Cargill v. Garland

Before Richman, Chief Judge, and Jones, Smith, Stewart, Dennis, Elrod, Southwick, Haynes, Graves, Higginson, Willett, Ho, Duncan, Engelhardt, Oldham, and Wilson, Circuit Judges. Jennifer Walker Elrod, Circuit Judge, joined by Richman, Chief Judge, and Jones, Smith, Stewart, Southwick, Haynes, Willett, Ho, Duncan, Engelhardt, Oldham, and Wilson, Circuit Judges:* Since the National Firearms Act of 1934, federal law has heavily regulated machineguns. Indeed, as proposed, that law was known to many as “the Anti-Machine Gun Bill.” The possession or transfer of a machinegun was eventually banned through the Gun Control Act of 1968 and the Firearms Owners’ Protection Act of 1986. Today, possession of a machinegun is a federal crime, carrying a penalty of up to ten years’ incarceration. This appeal concerns a regulation promulgated by the federal Bureau of Alcohol, Tobacco, Firearms, and Explosives, purporting to interpret the federal prohibition on machineguns as extending to bump stocks. A bump stock is a firearm attachment that allows a shooter to harness the natural recoil of a semi-automatic weapon to quickly re-engage the trigger after firing, enabling him to shoot at an increased rate of speed. When ATF first considered the type of bump stocks at issue here, it understood that they were not machineguns. ATF maintained this position for over a decade, issuing many interpretation letters to that effect to members of the public. But ATF reversed its longstanding position in 2018,subjectinganyone who possessed a bump stock to criminal liability. ATF reversed its position to a great extent in response to the tragic events that occurred in Las Vegas on October 1, 2017. On that day, a deranged gunman murdered dozens of innocent men and women, and injured hundreds more. To carry out this appalling crime, the gunman used many weapons and utilized many accessories—including bump stocks. Public pressure to ban bump stocks was tremendous. Multiple bills to that effect were introduced in both houses of Congress. But before they could be considered in earnest, ATF published the regulation at issue here, short-circuiting the legislative process. Appellant Michael Cargill surrendered several bump stocks to the Government following publication of the regulation at issue. He now challenges the legality of that regulation, arguing that a bump stock does not fall within the definition of “machinegun” as set forth in federal law, and thus that ATF lacked the authority to issue a regulation purporting to define the term as such. Cargill is correct. A plain reading of the statutory language, paired with close consideration of the mechanics of a semi-automatic firearm, reveals that a bump stock is excluded from the technical definition of “machinegun” set forth in the Gun Control Act and National Firearms Act. But even if that conclusion were incorrect, the rule of lenity would still require us to interpret the statute against imposing criminal liability. A rich legal tradition supports the “well known rule” that “penal laws are to be construed strictly.” United States v. Wiltberger, 18 U.S. (5 Wheat.) 76, 94–95 (1820). As Chief Justice Marshall explained long ago, the rule “is founded on the tenderness of the law for the rights of individuals; and on the plain principle that the power of punishment is vested in the legislative, not in the judicial department. It is the legislature, not the Court, which is to define a crime, and ordain its punishment.” Id. at 95. The Government’s regulation violates these principles. As an initial matter, it purports to allow ATF—rather than Congress—to set forth the scope of criminal prohibitions. Indeed, the Government would outlaw bump stocks by administrative fiat even though the very same agency routinely interpreted the ban on machineguns as not applying to the type of bump stocks at issue here. Nor can we say that the statutory definition unambiguously supports the Government’s interpretation. As noted above, we conclude that it unambiguously does not. But even if we are wrong, the statute is at least ambiguous in this regard. And if the statute is ambiguous, Congress must cure that ambiguity, not the federal courts. The definition of “machinegun” as set forth in the National Firearms Act and Gun Control Act does not apply to bump stocks. And if there were any doubt as to this conclusion, we conclude that the statutory definition is ambiguous, at the very least. The rule of lenity therefore compels us to construe the statute in Cargill’s favor. Either way, we must REVERSE. I A The Gun Control Act of 1968 provides that “it shall be unlawful for any person to transfer or possess a machinegun.” 18 U.S.C. § 922(o)(1). The Act defines machinegun as follows: The term “machinegun” means any weapon which shoots, is designed to shoot, or can be readily restored to shoot, automatically more than one shot, without manual reloading, by a single function of the trigger. The term shall also include the frame or receiver of any such weapon, any part designed and intended solely and exclusively, or combination of parts designed and intended, for use in converting a weapon into a machinegun, and any combination of parts from which a machinegun can be assembled if such parts are in the possession or under the control of a person. 