” 575 U.S. at 802. The government argued, however, that the use of the words “relating to” in the immigration statute rendered that mismatch irrelevant. The Kansas drug mis- demeanor of which Mellouli was convicted at least related to the federal controlled substance laws, and that (the gov- ernment contended) was enough. Not so, the Court responded. The majority accepted that “the last reasonable referent of ‘relating to,’ as those words appear in § 1227(a)(2)(B)(i), is ‘law or regulation.’” 575 U.S. at 811. Only the dissenting Justices accepted the argument that the words “relating to” are so broad that they covered this situation. The Court refused to go that far, warning in- stead that “those words, extended to the furthest stretch of their indeterminacy, stop nowhere.” Id. at 812 (cleaned up). “Context,” the majority continued, “may tug in favor of a narrower reading.” Id. (cleaned up). Both the historical background of § 1227(a)(2)(B)(i) and the language of the statute indicated to the Court that there had to be a “direct link between an alien’s crime of conviction” and a federally defined controlled substance. Id. at 812. The statute itself No. 21-3225 39

specified which controlled substances could be used for the immigration purpose at hand. Concluding, the Court held that “to trigger removal under § 1227(a)(2)(B)(i), the Gov- ernment must connect an element of the alien’s conviction to a drug ‘defined in [§ 802].’” Id. at 813. C That moves us closer to Liestman’s case. It is common ground between the majority and me that his Wisconsin conviction could not be used to enhance his federal sen- tence if we were to use the “generic crime” version of the categorical approach. Wisconsin’s statute covers more body parts—that is, more activity—than the federal statute does, and so it fails that test. But what about the “conduct” ap- proach? And how should the words “relating to” affect our analysis? Those two questions already have come up before this court. The first case I examine is United States v. Kraemer, 933 F.3d 675 (7th Cir. 2019), which was handed down on July 31, 2019. The facts in Kraemer were almost a carbon copy of those in Liestman’s case. After being charged with five counts of distribution of child pornography in violation of 18 U.S.C. § 2252(a)(2) and one count of possession in viola- tion 18 U.S.C. § 2252(a)(4)(B), Kraemer pleaded guilty to the possession count. 933 F.3d at 677. The government then dismissed the five distribution counts. At the sentencing stage, Kraemer’s two earlier convic- tions under Wisconsin law for first-degree sexual assault of a child became relevant. See Wis. Stat. § 948.02(1), (2) (1995). Citing 18 U.S.C. § 2252(b)(2) (which is functionally identical to (b)(1), the statute involved in Liestman’s case), the Proba- 40 No. 21-3225

tion Office advised that Kraemer’s Wisconsin convictions supported an enhancement because they “related to” the sexual abuse of a minor. But there was a wrinkle. Although Kraemer’s crime of conviction is located in chapter 110 of Title 18, that chapter did not contain any definition of the term “sexual abuse.” Lacking statutory guidance, the dis- trict court turned to the definition of sexual abuse in 18 U.S.C. § 2246(3), which specifically says that it provides def- initions for terms “[a]s used in this chapter [109A].” The court found four offenses that used the section 2246(3) defi- nition: 18 U.S.C. §§ 2241, 2242, 2243, and 2244. 933 F.3d at 677. Kraemer’s Wisconsin offenses (from a categorical per- spective, of course) covered more conduct than that identi- fied in sections 2241, 2243, and 2244. Id. 677–78. This left section 2242, which prohibits knowingly “engag[ing] in a sexual act with another person if that other person is … in- capable of appraising the nature of the conduct.” See id. at 678 (quoting 18 U.S.C. § 2242(2)). Finding this to be a cate- gorical match with Wis. Stat. § 948.02(2), on the theory that a person whose age is below the age of consent might be incapable of appreciating the nature of the act, the district court found the enhancement to be proper. Id. We upheld that decision, id. at 685, but our rationale dif- fered subtly from that of the district court. We emphasized that the prior conviction supporting an enhancement need- ed only to “relate to” aggravated sexual abuse, id. at 679, not to be “absolute[ly] congruen[t]” with the federal law, id. at 680. Importantly, there was no statutory definition bind- ing us, since chapter 110 (unlike chapter 109A) has no defi- nition of the terms “aggravated sexual abuse,” “sexual abuse,” or “abusive sexual conduct.” In that setting, we concluded that the words “relate to” have their default No. 21-3225 41

