Before JOLLY, WILLETT, and OLDHAM, Circuit Judges. E. GRADY JOLLY, Circuit Judge: The district court sentenced appellant Nonami Palomares to a 120-month “mandatory minimum” sentence for smuggling heroin. She argues the district court erred because 18 U.S.C. § 3553(f ), more commonly referred to as the First Step Act’s “safety valve” provision, exempts drug offenders like Palomares, with sufficiently minor criminal histories from mandatory minimum sentences. The relevant part of the statute states that criminal defendants are eligible for relief only if: the defendant does not have— more than 4 criminal history points, excluding any criminal history points resulting from a 1-point offense, as determined under the sentencing guidelines; a prior 3-point offense, as determined under the sentencing guidelines; and a prior 2-point violent offense, as determined under the sentencing guidelines[.] 18 U.S.C. § 3553(f)(1). Palomares argues that she was eligible for relief because her criminal history only ran afoul of sub-section (B)—she had a prior 3-point offense. Because the statute uses the word “and,” she argues that she would only be ineligible if her criminal history satisfied sub-sections (A), (B), and (C). The Government disagrees, arguing that defendants who run afoul of any one of the three requirements are not entitled to relief.[1] The First Step Act’s structure is perplexing. It opens with a negative prefatory phrase coupled with an em-dash (“does not have—”) followed by a conjunctive list (A, B, and C). But we conclude that the statute’s uncommon structure holds the key to unlocking its meaning. We agree with the Eighth Circuit that Congress’s use of an em-dash following “does not have” is best interpreted to “distribute” that phrase to each following subsection. To be eligible for safety valve relief, a defendant must show that she does not have more than 4 criminal history points, does not have a 3-point offense, and does not have a 2-point violent offense. Because Palomares had a previous 3-point offense, she is ineligible for safety valve relief. We AFFIRM. I. Nonami Palomares pleaded guilty to possession with intent to distribute one kilogram or more of a mixture or substance containing a detectable amount of heroin in violation of 21 U.S.C. § 841(a)(1) and (b)(1)(A) and 18 U.S.C. § 2. This offense carries a 10-year mandatory minimum sentence, with a maximum sentence of life imprisonment. 21 U.S.C. § 841(a)(1), (b)(1)(A); 18 U.S.C. § 2. The Presentence Investigation Report (PSR) calculated the advisory imprisonment range as 135 to 168 months, or if Palomares received a three-point reduction for acceptance of responsibility, 97 to 121 months. But because of the mandatory minimum, the PSR elevated its calculated guideline range to 120 to 121 months. Palomares objected to the PSR, arguing that she was eligible for relief under the safety valve. In particular, she argued that a plain reading of § 3553(f)(1) only requires mandatory minimum sentences for defendants whose history meets all three disqualifying criteria listed in subsections (A)– (C)—not just one. And because only one of the disqualifying criteria applied to her, she argued that she was eligible for relief. The district court overruled her objection. While the district court conceded that there was no controlling authority on this question, it agreed with the Government’s position that any of the disqualifying criteria in § 3553(f)(1) would render a defendant ineligible for safety valve relief. The district court granted Palomares a three-point reduction for acceptance of responsibility, agreed with the PSR’s calculation of the applicable guideline range of 120 to 121 months’ imprisonment, and sentenced Palomares to 120 months of imprisonment. Palomares timely appealed. II. A. We begin, as always, with the text of the statute. See In re DeBerry, 945 F.3d 943, 947 (5th Cir. 2019) (“In matters of statutory interpretation, text is always the alpha.”). But “we do not look at a word or a phrase in isolation. The meaning of a statutory provision ‘is often clarified by the remainder of the statutory scheme ‘” Ramos-Portillo v. Barr, 919 F.3d 955, 960 (5th Cir. 2019) (quoting Util. Air Regul. Grp. v. E.P.A., 573 U.S. 302, 321 (2014)). “We consider the text holistically, accounting for the ‘full text, language as well as punctuation, structure, and subject matter.’” Elgin Nursing & Rehab. Ctr. v. U.S. Dep’t of Health & Hum. Servs., 718 F.3d 488, 494 (5th Cir. 2013) (quoting U.S. Nat’l Bank of Or. v. Indep. Ins. Agents of Am., Inc., 508 U.S. 439, 455 (1993)). The ordinary meaning of “and,” which § 3553(f)(1) uses to join the three subsections, is conjunctive. See, e.g., ANTONIN SCALIA & BRYAN A. GARNER, READING LAW: THE INTERPRETATION OF LEGAL TEXTS 116–25 (2012). “Or” is disjunctive. Conjunctive/disjunctive canon, BLACK’S LAW DICTIONARY (10th ed. 2014) (“[I]n a legal instrument, and joins a conjunctive list to combine items, while or joins a disjunctive list to create alternatives.”). Palomares points to this straightforward linguistic rule and insists that because Congress used the word “and,” the government would need to prove that her criminal history included all the sub-sections, (A), (B), and (C). Or stated differently, because her criminal history only included (B), she is eligible for this sentencing relief. We cannot agree. “Authorities agree that when used as a conjunctive, the word “and” has “a distributive (or several) sense as well as a joint sense.” BRYAN A. GARNER, GARNER’S DICTIONARY OF LEGAL USAGE 639 (3d ed. 2011). That is, the phrase “A and B” could mean “A and B, jointly or severally.” Id. As applied to § 3553(f)(1), a “joint” sense of “and” would mean that a defendant is eligible for relief unless the court finds that he does not jointly have all three elements listed in (A), (B), and (C). The “distributive” sense of the word would mean that the requirement that a defendant “does not have” certain elements of criminal history is distributed across the three subsections, and a defendant is ineligible if he fails any one of the three conditions. United States v. Pulsifer, 39 F.4th 1018, 1021 (8th Cir. 2022). To determine whether “and” is used in a “joint” sense or a “distributive” sense in § 3553(f)(1), we must look to the context of the statute itself. In different words, the words here preceding the em-dash apply to each of the conditions that follow. Section 3553(f)(1) uses an em-dash preceding a list, with each item set off by semi-colons. To be eligible for safety valve relief the defendant must show that she: . . . does not have— more than 4 criminal history points, excluding any criminal history points resulting from a 1-point offense, as determined under the sentencing guidelines; a prior 3-point offense, as determined under the sentencing guidelines; and a prior 2-point violent offense, as determined under the sentencing guidelines[.] 