26 U.S.C. § 5845(b); see 18 U.S.C. § 921(a)(24) (incorporating the definition in the National Firearms Act). The traditional example of a machinegun[1] is a rifle capable of automatic fire, like the M-16. Semiautomatic rifles like the AR-15 are not machineguns. See Hollins v. Lynch, 827 F.3d 436, 440 n.2 (5th Cir. 2016) (The M-16 “is capable of automatic fire, that is to say, firing more than one round per trigger-action. . . . The AR-15 is essentially a semi-automatic version of the M-16, that is to say, it fires only one round per trigger-action.”). B To understand what a machinegun is, it is helpful to understand what a machinegun is not. To that end, the firing mechanism of a semi-automatic weapon is especially important. The relevant parts are as follows: The trigger is the interface between the gun’s internal mechanism and the human finger. The sear is the trigger’s top-forward geometric plane, which locks snugly into a groove near the spring of the hammer. The hammer is the spring-loaded element that strikes the firing pin, causing ignition of the charge and propulsion of the bullet. The disconnector is a part that sits on top of the trigger and serves to reset the hammer after a round is fired; this resetting is what makes a semi-automatic weapon semi-automatic. The mechanics of the firing process are as follows. First, the user pulls the trigger. Doing so disengages the hammer from the sear, allowing the spring to swing the hammer to strike the firing pin, which causes the charge to combust and propel the bullet. The firing of the bullet thrusts the bolt backward, which kicks the hammer into the disconnector on top of the still-depressed trigger. When the trigger is reset, the hammer is pulled back into the cocked position and secured by the trigger’s sear as it slips off the disconnector. The user may then fire again by pulling the trigger, without having to manually re-cock the hammer. The mechanics are viewed below:[2] The end result is that the user of a semi-automatic firearm can fire rapidly by means of repeated use of the trigger. Critically, use of the trigger necessarily corresponds one-to-one with bullets fired. That is, a single pull of the trigger results in a single bullet fired. Without resetting the trigger, the disconnector cannot reset the hammer to the fully cocked position. And unless the hammer is fully cocked, it will not be able to strike the firing pin with sufficient force to discharge the weapon a second time. (When a weapon fails to fire for this reason, it is said to experience a “hammer follow” mal-function.) In sum, both the hammer and the trigger-disconnector must in-variably return full circle before another round can be dispatched. This process may be contrasted with a fully automatic gun, which is equipped with something called an “auto sear”—a device that serves to re-cock and release the hammer in tandem with the motion of the bolt for so long as the trigger remains depressed. In other words, the auto sear enables a pendulum swing of the hammer in sync with the bolt without any further input from the user; with one pull of the trigger, an automatic weapon can shoot continuously until ammunition is depleted. C The statutory definition of a machinegun also includes devices that convert an ordinary firearm into a machinegun. See United States v. Camp, 343 F.3d 743 (5th Cir. 2003) (holding that a switch—which, if flipped, would cause a semi-automatic rifle to fire continuously—is a device that turns an ordinary firearm into a machinegun). The issue presented here is whether a bump stock is such a device. A bump stock is an accessory that attaches to a semi-automatic weapon and assists the shooter to engage in bump firing. Bump firing, in turn, is a technique whereby a shooter uses a firearm’s natural recoil to quickly reengage the trigger, resulting in an increased rate of fire. It is possible to bump fire an ordinary semi-automatic rifle without any assisting device, but a bump stock makes the technique easier. A typical bump stock consists of a sliding shoulder stock molded to a grip, a trigger ledge where the shooter places his finger, and a detachable rectangular receiver module that goes into the receiver well of the bump stock’s handle to guide the recoil of the weapon when fired. To begin bump firing, the shooter presses forward on the firearm’s forebody to bump into the trigger finger. The gun then slides back and forth, and the recoil energy forces the gun backward, re-engaging the trigger. The shooter maintains for-ward pressure on the gun’s forebody, again causing the trigger to bump into the trigger finger, maintaining fire. The firing process may be viewed as follows:[3] In summary, a bump stock combines with a semi-automatic weapon to facilitate the repeated function of the trigger. To be sure, it makes the process faster and easier. But the mechanics remain exactly the same: the firing of each and every round requires an intervening function of the trigger. This does not alter the form of manual input that the user must provide to discharge the weapon. Without a bump stock or the use of an alternative bump technique, the user must provide manual input by pulling the trigger with the muscles of his trigger finger. With a bump stock, the shooter need not pull and release his trigger finger. But the shooter must still apply forward pressure to the weapon’s forebody in order to maintain the shooting mechanism. Again, the manual input remains, even though its form changes. We note one important distinction. Some bump stocks—called mechanical bump stocks—are equipped with springs or other internal mechanical devices that automatically assist the shooter to engage in bump firing. For such a bump stock, the shooter does not have to maintain pressure on the barrel and trigger ledge in order to maintain this firing sequence. Only non-mechanical bump stocks are at issue in this case. D Bump stocks were first invented in the early 2000s. Historically, ATF distinguished between mechanical and non-mechanical bump stocks in categorizing a particular accessory as a machinegun. This categorization is done through ATF’s Firearms Technology Branch, which is authorized to issue classification letters upon request from members of the public. See 26 U.S.C. § 5841(c)(requiring firearm manufacturers and possessor to receive “authorization”); Bureau of Alcohol, Tobacco, Firearms and Explosives, National Firearms Handbook § 7.2.4 (setting forth the classification process). When ATF first considered mechanical bump stocks in 2006, it categorized them as machineguns: “[A] device attached to a semiautomatic firearm that uses an internal spring to harness the force of a firearm’s recoil so that the firearm shoots more than one shot with a single pull of the trigger is a machinegun.” 