broad meaning. Id. at 679 (citing Mellouli, 504 U.S. at 383). We distinguished Mellouli on the ground that it turned on “the particular removal statute’s surrounding text and his- tory.” 933 F.3d at 681. We could find no comparable history or text that informed the statutes at issue in Kraemer. Id. at 682. On the understanding that there was no contrary direc- tion from Congress, we felt free to conclude that a “slight difference in the maximum age of the victim” (Wisconsin used age 13, while the federal statute used age 12) did not prevent the state crime from being “related to” the federal offense, id. at 684, and so we were relieved of the need to use the normal categorical approach. Just three months later, we returned to this subject in United States v. Kaufmann, 940 F.3d 377 (7th Cir. 2019). Although Kaufmann purported to follow Kraemer’s reasoning, id. at 381 (“[W]e adhere to our Kraemer decision today”), a closer look reveals that it missed a critical point. Like Kraemer, Kaufmann had pleaded guilty to child pornography offenses—in Kaufmann’s case both receiving and possessing materials involving sexual exploitation of a minor, in violation of 18 U.S.C. §§ 2252(a)(2) and (a)(4). Rather than looking at the governing statutes, we assumed again that the “relating to” language in section 2252(b) exempted us from the need to conduct any comparison between the federal and the state laws. It was enough, we thought, that there was “at least substantial overlap in content” between the two. 940 F.3d at 380. That, we held, was enough to permit the enhancement to the federal law. Id. What we missed in Kaufmann was the importance of a statutory definition. Although there was no governing stat- 42 No. 21-3225

utory definition in Kraemer, the opposite was true in Kauf- mann. The key term in Kaufmann was “child pornography,” not “sexual abuse” or one of its variants. The term “child pornography” is defined in 18 U.S.C. § 2256(8), 1 which con- tains the definitions “[f]or the purposes of this chapter [110].” Mellouli holds that when a statutory definition such as this one exists, that is what we must use in carrying out the categorical approach. There was an argument that the Indiana statute under which Kaufmann had been convicted swept more broadly than the statutory definition of “child pornography,” insofar as it covered possession of images that did not depict an actual minor. That was a critical issue; if Kaufmann had been able to demonstrate that such a dif- ference existed (a question on which I take no position), the enhancement would have been improper.

1 The definition reads as follows in its entirety:

(8) “child pornography” means any visual depiction, including any photograph, film, video, picture, or computer or computer-generated image or picture, whether made or produced by electronic, mechanical, or other means, of sexually explicit conduct, where—

(A) the production of such visual depiction involves the use of a minor engaging in sexually explicit conduct; (B) such visual depiction is a digital image, computer im- age, or computer-generated image that is, or is indistinguisha- ble from, that of a minor engaging in sexually explicit conduct; or (C) such visual depiction has been created, adapted, or modified to appear that an identifiable minor is engaging in sexually explicit conduct. No. 21-3225 43

The third case I wish to highlight is United States v. Ruth, 966 F.3d 642 (7th Cir. 2020). Defendant Ruth pleaded guilty to one count of possession of a firearm by a felon, 18 U.S.C. § 922(g), and one count of possession of cocaine with intent to distribute, 21 U.S.C. § 841(a)(1), (b)(1)(C). The govern- ment notified him that it intended to use a 2006 Illinois con- viction for possession of a controlled substance (cocaine) with intent to distribute, 720 ILCS 570/401(c)(2), to enhance his federal sentence. Ruth objected to that enhancement, be- cause the Illinois definition of “controlled substance” is broader than the federal definition. The Controlled Substances Act defines the term “felony drug offense” as “an offense that is punishable by impris- onment for more than one year under any law of the United States or of a State or foreign country that prohibits or re- stricts conduct relating to narcotic drugs, marihuana, anabolic steroids, or depressant or stimulant substances.” 21 U.S.C. § 802(44) (emphasis added). That definition, however, did not play a role in our analysis. Instead, in my opinion cor- rectly, we compared the federal definition of the term “nar- cotic drugs” which includes “[c]ocaine, its salts, optical and geometric isomers, and salts of isomers,” id. § 802(17), with the state definition, which criminalizes not only optical and geometric isomers of cocaine, but also positional isomers. Ruth attached no significance to the use of the phrase “relat- ing to” in the definition of “felony drug offense.” We held, following Shular’s guidance, that the variant of the categori- cal approach that applied was the one that relied on con- duct, that the Illinois statute is categorically broader than the federal law, and thus that the state conviction could not be used to enhance Ruth’s sentence. 966 F.3d at 654. 44 No. 21-3225