18 U.S.C. § 3553(f)(1)(A)–(C). This structure, utilizing a negative preceding an em-dash followed by a conjunctive list, makes it likely that the phrase “does not have” independently applies to each item in the list (does not have (A), does not have (B), and does not have (C)). See Carroll v. Trump, 498 F. Supp. 3d 422, 433 n.42 (S.D.N.Y. 2020) (“[A]n em dash . . . signif[ies] that the . . . clause” that immediately precedes the dash “applies to all . . . of the [items] that follow.” (citing Act of June 25, 1948, ch. 646, 62 Stat. 982 (1948))), rev’d in part & vacated in part, 49 F.4th 759 (2d Cir. 2022). Read in this way, § 3553(f)(1) serves as an “eligibility checklist” for defendants who seek to avail themselves of the First Step Act’s safety valve relief. Pulsifer, 39 F.4th at 1022. Suppose for example that you were about to enter a baseball stadium and you saw a sign that read: To enter the stadium, you must not have— a weapon; any food; and any drink. Readers would quickly understand that the phrase “must not have—” independently modifies each item in the list and thus creates a checklist of prohibited items. No baseball fan would insist that they could enter the stadium with a weapon just because they didn’t have food or a drink. i. Such a natural reading was rejected by the Lopez majority. This distributive approach was described as “far-fetched and quixotic” for two reasons. Lopez, 998 F.3d at 441 n.11. First, it noted that no Ninth Circuit precedents had ever endorsed the distributive approach. Id. But neither did it cite a case rejecting it. Nor can we find one. The statute’s unusual and grammatically difficult structure (a negative followed by an em-dash introducing a list of items set off with semi-colons joined by “and”) is not common. The absence of any authority cuts neither way. Second, Lopez rejected the distributive approach because it reasoned that, applied consistently, it “would destroy the entire safety-valve structure and allow a defendant to receive safety valve relief if he or she met the criteria in § 3553(f)(1), § 3553(f)(2), § 3553(f)(3), § 3553(f)(4), or § 3553(f)(5).” Id. (emphasis in original). But this conclusion does not follow. The distributive reading cannot affect the rest of the statute because the list in § 3553(f)(1) works differently due to its negative clause “does not have” that precedes an em-dash. By contrast, § 3553(f) contains a list of affirmative requirements. Only reading “and” to mean “or” would imperil the rest of the safety valve. Additionally, Palomares challenges this distributive approach as being inconsistent with our holding in Modica v. Taylor, where we considered the FMLA’s definition of “employer.” 465 F.3d 174, 183–88 (5th Cir. 2006). But that case is inapposite. Modica was concerned with whether one sub-section in a conjunctive list modified another sub-section. Id. The court concluded it did. Id. But here we must decide how a higher-level provision applies to each of its sub-parts, not how the sub-parts modify each other. Moreover, the phrase preceding the em-dash in Modica did not include a negative (like “not”). Although we agree with Modica that the use of an em-dash “suggests that there is some relationship between the [sub-sections,]” that tells us little about what exactly that relationship is here. Id. B. The distributive meaning of “and” such that “does not have” independently applies to each item in the list (does not have (A), does not have (B), and does not have (C)) is the preferred interpretation because it avoids violating the canon against surplusage. The canon against surplusage is the interpretive principal that courts prefer interpretations that give independent legal effect to every word and clause in a statute. Williams v. Taylor, 529 U.S. 62, 404 (2000) (stating that it is “a cardinal principle of statutory construction that we must give effect, if possible, to every clause and word of a statute.” (quoting United States v. Menasche, 348 U.S. 528, 538–39 (1955)) (cleaned up)); Latiolais v. Huntington Ingalls, Inc., 951 F.3d 286, 294 (5th Cir. 2020) (en banc) (“[T]he canon against surplusage . . . expresses courts’ ‘general “reluctan[ce] to treat statutory terms as surplusage.”‘” (quoting Bd. of Trs. of Leland Stanford Junior Univ. v. Roche Molecular Sys., Inc., 563 U.S. 776, 788 (2011))). As discussed supra, to be eligible for safety valve relief the defendant must show that she: . . . does not have— more than 4 criminal history points, excluding any criminal history points resulting from a 1-point offense, as determined under the sentencing guidelines; a prior 3-point offense, as determined under the sentencing guidelines; and a prior 2-point violent offense, as determined under the sentencing guidelines[.] 18 U.S.C. § 3553(f)(1)(A)–(C). If we accepted Palomares’s reading of § 3553(f)(1), that she would only be ineligible if her criminal history satisfied sub-sections (A), (B), and (C), subsection 3553(f)(1)(A) would be surplusage because every criminal defendant who has a 2-point violent offense and a 3-point offense (satisfying (B) and (C)) will have at least 5 criminal history points, satisfying (A). As a result, we could strike out (A) without changing § 3553(f)(1)’s legal effect. Put simply: 3 + 2 > 4. Nonetheless, Palomares offers two arguments why subsection (A) is not surplusage. Neither are availing. i. Palomares’s first argument is the argument endorsed by the Ninth Circuit in Lopez. Lopez reasoned that the surplusage problem can be avoided by supposing that a single 3-point violent offense could satisfy subsections (B) and (C) simultaneously. It cited legislative history indicating that 2-point offenses are defined as those carrying a sentence of “at least 60 days”— which would include both 2-and 3-point offenses. Lopez, 998 F.3d at 440 n.10. This approach is attractive because it would give Congress’s choice of the word “and” its ordinary meaning, without rendering subsection (A) surplusage. But as Judge Smith noted in his concurrence in Lopez, this approach violates the plain wording of § 3553(f)(1)(C). Id. at 444–47 (Smith, J., concurring in part, dissenting in part, and concurring in the judgment). Subsection (C) incorporates the sentencing guidelines’ definition of a “2-point violent offense.” 