83 Fed. Reg. at 66514. ATF maintains that categorization. But from the time ATF first considered non-mechanical bump stocks to 2017, it categorized that type of bump stock as not being a machinegun. In that time, the Firearms Technology Branch issued dozens of classification letters regarding non-mechanical bump stocks, each time arriving at the same conclusion. One letter from 2010 is illustrative: Dear [Applicant], This is in reference to your submission . . . asking for an evaluation of a replacement shoulder stock for an AR-15 type rifle. Your letter advises that the stock (referenced in this reply as a “bump-stock”) is intended to assist persons whose hands have limited mobility to “bump-fire” an AR-15 type rifle. . . . The stock has no automatically functioning mechanical parts or springs and performs no automatic mechanical function when installed. In order to use the installed device, the shooter must apply constant forward pressure with the non-shooting hand and constant rearward pressure with the shooting hand. Accordingly, we find that the “bump-stock” is a firearm part and is not regulated as a firearm under the Gun Control Act or the National Firearms Act. However, bump-stock classification reached a point of inflection on October 1, 2017. On that day, a gunman murdered over 50 innocent men and women in Las Vegas, and injured 500 more. He used several weapons, many of which were equipped with extended magazines and bump stocks. These tragic events thrust bump stocks into the center of national attention. Within ten days of the shooting, two bills prohibiting bump-stock devices were proposed in Congress. See Automatic Gunfire Prevention Act, H.R. 3947, 115th Cong. (2017)[4]; To Amend Title 18, United States Code, To Prohibit the Manufacture, Possession, or Transfer of Any Part or Combination of Parts That is Designed and Functions to Increase the Rate of Fire of a Semiautomatic Rifle, H.R. 3999, 115th Cong. (2017). While Congress debated the bills, ATF published a notice of proposed rulemaking, intending to reverse its previous interpretation that non-mechanical bump stocks are not machineguns for purposes of the National Firearms Act and Gun Control Act. Notice of Proposed Rulemaking, Bump-Stock-Type Devices, 83 Fed. Reg. 13442 (Mar. 29, 2018). Senator Diane Feinstein—who sponsored one of the bills mentioned above—expressed concern with the proposed rule: Until today, the ATF has consistently stated that bump stocks could not be banned through regulation because they do not fall under the legal definition of a machine gun. Now, the department has done an about face, claiming that bump stocks do fall under the legal definition of a machine gun and it can ban them through regulations. The fact that ATF said as recently as April 2017 that it lacks this authority gives the gun lobby and its allies even more reason to file a lawsuit to block the regulations. Unbelievably, the regulation hinges on a dubious analysis claiming that bumping the trigger is not the same as pulling it. The gun lobby and manufacturers will have a field day with this reasoning. What’s more, the regulation does not ban all devices that accelerate a semi-automatic weapon[']s rate of fire to that of a machine gun. Both Justice Department and ATF lawyers know that legislation is the only way to ban bump stocks. The law has not changed since 1986, and it must be amended to cover bump stocks and other dangerous devices like trigger cranks. Our bill does this—the regulation does not. Press Release, Sen. Dianne Feinstein, Feinstein Statement on Regulation to Ban Bump Stocks (Mar. 23, 2018). ATF continued with the rulemaking process, publishing the final rule later that year. Final Rule, Bump-Stock-Type Devices, 83 Fed. Reg. 66514 (Dec. 26, 2018).[5] The Final Rule purported to modify the definition of machinegun as follows: A “machinegun,” “machine pistol,” “submachinegun,” or “automatic rifle” is a firearm which shoots, is designed to shoot, or can be readily restored to shoot, automatically more than one shot, without manual reloading, by a single function of the trigger. The term shall also include the frame or receiver of any such weapon, any part designed and intended solely and exclusively, or combination of parts designed and intended, for use in converting a weapon into a machinegun, and any combination of parts from which a machinegun can be assembled if such parts are in the possession or under the control of a person. For purposes of this definition, the term “automatically” as it modifies “shoots, is designed to shoot, or can be readily restored to shoot,” means functioning as the result of a self-acting or self-regulating mechanism that allows the firing of multiple rounds through a single function of the trigger; and “single function of the trigger” means a single pull of the trigger and analogous motions. The term “machinegun” includes a bump-stock-type device, i.e., a device that allows a semi-automatic firearm to shoot more than one shot with a single pull of the trigger by harnessing the recoil energy of the semi-automatic firearm to which it is affixed so that the trigger resets and continues firing without additional physical manipulation of the trigger by the shooter. 