I take several lessons from all of these cases. First, the Supreme Court has steadfastly adhered to the categorical approach, and so any step we take must be consistent with it. The only thing that Shular did was to ensure that courts did not think, after Taylor, that it is always necessary to con- jure up a generic offense. Second, whether federal and state laws can be compared depends on the conduct that each covers. Third, when it comes to identifying pertinent con- duct, we must pay close attention to the text of the federal statute under consideration. Sometimes the statute defines the actionable conduct quite broadly, as simply “relating to” airline pricing, controlled-substance offenses, child abuse, or whatever else Congress wishes to use as the point of comparison. Sometimes it has a narrower definition, and sometimes there is no definition at all. Another case on which the majority relies, Pugin v. Garland, 599 U.S. 600 (2023), is an example of the last type of case. In Pugin, the question was whether a state offense “re- lated to” obstruction of justice if there was no pending in- vestigation or proceeding. The term “obstruction of justice” was undefined in the statute, and so the Court turned to definitions appearing in various state laws as a hint to what Congress may have had in mind when it used that term. See id. at 607. The majority does the same here. Ante at 13. But while 50-state surveys may be appropriate in cases involv- ing the generic categorical approach where Congress leaves a term undefined, they cannot take precedence when Con- gress tethers the comparison to a defined term in federal law. We should presume that Congress knows what it is do- ing, even if it adopts a definition that only a minority of states follow. In Liestman’s case, we need not guess what offenses are reached by the phrase “child pornography,” No. 21-3225 45

because the definition in section 2256 tells us exactly what conduct Congress had in mind. We are not free to ignore that definition even if (as in Mellouli) the phrase “related to” occurs elsewhere in the law. As Mellouli held, we must stay within the definition Congress gave us. III All that remains is to apply this law to Liestman’s case. His crime of conviction is the transportation of child por- nography. As I noted earlier, the term “child pornography” is defined for purposes of chapter 110 in 18 U.S.C. § 2256(8). That is enough, in my view, to bring this case within the scope of the Mellouli rule. The majority refuses to do so, on the ground that the statute in Mellouli contained an express cross-reference to the governing definition in the Controlled Substances Act, but that distinction elevates form over sub- stance. When Congress specifies that a definition applies throughout a certain chapter, it should not also have to in- sert a cross-reference to the defined term every time it aris- es. Here is what section 2252(b)(1), Liestman’s sentencing statute, would look like if we were to insist on such a rule: Whoever violates, or attempts or conspires to vio- late, paragraph (1) … of subsection (a) shall be fined under this title and imprisoned not less than 5 years and not more than 20 years, but if such person has a prior conviction … under the laws of any State relat- ing to aggravated sexual abuse, sexual abuse, or abu- sive sexual conduct involving a minor (as defined in section 2256 of this title) or ward, or the production, possession, receipt, mailing, sale, distribution, ship- ment, or transportation of child pornography (as de- fined in section 2256 of this title), or sex trafficking of 46 No. 21-3225

children, such person shall be fined under this title and imprisoned for not less than 15 years nor more than 40 years. The first line of section 2256 already does the work of the italicized parentheticals. Liestman’s case is thus indistin- guishable from Mellouli. The proper comparison, using the conduct-based cate- gorical approach, is between the conduct defined to be child pornography by the governing federal statute, and the con- duct covered by Liestman’s predicate Wisconsin conviction. We all agree that the Wisconsin statute criminalizes more conduct than the federal statute. That should be enough to require a holding that Liestman’s earlier conviction cannot be used to enhance his sentence in this case. I would remand for resentencing. I therefore respectfully dissent from the holding of the en banc court.