18 U.S.C. § 3553(f )(1)(C) (“[T]he court shall impose a sentence . . . without regard to any statutory minimum sentence, if . . . [inter alia] the defendant does not have . . . a prior 2-point violent offense, as determined under the sentencing guidelines.” (emphasis added)). And § 4A1.1 of the Sentencing Guidelines defines 2-and 3-point offenses in mutually exclusive terms. U.S.S.G. § 4A1.1(b); see also Lopez, 998 F.3d at 444–47 (Smith, J., concurring in part, dissenting in part, and concurring in the judgment). Courts add 3 points for sentences exceeding 13 months, 2 points for sentences between 60 days and 13 months, and 1 point for sentences less than 60 days. U.S.S.G. § 4A1.1(a)–(c). So a 3-point violent offense could not satisfy subsection (C) because by definition its sentence was more than 13 months—not between 60 days and 13 months. As Judge Smith succinctly put it: “Two points is two points. Two points is not three points.” Lopez, 998 F.3d at 445 (Smith, J., concurring in part, dissenting in part, and concurring in the judgment). The Lopez majority’s first response to this argument is unpersuasive. The majority noted that the sentencing guidelines were designed to “add” criminal history points together as a part of a calculation. Id. at 440 n.10 (majority opinion). The majority reasoned that context compels us to count 3-point violations only once, because counting a 3-point violation as both a 3-and 2-point offense “would overstate a defendant’s criminal history and cause an inflated Guidelines range.” Id. But “[h]ere, in the safety-valve context, we are not ‘adding’ criminal-history points to form a Guidelines calculation.” Id. Because Congress must have meant to target “more serious violent offenses (three-point violent offenses)” along with less serious ones (two-point offenses), the Lopez majority reasoned it makes little sense to rely on the Sentencing Guidelines’ definitions of those terms. Id. Except that is what Congress plainly did. Sensibly or not, § 3553(f)(1)(C) explicitly incorporates the Sentencing Guidelines by reference. Sub-section (C) is triggered by “a prior 2-point violent offense, as determined under the sentencing guidelines.” See 18 U.S.C. § 3553(f)(1)(C) (emphasis added). While Lopez‘s reliance on the canon against surplusage is understandable, “such interpretative canon[s are] not a license for the judiciary to rewrite language enacted by the legislature.” Asadi v. G.E. Energy (USA), L.L.C., 720 F.3d 620, 622 (5th Cir. 2013) (alteration in original) (quoting United States v. Monsanto, 491 U.S. 600, 611 (1989)). The Lopez majority’s second response to the unambiguous text was to rely on legislative history. Lopez, 998 F.3d at 440 n.10 (citing COMM. ON THE JUDICIARY, 115TH CONGRESS, THE REVISED FIRST STEP ACT OF 2018 (S.3649)). But, the Supreme Court has repeatedly and emphatically rejected the use of legislative history where the text is unambiguous. See, e.g., Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1750 (2020) (“[L]egislative history can never defeat unambiguous statutory text “). So have we. See, e.g., Franco v. Mabe Trucking Co., 3 F.4th 788, 795 (5th Cir. 2021) (“[N]o amount of legislative history can defeat unambiguous statutory text . . . .” (citing Bostock, 140 S. Ct. at 1750)). And, so did the Lopez majority—notably, in a part of its opinion where the legislative history undermined its interpretation. See Lopez, 998 F.3d at 442 (“Because § 3553(f)(1)’s ‘and’ is not ambiguous, we need not consult legislative history.” (citing Food Mktg. Inst. v. Argus Leader Media, 139 S. Ct. 2356, 2364 (2019)). In fact, the Lopez majority rejected the exact same piece of legislative history where it indicated that Congress meant to use “or” instead of “and.” See id. (discussing a Senate Judiciary Committee summary of a prior version of § 3553(f)(1) saying that “[o]ffenders with prior ’3 point’ felony convictions . . . or prior ’2 point’ violent offenses . . . will not be eligible for the safety valve” (quoting COMM. ON THE JUDICIARY, 115TH CONGRESS, THE REVISED FIRST STEP ACT OF 2018 (S.3649)). The Lopez majority was right the second time. Id. (citing Food Mktg. Inst., 139 S. Ct. at 2364). ii. That leads us to Palomares’s second argument: Simply admit that sub-section (A) is surplusage. That is the approach the Lopez majority endorsed as a fallback option, and which the concurrence defended. As both those opinions note, the canon against surplusage is not absolute. See Lopez, 998 F.3d at 441 (first citing Conn. Nat’l Bank v. Germain, 503 U.S. 249, 253 (1992); and then citing Chickasaw Nation v. United States, 534 U.S. 84, 94 (2001)); id. at 446 (Smith, J., concurring in part, dissenting in part, and concurring in the judgment). “[O]ur hesitancy to construe statutes to render language superfluous does not require us to avoid surplusage at all costs.” United States v. Atl. Rsch. Corp., 551 U.S. 128, 137 (2007). The Lopez concurrence was on firm ground in concluding that it is better to admit surplusage than to rewrite the statute. See id. at 137 (“It is appropriate to tolerate a degree of surplusage rather than adopt a textually dubious construction “). But here, that is unnecessary because the distributive approach gives “and” its ordinary conjunctive meaning without rendering sub-section (A) meaningless. Palomares’s second argument would only be persuasive if we had no other option. To sum up, the “distributive approach” is the most natural and indeed the most likely intent behind Congress’s choice of a unique structure for § 3553(f)(1). That structure is understandably read as an eligibility checklist of conditions that the defendant cannot possess to be eligible for safety valve relief. The alternative readings are implausible because they each fail to account for the plain meaning of the statute in some way.