83 Fed. Reg. at 66553–54. E Plaintiff Michael Cargill lawfully acquired two non-mechanical bump stocks but surrendered them to ATF after passage of the Final Rule. He then sued ATF and other federal defendants, bringing several claims under the Administrative Procedure Act. First, he contends that ATF lacked authority to promulgate the Final Rule because its interpretation of machinegun conflicts with the unambiguous statutory definition. And even if the statute is ambiguous, Cargill says, it should be construed in his favor because of the rule of lenity. And because the statute concerns criminal penalties, the Government’s interpretation is not entitled to deference under Chevron USA, Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984). Cargill also argues that the Final Rule constitutes an unconstitutional exercise of legislative power by an administrative agency. After a one-day bench trial, the district court entered judgment for the Government. Cargill v. Barr, 502 F. Supp. 3d 1163 (W.D. Tex. 2020). It declined to apply Chevron but found that the Government’s new interpretation of machinegun is the best reading of the relevant statute. The district court also rejected Cargill’s nondelegation claim. A panel of this court affirmed, concluding that the Final Rule’s interpretation of machinegun is the best reading of the statute, and declining to reach Chevron or the nondelegation question. Cargill v. Garland, 20 F.4th 1004 (5th Cir. 2021). We granted rehearing en banc, vacating the panel opinion. 37 F.4th 1091. II Three of our sister circuits have reviewed preliminary-injunction motions relating to the Final Rule. The issues engendered great disagreement, but each circuit that has addressed them agrees that the definition of machinegun within the National Firearms Act and Gun Control Act does not unambiguously mean what the Government says it means. Each circuit ultimately denied preliminary injunctive relief, and the Supreme Court denied each of the certiorari petitions. See, e.g., Guedes v. ATF, 140 S. Ct. 789, 791 (2020) (Gorsuch, J., statement respecting denial of certiorari) (“Despite these concerns, I agree with my colleagues that the interlocutory petition before us does not merit review.”). A divided D.C. Circuit panel determined that the Final Rule is ambiguous, but applied Chevron deference to the Government’s statutory interpretation. Guedes v. ATF, 920 F.3d 1 (D.C. Cir. 2019). The dissent contended that the Final Rule contradicts the statute’s plain meaning. Id. at 35 (Henderson, J., concurring in part and dissenting in part). Although the Supreme Court denied certiorari, 140 S. Ct. 789 (2020), one Justice wrote separately to explain his view that Chevron does not apply because (i) the Government had expressly waived its application, (ii) it does not apply to regulations bearing criminal sanctions, and (iii) the Final Rule directly contradicts the Government’s previous interpretation. Id. at 789–91 (Gorsuch, J., statement respecting denial of certiorari). In another divided opinion, the Tenth Circuit reached the same conclusion as the D.C. Circuit: that the Final Rule was ambiguous and entitled to Chevron deference. Aposhian v. Barr, 958 F.3d 969 (10th Cir. 2020); see id. at 991 (Carson, J., dissenting). The court initially granted rehearing en banc, vacating the panel decision, 973 F.3d 1151 (10th Cir. 2020) (en banc), but later vacated the order as improvidently granted, reinstating the former opinion. Aposhian v. Wilkinson, 989 F.3d 890 (10th Cir. 2021) (en banc), cert denied sub nom. Aposhian v. Garland, 143 S. Ct. 83 (2022). Three dissents were written, each of which was joined by five of the eleven participating judges. The first dissent would have held that (i) the statute unambiguously does not apply to bump stocks, (ii) Chevron does not apply either because the Government waived it or because it does not apply in the criminal context, and (iii) the Final Rule may violate nondelegation principles. Id. at 891–903 (Tym-kovich, C.J., dissenting); see also id. at 903–04 (Hartz, J., dissenting); id. at 904–06 (Eid, J., dissenting); id. at 906–08 (Carson, J., dissenting). Finally, in yet another divided opinion, a Sixth Circuit panel ruled against the Government, declining to apply Chevron deference and holding the statutory definition of machinegun does not include bump stocks. Gun Owners of America, Inc. v. Garland, 992 F.3d 446, 450 (6th Cir. 2021); but see id. at 475-92 (White, J., dissenting) (arguing that Chevron applies and that the Government’s interpretation is reasonable). The court granted rehearing en banc, 2 F.4th 576 (6th Cir. 2021) (en banc), and an evenly-divided court affirmed the district court’s denial of the preliminary injunction. 19 F.4th 890 (6th Cir. 2020), cert denied, 143 S. Ct. 83 (2022). In addition to these circuit decisions, the Navy-Marine Corps Court of Criminal Appeals considered the Final Rule in the context of a criminal prosecution for possession of a machinegun. See United States v. Alkazahg, 81 M.J. 764, 780–81 (N–M. Ct. Crim. App. 2021). That court determined that, under the best reading of the statutory language, a bump stock is not a machinegun. But it ultimately found that the statute is ambiguous and applied the rule of lenity to construe the statute against imposing criminal liability. It dismissed the charge for possession of a machinegun. III Our primary task is to interpret the meaning of machinegun as defined in 26 U.S.C. § 5845(b). Of course, the Government has sponsored its own interpretation, as expressed in the Final Rule. Ordinarily, that action would invoke the two-step Chevron framework. As we recently summarized, “[a]t step one, we ask whether Congress has directly spoken to the precise question at issue, in which case we must give effect to the unambiguously expressed intent of Congress and reverse an agency’s interpretation that fails to conform to the statutory text.” Huawei Technologies USA, Inc. v. FCC, 2 F.4th 421, 433 (5th Cir. 2021) (quotations omitted). But here, the Government has declined to invoke Chevron in any of the lawsuits challenging the Final Rule. Nonetheless, several circuits have applied Chevron deference to challenges of the Final Rule, and so we will consider Chevron‘s applicability below. But before we do, we determine the statute’s meaning using traditional statutory-interpretation tools. That is, “the old-fashioned way.” Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari); see also Guedes, 920 F.3d at 42 (Henderson, J., concurring in part and dissenting in part); Aposhian, 989 F.3d at 898 (Tym-kovich, C.J., dissenting). If the statute is unambiguous, it does not matter whether Chevron applies. See, e.g., Western Refining Southwest, Inc. v. FERC, 636 F.3d 719, 727 (5th Cir. 2011) (“[If] the statute’s text is unambiguous, we need not proceed to Step Two of Chevron.”). Recall that in this circumstance, Congress has defined machinegun to mean “any weapon which shoots . . . automatically more than one shot . . . by a single function of the trigger,” or any accessory that allows a firearm to shoot in that manner. 26 U.S.C. § 5845(b). The parties dispute whether the fire created by a semi-automatic rifle equipped with a non-mechanical bump stock is produced both “automatically” and “by single function of the trigger.” We address those components in reverse order.[6] A The first phrase we consider is “by a single function of the trigger.” At the time the statute was passed, “function” meant “action.” Webster’s New International Dictionary 1019 (2d ed. 1934); see Guedes, 920 F.3d at 43 (Henderson, J., concurring in part and dissenting in part); Aposhian, 989 F.3d at 895 (Tymkovich, C.J., dissenting). Thus, the relevant question is whether a semi-automatic rifle equipped with a non-mechanical bump stock fires more than one shot each time the trigger “acts.” It does not. As illustrated above, a semi-automatic weapon utilizes a simple mechanical process: the trigger disengages the hammer from the sear, the hammer strikes the firing pin, the bullet fires, and the recoil pushes the hammer against the disconnector, which resets the trigger. This process happens every single time one bullet is fired. To be sure, a non-mechanical bump stock increases the rate at which the process occurs. But the fact remains that only one bullet is fired each time the shooter pulls the trigger. The Government contends that “single function of the trigger” means “a single pull of the trigger and analogous movements.” 83 Fed. Reg. at 66553. That is, according to the Government, “function” means “pull.” But that argument fails on its face because a shooter still pulls the trigger of a semi-automatic weapon equipped with a non-mechanical bump stock each time he or she fires a bullet. Without a bump stock, the trigger activates because the shooter flexes his or her finger; with a bump stock, the trigger activates because the recoil of the previous shot re-engages the trigger and the shooter’s maintained force on the gun’s forebody bumps the trigger against the shooter’s finger. This is a distinction without a difference—the end result in both cases is that the trigger is pulled. See Guedes, 920 F.3d at 48 (Henderson, J., concurring in part and dissenting in part) (“A semiautomatic rifle shoots a single round per pull of the trigger and the bump stock changes only how the pull is accomplished.”); Gun Owners of America, 992 F.3d at 469–73, vacated, 2 F.4th 576. Even if “single function” meant “single pull,” the definition would still not include a non-mechanical bump stock. Moreover, even though pulling the trigger can sometimes begin the bump firing sequence, the process is more typically begun by pushing forward on the fore-body of the firearm. For several of our sister circuits, however, the plain language is not so plain. They reason that single function of the trigger “could mean ‘a single pull of the trigger from the perspective of the shooter.’” Guedes, 920 F.3d at 29; see also Gun Owners of America, 19 F.4th at 905 (White, J., in support of affirmance). Considering the definition of “function,” one court understood the issue as such: “[T]hat definition begs the question of whether ‘function’ requires our focus upon the movement of the trigger, or the movement of the trigger finger. The statute is silent in this regard.” Aposhian, 958 F.3d at 986. According to that logic, for a semi-automatic rifle equipped with a non-mechanical bump stock, the act of pulling the trigger—which begins the bump firing sequence—is a single pull for purposes of the Gun Control Act and National Firearms Act. The problem with that interpretation is that it is based on words that do not exist in the statute. The statute “uses ‘single function of the trigger,’ not single function of the shooter’s trigger finger.” Guedes, 920 F.3d at 48 (Henderson, J., concurring in part and dissenting in part); see also Aposhian, 989 F.3d at 895 (Tymkovich, C.J., dissenting) (“The statute speaks only to how the trigger acts, making no mention of the shooter.”). The Navy-Marine Corps Court of Appeals likewise refused to read words into the statute: The best read implies that the shooter initiates the trigger function by some action, such as pulling the trigger—or it could be by just pushing a button—and it is the follow-on action where the trigger acts out its mechanical design or purpose that speaks to the “function of the trigger.” The statute does not say “by a single function of the trigger finger” nor does it say “by a single pull of the trigger in addition to external pressure from the shooter’s non-firing hand.” . . . . Had Congress wanted to use the phrase “by a single pull of the trigger” for machine guns, it could have. But it did not. Alkazahg, 81 M.J. at 780–81. We agree. The statutory definition of machinegun utilizes a grammatical construction that ties the definition to the movement of the trigger itself, and not the movement of a trigger finger. Nor do we rely on grammar alone. Context firmly corroborates what grammar initially suggests by demonstrating that Congress knew how to write a definition that is keyed to the movement of the trigger finger if it wanted to. But it did not. The Government offers nothing to overcome this plain reading, so that we are obliged to conclude that the statutory definition of machinegun unambiguously turns on the movement of the trigger and not a trigger finger. Grammar rejects a reading based on the shooter’s perspective. Each component of the statutory definition supports the mechanical perspective, not a shooter’s perspective. Again, the definition reads as follows: “[M]achinegun means . . . any weapon which shoots . . . automatically more than one shot . . . by a single function of the trigger.” 