[2] C. Finally, Palomares argues that the Court should apply the rule of lenity in applying § 3553(f)(1). The rule of lenity “requires ambiguous criminal laws to be interpreted in favor of the defendants subjected to them.” United States v. Bittner, 19 F.4th 734, 748 (5th Cir. 2021) (quoting United States v. Santos, 553 U.S. 507, 514 (2008) (plurality opinion)), cert. granted, 142 S. Ct. 2833 (2022). But courts only apply the rule of lenity when faced with a “grievous ambiguity or uncertainty.” See Maracich v. Spears, 570 U.S. 48, 76 (2013) (emphasis added) (quoting Barber v. Thomas, 560 U.S. 474, 488 (2010)). And an ambiguity is only grievous if it remains after the court considers the statute’s “text, structure, history, and purpose,” id., including all the “traditional canons of statutory construction,” Shular v. United States, 140 S. Ct. 779, 787 (2020) (quoting United States v. Shabani, 513 U.S. 10, 17 (1994)). Such a conclusion is not true of § 3553(f)(1). After studying the text and structure of the statute, as informed by the various canons of construction, one approach stands prominently above the other interpretations. Because we need not “guess” at the statute’s meaning, the rule of lenity does not apply. See Maracich, 570 U.S. at 76 (“[T]he rule of lenity only applies if, after considering text, structure, history, and purpose, there remains a grievous ambiguity or uncertainty in the statute such that the Court must simply guess as to what Congress intended.” (quoting Barber, 560 U.S. at 488)). IV. We hold that the phrase “does not have” independently applies to each subsection in 18 U.S.C. § 3553(f)(1), rendering criminal defendants ineligible for safety valve relief if they run afoul of any one of its requirements. Because Palomares has a prior 3-point offense, we AFFIRM. ANDREW S. OLDHAM, Circuit Judge, concurring in the judgment: In my view, Nonami Palomares was ineligible for the safety valve codified in 18 U.S.C. § 3553(f). That’s because she had a prior 3-point offense and hence could not satisfy § 3553(f)(1)(B). I write separately to make two points. The first is a general point about textualism. The second is a specific point about § 3553(f)’s text. I. I am an ardent textualist. As I’ve said many times, “[i]n matters of statutory interpretation, text is always the alpha.” In re DeBerry, 945 F.3d 943, 947 (5th Cir. 2019); see also ibid. (noting “it’s also the omega”); Cochran v. SEC, 20 F.4th 194, 214 (5th Cir. 2021) (en banc) (Oldham, J., concurring) (“First, as should go without saying by now, our inquiry begins with the statutory text, and ends there as well if the text is unambiguous.” (quotation omitted)); United States v. Koutsostamatis, 956 F.3d 301, 306 (5th Cir. 2020) (“In statutory interpretation, we have three obligations: ‘(1) Read the statute; (2) read the statute; (3) read the statute!’” (quoting HENRY J. FRIENDLY, BENCHMARKS 202 (1967))); Djie v. Garland, 39 F.4th 280, 285 (5th Cir. 2022) (“When a regulation attempts to override statutory text, the regulation loses every time—regulations can’t punch holes in the rules Congress has laid down.”); Hoyt v. Lane Constr. Corp., 927 F.3d 287, 295 (5th Cir. 2019) (“We start with the only easy part—the statutory text.”); Heinze Tesco Corp., 971 F.3d 475, 484 (5th Cir. 2020) (“Of course, when a statute’s text is clear, courts should not resort to legislative history.” (quotation omitted)); United States v. Graves, 908 F.3d 137, 141 (5th Cir. 2018) (“We start, of course, with the statutory text.” (quotation omitted)); United States ex rel. Drummond v. BestCare Lab’y Servs., LLC, 950 F.3d 277, 281 (5th Cir. 2020) (“The statutory text is what matters “). And as a textualist, I subscribe to Justice Scalia’s understanding of textualism: “In their full context, words mean what they conveyed to reasonable people at the time they were written—with the understanding that general terms may embrace later technological innovations.” Antonin SCALIA & BRYAN A. GARNER, READING LAW: THE INTERPRETATION OF LEGAL TEXTS 16 (2012); see also ibid. (“Textualism, in its purest form, begins and ends with what the text says and fairly implies.”). I also agree with him that one of the virtues of careful adherence to text is that “most interpretive questions have a right answer.” Id. at 6. But I’ve never understood textualism to mean hyper-literalism. As my law school mentor often said, “textualists . . . are not literalists.” John F. Manning, Textualism and the Equity of the Statute, 101 COLUM. L. REV. 1, 108 (2001); see also John F. Manning, The Absurdity Doctrine, 116 HARV. L. REV. 2387, 2392–93 (2003) (“Even the strictest modern textualists properly emphasize that language is a social construct. They ask how a reasonable person, conversant with the relevant social and linguistic conventions, would read the text in context.”). Justice Scalia said the same thing. See SCALIA & GARNER, supra, at 356 (“Adhering to the fair meaning of the text (the textualist’s touchstone) does not limit one to the hyperliteral meaning of each word in the text.”). Hyper-literalism is bad for two reasons. The first is perhaps the most obvious: hyper-literalism is bad textualism. A statute’s “text may never be taken out of context.” Graves, 908 F.3d at 142. That’s because “words are given meaning by their context, and context includes the purpose of the text.” SCALIA & GARNER, supra, at 56. As Justice Scalia once quipped, without context, we could not tell whether the word draft meant a bank note or a breeze. See ibid. Such nuance is lost on the hyper-literalist. The second indictment of hyper-literalism is perhaps more subtle: it opens textualism to the very criticism that necessitated textualism in the first place. In one of the most influential law review articles ever written, Karl Llewellyn denigrated the late nineteenth century “Formal Period,” in which “statutes tended to be limited or even eviscerated by wooden and literal reading, in a sort of long-drawn battle between a balky, stiff-necked, wrong-headed court and a legislature which had only words with which to drive that court.” Karl N. Llewellyn, Remarks on the Theory of Appellate Decision and the Rules or Canons About How Statutes Are to