26 U.S.C. § 5845(b). The subject of the sentence, of course, is machinegun. The linking verb means connects the subject to the subject complement—weapon. Next, the adjectival phrase which shoots modifies weapon. The adverbial phrase automatically more than one shot then modifies shoots. Finally, two prepositional phrases follow. The first, by a single function, modifies the adverbial phrase. The second, of the trigger, modifies the first prepositional phrase. See also Guedes, 920 F.3d at 44 n.13 (Henderson, J., concurring in part and dissenting in part) (diagramming the statutory definition). The first thing to note is that the ultimate subject is machinegun, and the subject complement is weapon. In other words, a machinegun is defined by reference to what kind of weapon it is. But identifying the subject of the sentence is only our first step. We next look, second, to the fact that the term weapon is defined by how it shoots. So, again, the definition refers to the device being made to shoot, not the person or thing doing the shooting. Third, the manner of shooting must be automatic. Fourth—and critically—the  prepositional phrases define the firing process’s requirements from a mechanical perspective. The process must occur by a single function, and the single act must be by the trigger. In short, there is no mention of a shooter. The grammatical structure continuously points the reader back to the mechanics of the firearm. The statute does not care what human input is required to activate the trigger—it cares only whether more than one shot is fired each time the trigger acts. We do not stop with the grammar. With statutes, “[c]ontext is a primary determinant of meaning. Antonin Scalia & Bryan A. Garner, Reading Law: The Interpretation of Legal Texts 167 (2012). So, we look next to context for further clues. And context confirms that the statute must be read from the mechanical perspective. Specifically, context tells us that Congress knew how to write a definition that explicitly turns on the action of a shooter rather than the action of a trigger, but chose not to do so here. Immediately following the definition of machinegun provided in 26 U.S.C. § 5845(b), Congress defined the term “rifle” to mean a weapon designed “to use the energy of the explosive in a fixed cartridge to fire only a single projectile through a rifled bore for each single pull of the trigger.” § 5845(c) (emphasis added). The statute next defines “shotgun” to mean a weapon designed “to use the energy of the explosive in a fixed shotgun shell to fire through a smooth bore either a number of projectiles (ball shot) or a single projectile for each pull of the trigger.” Id. § 5845(d) (emphases added). “[W]here the document has used one term in one place, and a materially different term in another, the presumption is that the different term denotes a different idea.” Reading Law at 170. To summarize, the definition of machinegun must turn on the action (or “function”) of the trigger because no other actor is mentioned or implied. This conclusion is only strengthened by the fact that other definitions within the same statutory provision explicitly turn on the action of a shooter, showing that Congress knew how to write a definition that proceeds from a  shooter’s perspective, rather than a mechanical one, if it had wanted to. The notion that the definition turns on the action of an unnamed shooter is inconsistent with both the grammatical and statutory contexts. The Government says that this straightforward interpretation defies common sense. It would not have been prudent for Congress to “zero[] in on the mechanistic movement of the trigger,” the Government says, because the problem sought to be remedied was “the ability to drastically increase a weapon’s rate of fire.” Aposhian v. Barr, 374 F. Supp. 3d 1145, 1152 (D. Utah 2019)). Perhaps Congress’s choice of words was prudent, or perhaps it was not. That is not for us to decide. But the Government’s objection only accentuates the fact that Congress did not use words describing the shooter’s perspective or the weapon’s rate of fire. See Alkazahg, 81 M.J. at 781 (“Congress could have suggested that a shooter-focused approach or even a rate-of-fire approach was the way to read the statute by enacting those words. Even the term ‘machine gun’ suggests a mechanical approach where the shooter interaction is extremely limited.”). Instead, it made up an entirely new phrase—by a single function of the trigger—that specifically pertains to the mechanics of a firearm. Prudent or not, Congress defined the term “machinegun” by reference to the trigger’s mechanics. We are bound to apply that definition as written.[7] The Government also points to our decision in United States v. Camp, arguing that it controls here. 343 F.3d 734. It does not. But to the extent Camp applies, it supports Cargill’s position. The issue presented in Camp was whether something other than the metal lever that ordinarily begins the firing process can be a “trigger” for purposes of the Gun Control Act and National Firearms Act. The defendant there modified a semi-automatic rifle, building a switch behind the original trigger that, when pulled, “supplied electrical power to a motor connected to the bottom of a fishing reel that had been placed inside the weapon’s trigger guard; the motor caused the reel to rotate; and that rotation caused the original trigger to function in rapid succession.” 343 F.3d at 744. We held that the weapon was a machinegun even though the gun’s original trigger activated each time a bullet was fired. That was so because the gun had been modified such that it had a new trigger. As we held in United States v. Jokel, 969 F.2d 132, 135 (5th Cir. 1992), a trigger is just the “mechanism . . . used to initiate the firing sequence.” The mechanism used to initiate the firing sequence in Camp was the new switch. That switch operated by a single function, and so the firearm met the statutory definition of a machinegun. Camp, 343 F.3d at 745. Here, no party cites Camp for the proposition that the legally relevant trigger is anything other than the traditional trigger. And for good reason. All a non-mechanical bump stock does is allow the shooter to fire at an in-creased rate by harnessing a weapon’s natural recoil to re-engage the trigger, and by using the shooter’s maintained forward force. The case might well be different if we were considering a semi-automatic weapon equipped with a mechanical bump stock. It could be the case that a switch activating a mechanical bump stock would be the legal trigger. But we are not considering that case.[8] Here, the definition of “trigger” is not in dispute. If anything, Camp supports our conclusion because the trigger at issue in this case is different in kind from the trigger at issue there. B Even if a non-mechanical bump stock caused a semi-automatic rifle to operate by a single function of the trigger, the rifle would still need to operate automatically in order to be a machinegun.[9] All generally agree that here, automatically means “self-acting.” Oxford English Dictionary at 574 (1933) (“[s]elf-acting under conditions fixed for it, going of itself”); see also Cargill, 20 F.4th at 1012; Guedes, 920 F.3d at 30; id. at 43 (Henderson, J. concurring in part and dissenting in part); Aposhian, 958 F.3d at 986; Aposhian, 989 F.3d at 895 (Tymkovich, C.J., dissenting); Gun Owners of America, 19 F.4th at 905–06 (White, J., in support of affirmance); id. at 912–13 (Murphy, J., in opposition to affirmance). But the parties dispute whether the firing process enabled by a non-mechanical bump stock is self-acting. It is not. As an initial matter, we must remember that the phrase “by a single function of the trigger” modifies the adverb “automatically.” Thus, the condition is satisfied only if it is the trigger that causes the firearm to shoot automatically. See Guedes, 920 F.3d at 43 (Henderson, J., concurring in part and dissenting in part) ( “‘Automatically’ cannot be read in isolation. On the contrary, it is modified—that is, limited—by the clause ‘by a single function of the trigger.’”); Aposhian, 989 F.3d at 896 (Tymkovich, C.J., dissenting). That is not how a bump stock works. Bump firing does not maintain if all a shooter does is initially pull the trigger. Rather, to continue the firing after the shooter pulls the trigger, he or she must maintain manual, forward pressure on the barrel and manual, backward pressure on the trigger ledge. The Government argues that, taken together, those actions create automatic fire. But Cargill would prevail even if that were true because those actions are not “a single function of the trigger.” For example, the ATF’s treatment of the Ithaca Model 37 “slam fire” shotgun confirms that bump stocks do not enable automatic fire. With the Model 37, a shooter can pull the trigger once and hold it. Then, after each pump with the shooter’s non-trigger hand, a new shell is loaded and immediately discharged. According to the ATF, the Model 37 fires multiple shots by a single function of the trigger, but it does not do so automatically because the shooter must manually pump the shotgun with his non-trigger hand. See 83 Fed. Reg. at 66,534. By this same logic, a rifle equipped with a non-mechanical bump stock does not fire automatically because the shooter must manually apply forward pressure on the barrel with his or her non-trigger hand. The Government recognizes this logic but argues that it proves too much. After all, the Government says, to operate a traditional automatic rifle, the shooter must pull and hold the trigger to fire more than one round. No one doubts that a traditional automatic weapon is a machinegun for purposes of federal law. And so it cannot be that a process is not automatic simply because it requires sustained input. See Guedes, 920 F.3d at 31; Aposhian, 958 F.3d at 987. That argument makes the same mistake as before: it untethers “single function of the trigger” from “automatically.” Restated, the statute requires that a machinegun be capable of firing automatically once thetrigger performs a single function. An automatic weapon satisfies this requirement because the act of pulling and holding the trigger is one function, and that function produces more than one shot. That force must be maintained on the trigger does not change this conclusion. Stated succinctly: [A] gun shoots automatically by a single function of the trigger as long as the shooter need only manually cause the trigger to engage in a “single” function in order to fire multiple shots . . . So a typical machine gun qualifies even though the shooter pulls the trigger and keeps it pressed down because that combined external influence still does no more than result in one action of the trigger. Gun Owners of America, 19 F.4th at 915 (Murphy, J., in opposition to affirmance); see also Guedes, 920 F.3d at 44 (Henderson, J., concurring in part and dissenting in part) (“The statutory definition of machinegun does not include a firearm that shoots more than one round automatically by a single pull of the trigger and then some (that is, by constant forward pressure with the non-trigger hand).”) (internal quotation marks omitted and emphasis omitted). As understood by the Navy-Marine Corps Court of Appeals: It is incorrect to equate the holding of the trigger in an automatic weapon with the holding of the trigger and the forward motion in a semi-automatic weapon equipped with a bump stock. That is because the former is shooting automatically by a single function