Be Construed, 3 VAND. L. REV. 395, 400 (1950). That criticism—and Llewellyn’s famous indictment of the canons of construction, see id. at 401–06—”largely persuaded two generations of academics that the canons of construction were not to be taken seriously.” John F. Manning, Legal Realism & the Canons’ Revival, 5 GREEN BAG 2D 283 (2002). It’s remarkable how much of modern textualism aims to overcome Llewellyn’s criticisms 72 years later. See, e.g., SCALIA & GARNER, supra, at 59–62 (devoting a chapter to the project). While Llewellyn and his fellow legal realists were wrong about an awful lot, one of his arguments merits continued attention: wooden and hyper-literal textualism, Llewellyn argued, generates a “foolish pretense” that there’s “only one single correct answer possible” in every single statutory-interpretation dispute. Llewellyn, supra, at 399. Again, it’s true that, under careful textualist analysis, “most interpretive questions have a right answer.” SCALIA & GARNER, supra, at 6 (emphasis added). But not all of them. Some textualist inquiries generate a range of potentially right answers, and it’s the judge’s job to pick the best one and explain it. We do the law a disservice when we suggest that textualist exegeses are reducible to math problems, logic puzzles, or hyper-literalist readings of the word and. II. In my view, the proper interpretation of § 3553(f)(1) does not hinge on an inferential rule of Boolean algebra called “De Morgan’s Theorem.” Nor does it hinge on a hyper-literalist interpretation of the conjunction “and.” It hinges instead on a context-sensitive interpretation of § 3553(f) as a whole. Let’s start, as always, with the statutory text. Subsection (f) creates a “safety valve” by eliminating mandatory minimums for certain drug offenses: if the court finds at sentencing . . . that— the defendant does not have— more than 4 criminal history points, excluding any criminal history points resulting from a 1-point offense, as determined under the sentencing guidelines; a prior 3-point offense, as determined under the sentencing guidelines; and a prior 2-point violent offense, as determined under the sentencing guidelines; the defendant did not use violence or credible threats of violence or possess a firearm or other dangerous weapon (or induce another participant to do so) in connection with the offense; the offense did not result in death or serious bodily injury to any person; the defendant was not an organizer, leader, manager or supervisor of others in the offense, as determined under the sentencing guidelines and was not engaged in a continuing criminal enterprise, as defined in section 408 of the Controlled Substances Act; and [before] the time of the sentencing hearing, the defendant has truthfully provided to the Government all information and evidence the defendant has concerning the offense. 18 U.S.C. § 3553(f). Subsection (f) thus has two different distributive clauses and two different conjunctive lists. Conjunctive List 1. The first conjunctive list provides safety-valve relief if “the court finds” that five conditions are satisfied. Those five conditions are separated into five statutory paragraphs, numbered (f)(1) through (f)(5). The five paragraphs follow an em dash, are separated by semicolons, and are written in a conjunctive list (1; 2; 3; 4; and 5). The affirmative prefatory clause in Conjunctive List 1 distributes to each of the five paragraphs. Thus, for a defendant to get relief under § 3553(f), the court needs to “find at sentencing that” (1) is met, “find at sentencing that” (2) is met, “find at sentencing that” (3) is met, “find at sentencing that” (4) is met, and “find at sentencing that” (5) is met. Conjunctive List 2. Within paragraph (1), there is also a second nested list. This one has three items. Those three items are separated into three statutory “subparagraphs” labeled (A) through (C). The three subparagraphs also follow an em dash, are separated by semicolons, and are presented in a conjunctive list (A; B; and C). Conjunctive List 1 and Conjunctive List 2 are structurally identical. Both are introduced by a prefatory clause followed by an em dash. Both offer a multi-item list, separated by semicolons, and conjoined with “and.” The only difference between them is that Conjunctive List 1 has an affirmative prefatory clause (“if the court finds at sentencing . . . that”) whereas Conjunctive List 2 has a negative prefatory clause (“the defendant does not have”). I don’t see why that difference matters. The text and structure of both Conjunctive List 1 and Conjunctive List 2 indicate that the language preceding the em dashes distributes throughout the statutory sentence. And hence subsection (f)(1) reads, in effect: The safety valve applies, and hence a mandatory minimum sentence does not, (1)(A) if the court finds at sentencing that the defendant does not have more than 4 criminal history points, excluding any criminal history points resulting from a 1-point offense, as determined under the sentencing guidelines; (1)(B) if the court finds at sentencing that the defendant does not have a prior 3-point offense, as determined under the sentencing guidelines; and (1)(C) if the court finds at sentencing that the defendant does not have a prior 2-point violent offense, as determined under the sentencing guidelines.[3] The “if the court finds at sentencing” language from the umbrella clause of (f) distributes throughout, just as the “defendant does not have” language from the umbrella clause of (f)(1) distributes throughout. That double distribution of the prefatory clauses creates a cogent statutory sentence.[4] And none of it hinges on symbolic logic, Boolean algebra, or whether and means “and” or “or.” It hinges only on the context of the statute as a whole. And a recognition that human beings—and hence members of Congress—do not speak or legislate like computers. * * * As this question continues to percolate through the federal courts, I am sure that the pages of the Federal Reporter and eventually the United States Reports will teem with colorful hypotheticals including and adding to those offered by my esteemed colleagues today. See supra, at 6 (weapons at baseball games); post, at 24–25 (Willett, J., dissenting) (grocery lists, drunk driving, and fire). But all these hypotheticals prove, with greatest respect, is that language is context-dependent. It’s far easier, I’d submit, to approach the statute that Congress actually wrote and to understand it as an ordinary person would. Subsection (f) constitutes one (admittedly long) statutory sentence. And I’d read it from the beginning to the end—distributing the prefatory clauses as Congress wrote them along the way. Perhaps I’m wrong about how best to read § 3553(f). But if I am, it only proves that ordinary English does not beget the sort of epistemic certainty that De Morgan invoked. DON R. WILLETT, Circuit Judge, dissenting: With deepest respect, I dissent. The majority opinion holds that “and” can have either a “distributive” or a “joint” sense and that only context can resolve the inherent ambiguity. Piercing the legalese, the idea is that “and” can mean either “and” or “or.” To be fair, the majority isn’t alone. Courts facing clumsy drafting have sporadically reached that conclusion.[5] The English language is never in stasis. Witness the off-definition misuse of “literally,” which has literally come to mean “figuratively.” But interchanging “and” and “or” is a mistake. “We give our language, and our language-dependent legal system, a body blow when we hold that it is reasonable to read ‘or’ for ‘and’”—or “and” for “or.”[6] Manufactured ambiguity poses a special threat to our language’s elemental particles. How can Congress express its will if everyday words slip into linguistic black holes so dense that settled language rules break down? When judges say that certain words are inherently ambiguous, we beget a self-fulfilling prophecy. And when we use complicated semantic bracework to augment ordinary meaning, we risk creating a negative feedback loop if Congress sees the favor as an invitation rather than a one-off. Congress said that Palomares is ineligible for safety valve relief if her criminal history runs afoul of § 3553(f)(1)(A), (B), and (C). While the First Step Act “is far from a chef d’oeuvre of legislative draftsmanship,”[7] we must assume that Congress meant what it said. Congress said “and.” If it wished to withhold safety valve relief from defendants who failed any one of the three sub-sections, it would have (maybe should have) joined them together with “or.” I would vacate Palomares’s sentence and remand for resentencing. I As the majority concedes, the plain meaning of “and” is conjunctive. Dictionaries and treatises aren’t needed to prove the point, but they uniformly define “and” this way.[8] The definition even lends its name to the “conjunctive/disjunctive canon” of construction.[9] That interpretive rule says what ordinary English speakers already know: When the word “and” joins a list, all the things listed are required.[10] A parent who tells you to pick up milk, eggs, and cheese will rightly be upset if you return with just milk. “And” is still conjunctive when it follows a negative like “not” or “no.” When a negative precedes a conjunctive list, “the listed things are individually permitted but cumulatively prohibited.”[11] “[D]on’t drink and drive” means that you can “do either one, but you can’t do them both.”[12] Thus a concerned parent might tell their child: “Don’t drink or drive.” Expanding the list beyond two items leaves the underlying principle unchanged. “Do not mix heat, fuel, and oxygen” instructs the reader to prevent the unity of all three ingredients unless she wants a fire. A speaker who wishes to individually prohibit each item in a list must use “or.” This common-sense rule travels more stuffily as “De Morgan’s law,”[13] a logical precept which holds that (1) the negation of a conjunction is equivalent to the disjunction of the negations, and (2) the negation of a disjunction is equivalent to the conjunction of the negations.[14] I recite the precept not because this case requires it, but rather to show that “and” is conjunctive at all points along the spectrum from friendly to formal. What this means for § 3553(f)(1) is simple. If Congress wanted “not” to independently modify each item in the list, the proper word was “or.” II The majority agrees with me about the ordinary meaning of “and” but argues that § 3553(f)(1) has additional elements that make all the difference. That section ends its prefatory clause with an em-dash—the wonderfully versatile Swiss Army knife of punctuation marks—and it separates the subsections that follow with line breaks and semi-colons. The majority concludes that the language before the em-dash is “distributed” to independently modify each following subsection, as so: the defendant does not have— [the defendant does not have] more than 4 criminal history points, . . . ; [the defendant does not have] a prior 3-point offense . . . ; and [the defendant does not have] a prior 2-point violent offense [15] This interpretation has some merit. For one, it has the benefit of obscurity. Style guides, dictionaries, books on grammar, and the like are silent on whether putting an em-dash after the negative phrase changes its meaning. Like the majority, I have been unable to find any case construing a statute with a similar structure.[16] It is plausible that Congress’s choice to include subsections (A)–(C) on separate lines set off by semi-colons, rather than a more conventional list set off by commas, is best read as a “checklist” of items that a defendant must comply with to be eligible for safety valve relief. One of our sister circuits recently adopted this interpretation.[17] The problem is that this approach runs afoul of the canon of consistent usage—the principle that “a given term is used to mean the same thing throughout a statute.”[18] Section 3553(f), which the Government dubs the “umbrella clause,” contains an introduction set off by an em-dash just like § 3553(f)(1). The umbrella clause is long, but in essence it says: “The court shall not apply any mandatory minimum sentence if—”. A five-part list follows, again separated by line breaks and semi-colons, and again with an “and” at the end of the list’s penultimate item. If we buy the Government’s argument for the “and” in § 3553(f)(1)(B), then consistency requires us to do the same for the “and” that closes § 3353(f)(4). But treating that “and” as an “or” would tell district courts to disregard mandatory minimums in five separate scenarios—not one scenario consisting of five elements. As a result, the majority’s “distributive” theory—applied consistently—would make it harder for defendants to meet sub-section (f)(1) but would make it far easier for them to qualify for the safety valve in general. In fact, it would effectively eliminate all mandatory minimums for drug crimes. If the majority is right that em-dashes mean everything before them independently modifies what follows, then Palomares should still win. The majority’s response to this conundrum is not convincing. The majority notes that the phrase “does not have” appears before § 3553(f)(1) but not § 3553(f). In more academic terms, § 3553(f)(1) is a negative conjunctive list while § 3553(f) is just an “ordinary” conjunctive list. But why should that matter? Why should an em-dash function one way when it is preceded the word “not,” and another way when it isn’t? Either an em-dash signifies that the preceding language independently modifies each sub- section that follows, or it does not. The majority does not cite a single grammarian, dictionary, or case endorsing its on-again off-again view of em- dashes. Making up new grammatical rules on the fly isn’t statutory interpretation, it’s statutory Calvinball.[19] III The majority’s next argument is that a conjunctive interpretation of “and” would violate the canon against surplusage—the interpretive principal that courts prefer interpretations that give independent legal effect to every word and clause in a statute.[20] The majority reasons that reading “and” conjunctively would result in surplusage because every time subsections (B) and (C) are satisfied, so is (A). Every criminal defendant who has a 2-point violent offense and a 3-point offense (satisfying (B) and (C)) will have at least 5 criminal history points, satisfying (A). This view would allow us to strike out (A) without changing § 3553(f)(1)’s legal effect. I agree that Palomares’s first argument against surplusage—which the majority in Lopez adopted—is not convincing for exactly the reasons set forth in the majority’s opinion.[21] But we asked the parties to brief another argument for why § 3553(f)(1)(A) can retain effect without requiring “and” to forsake its ordinary meaning. Criminal history points are computed in a two-step process. The first step is governed by § 4A1.1 of the Sentencing Guidelines, which tells courts to add three points for prior sentences exceeding 13 months, two points for sentences between 60 days and 13 months, and one point for sentences less than 60 days.[22] The second step is governed by § 4A1.2, which tells courts to not count certain kinds of convictions. For example, § 4A1.2 tells courts to not count 3-point offenses if the defendant was released from incarceration more than fifteen years before committing the instant offense, and to not count 1- or 2-point sentences imposed more than ten years before the instant offense was committed.[23] Section 4A1.2 also instructs courts to never count certain misdemeanors like speeding, loitering, or fish and game violations.[24] Palomares argues that the upshot of this complex system is that subsection (A) is not surplusage because some defendants will have a 2- or 3- point conviction that is ineligible for inclusion in the criminal history calculation. For example, Palomares says that a defendant who completed her sentence for a 3-point drug offense more than 15 years ago, and who committed a 2-point violent offense within the last 10 years, will satisfy § 3553(f)(1)(B) and (C)—she has a prior 3-point offense and a prior 2-point violent offense. But she will not run afoul of subsection (A), because § 4A1.2 tells courts to not count 3-point offenses that have “gone stale.” This hypothetical defendant would satisfy subsections (B) and (C), but not (A). The Government responds that a stale conviction is a 0-point offense, not a 3-point offense. Thus, the Government argues that an offense triggering subsections (B) or (C) always counts for purposes of subsection (A). It notes that § 3553(f)(1) defines 2- and 3-point offenses by reference to § 4A1.1 and § 4A1.2—not just § 4A1.1. A 2- or 3-point offense that is excluded at the second step cannot be used to satisfy any of § 3553(f)(1)’s subsections. The Eighth Circuit recently agreed with similar reasoning.[25] I think Palomares has the better argument. Most readers would not give a 2- or 3-point offense a different name just because it is excluded at the second step. They would not call it a “0-point offense.” Many judges haven’t either.[26] And the legislative history that the Government relies on defines 2-and 3-point offenses in the same way.[27] A stale 3-point offense is still a 3-point offense—it just isn’t counted in the criminal history point calculation.[28] Even if subsection (A) were surplusage, that would not change my view. As Judge Smith noted in his concurrence in Lopez, “our hesitancy to construe statutes to render language superfluous does not require us to avoid surplusage at all costs.”[29] Our task is to follow the plain text of § 3553(f)(1). Congress joined the subsections with “and,” not “or.” The fact that “and” is conjunctive while “or” is disjunctive is one of the elemental aspects of the English language. While I cannot say that the conjunctive/disjunctive canon should always win out over the canon against surplusage, it is a better indication of plain meaning here for at least three reasons. First, ignoring Congress’s choice of the word “and” also violates the canon against surplusage.[30] If the em-dash “distributes” the prefatory clause, then subsections (A)–(C) operate independently regardless of what word appears between them. Under the majority’s logic, that word could be “and,” “or,” or no word at all. But “the canon against superfluity assists only where a competing interpretation gives effect ‘to every clause and word of a statute.’”[31] The majority’s approach fails that test because it ignores the word “and,” and that means the canon against surplusage can do no work. Second, reading “and” out of sub-section (f)(1) violates the canon of consistent usage—but not solely as discussed above. By my count, Congress used “and” to join a list of elements 8 times in this very statute.[32] “Or” joins a list of elements 3 times where Congress wanted to produce the opposite effect.[33] That does not include the countless other uses of “and” and “or” in the same statute that do not join a list of elements where, again, no party disagrees that the words appear in their ordinary sense.[34] By the majority’s logic, we would have to believe that Congress meant to invoke the plain meaning of these words every time except in subsection (f)(1). The majority is 0-2 in complying with the canon for consistent usage. Finally, ignoring the plain meaning of a clearly understood word like “and” is a more obvious and palpable problem than reading part of the statute as redundant. One need only look at the face of the statute to understand that “and” ordinarily means “and,” not “or.” In contrast, the majority’s surplusage argument is apparent only after a reader consults the definitions in the Sentencing Guidelines and pauses to conclude that defendants who trigger subsections (B) and (C) will also always trigger (A). Courts prefer obvious meanings to “ingen[ious]” or elaborate meanings that emerge only after careful reflection.[35] That principle favors tolerating non- obvious surplusage rather than ignoring rudimentary grammar. So it’s not surprising that courts have repeatedly relied on the legislature’s choice of “and” or “or,” even when doing so created some statutory surplusage.[36] IV Because the majority accepts the Government’s plain language argument, it need not consider whether Palomares’s position leads to absurdities. For completeness, I will explain why the Government’s argument on this point is not a feasible fallback. Absurdity arguments face a steep climb. “In statutory interpretation, an absurdity is not mere oddity. The absurdity bar is high, as it should be. The result must be preposterous, one that ‘no reasonable person could intend.’”[37] The result must be so preposterous that it “shock[s] the general moral or common sense.”[38] The Government’s absurdity argument relies on three hypothetical defendants. The first has a “lifetime of serious drug offenses” (but no 2-point violent offense). The second has multiple 3-point offenses (one of which was violent). The third has both a single 2-point violent offense and a 3-point offense. It is absurd, the Government says, that the first two should be eligible for safety valve relief while the third must face mandatory minimums. These hypotheticals do not present any absurdities or demonstrate that Congress could not have meant what it said. The first defendant shows only that drug offenses, standing alone, do not bar a defendant from safety valve relief. Congress’s evident conclusion—only violent drug offenders should receive mandatory minimum sentences—is perfectly rational. The second defendant’s situation is somewhat more troubling. This defendant can use the safety valve even though her offenses are more serious than those of the third defendant, who cannot. That hardly seems fair. But on the other hand, the rule of lenity prevents us from “giv[ing] the text a meaning that is different from its ordinary, accepted meaning, and that disfavors the defendant.”[39] We cannot fix the unfairness that the Government posits by preventing the third defendant from using the safety valve. In other words, the absurdity canon must yield to the rule of lenity. The authoritative case is United States v. Wiltberger.[40] There, Congress had passed a statute criminalizing murder committed “upon the high seas, or in any river, haven, basin or bay.”[41] But the statute criminalized manslaughter only if it was committed on the “high seas.”[42] Chief Justice Marshall, for the Court, conceded that it was “extremely improbable” that Congress meant to ignore shallow-water manslaughter.[43] Nevertheless, the Court was unanimous that it could not enlarge the statute to avoid this apparent absurdity.[44] “To determine that a case is within the intention of a statute, its language must authorise us to say so. It would be dangerous, indeed . . . to punish a crime not enumerated in the statute, because it is of equal atrocity, or of kindred character, with those which are enumerated.”[45] The Government invites us to make that dangerous move here— enlarge the scope of criminal liability because, in some small class of cases, the statute’s plain meaning might generate comparative lenience. The rule of lenity prevents courts from using the absurdity doctrine to that end. And even if I agreed with the Government’s absurdity argument, that would not mean the Government should prevail. After all, the strange conundrum the Government points to comes from the fact that the Sentencing Guidelines define 2- and 3-point offenses to be mutually exclusive. I agree with the majority and the Government that, according to the plain meaning of the Sentencing Guidelines, a 3-point violent offense cannot satisfy both subsections (B) and (C)—even if the Lopez court’s argument to the contrary makes some intuitive sense.[46] But as long as we’re using the absurdity doctrine to rewrite the statute, why not overlook this textual wrinkle rather than Congress’s choice of the word “and”? Both solutions solve the absurdity problem. When choosing between two equally plausible interpretations, “the tie must go to the defendant.”[47] The Government has one final absurdity argument, based on its concern that Palomares’s interpretation would “all but eliminate” the criminal history requirement. Again, I am not persuaded. The Government presents no evidence that it would be rare for defendants to run afoul of all three of conditions in § 3553(f)(1). To the contrary, I would not be surprised to learn that a significant number of career criminals have a 2-point violent offense on their records. And even if the Government’s assertion were true, the absurdity doctrine would still stand in the way. Section 3553(f)(1) is only one of five requirements that a defendant must satisfy to be eligible for safety valve relief. The other four turn on the defendant’s instant offense.[48] Congress could have concluded that access to the safety valve should usually hinge on the instant offense’s severity rather than the defendant’s criminal history. “This is, at minimum, a ‘rational’ policy result.”[49] V We must take Congress at its word: “and.” I respectfully dissent.