of the trigger, while the latter is relying on an additional human action beyond the mechanical self-acting and impersonal trigger function. Alkazahg, 81 M.J. at 782–83. We reiterate that a shooter can bump fire an ordinary semi-automatic rifle even without a bump stock. But nobody, not even the Government, contends that semi-automatic rifles are machineguns. That concession damns the Government’s position. As Cargill recognizes, if ordinary bump firing constituted automatic fire, the Final Rule would “convert a semiautomatic weapon into a machinegun simply by how a marksman used the weapon.” That absurd result reveals the flaw in the Government’s line of reasoning. In addition to implying absurd results, the Government’s position is quite telling. It would allow the use of semi-automatic rifles, which can bump fire, but prohibit the use of non-mechanical bump stocks, even though there is no mechanical difference between the two forms of gunfire. Rather, the meaningful difference is that, with a non-mechanical bump stock, bump firing is easier and can occur at a faster rate. That is  a distinction Congress certainly could have addressed in the National Firearms Act and Gun Control Act. But Congress did not prohibit machineguns according to how quickly they fire. It prohibited machineguns according to the way that they fire. And semi-automatic weapons do not fire “automatically,” even when equipped with a non-mechanical bump stock. * * * The definition of machinegun as set forth in the Gun Control Act and National Firearms Act establishes two conditions that must obtain in order for a weapon to qualify. The weapon must operate “automatically” and “by single function of the trigger.” According to the statute’s unambiguous language, neither condition obtains as applied to a semi-automatic rifle equipped with a non-mechanical bump stock. The failure of either condition is sufficient to entitle Cargill to judgment. IV As introduced above, several of our sister circuits applied Chevron to challenges to this Final Rule, even though no party requested its application. Because we hold that the statute is unambiguous, Chevron deference does not apply even if the Chevron framework does. See Western Refining Southwest, 636 F.3d at 727. But if the statute were ambiguous, Chevron would not apply for any of the three reasons explained below. A First, Chevron does not apply for the simple reason that the Government does not ask us to apply it. Indeed, the Government affirmatively argued in the district court that Chevron deference is unwarranted. As other jurists have recognized in this context, that means that the Chevron argument has been waived—not merely forfeited. United States v. Olano, 507 U.S. 725, 733 (1993) (“Whereas forfeiture is the failure to make the timely assertion of a right, waiver is ‘the intentional relinquishment or abandonment of a known right.’”) (quoting Johnson v. Zerbst, 304 U.S. 458, 464 (1938)); Guedes, 920 F.3d at 21 (“To the extent Chevron treatment can be waived, we assume that the government’s posture in this litigation would amount to a waiver rather than only a forfeiture.”); Aposhian, 989 F.3d at 897 (“[W]hen a party chooses not to pursue a legal theory potentially available to it, we generally take the view that it is ‘inappropriate’ to pursue that theory in our opinions.”) (internal citation and quotation marks omitted) (Tymkovich, C.J., dissenting). That would seem to be the end of the inquiry, but we recognize that one of our sister circuits has held that Chevron cannot be waived. Guedes, 920 F.3d at 21–23; see also Gun Owners of America, 19 F.4th at 899 n.5 (White, J., in support of affirmance). To be sure, we have never held in a published case that Chevron must be raised by the Government in order to apply. See Albanil v. Coast 2 Coast, Inc., 444 F. App’x 788, 796 (5th Cir. 2011). But the conclusion is obvious, and flows from well-settled waiver principles. After all, that a court should defer to the Government’s expressed interpretation is just a legal argument, and a party waives a legal argument if it fails to raise the argument when presented with the opportunity. See, e.g., Sindhi v. Raina, 905 F.3d 327, 334 (5th Cir. 2018); Fruge v. Amerisure Mut. Ins. Co., 663 F.3d 743, 747 (5th Cir. 2011). As explained in the Tenth Circuit’s consideration of the Final Rule, “We refuse to consider arguments a party fails to make because we ‘depend on the adversarial process to test the issues for our decision’ and are concerned ‘for the affected parties to whom we traditionally extend notice and an opportunity to be heard on issues that affect them.’” Aposhian, 989 F.3d at 897 (Tymkovich, C.J., dissenting) (quoting Hydro Resources, Inc. v. EPA, 608 F.3d 1131, 1146 n.10 (10th Cir. 2010) (en banc)); see also Carducci v. Regan, 714 F.2d 171, 177 (D.C. Cir. 1983) (Scalia, J.) (“The premise of our adversarial system is that appellate courts do not sit as self-directed boards of legal inquiry and research, but essentially as arbiters of legal questions presented and argued by the parties before them.”). We must not defer to the Government’s interpretation here for the simple reason that no party argues that we should. If ordinary waiver principles were not enough, we note also that it would contradict Chevron‘s central justification to defer to the Government’s interpretation without its urging us to do so. The justification is that “‘policy choices’ should be left to executive branch officials ‘directly accountable to the people.”‘ Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari) (quoting Epic Systems v. Lewis, 138 S. Ct. 1612, 1630 (2018) and Chevron, 467 U.S. at 865)). Here, the Government made a clear policy choice by declining to seek Chevron deference. The very interest underlying Chevron demands that we respect the Government’s choice and interpret the statute according to traditional principles of statutory interpretation. See Aposhian, 989 F.3d at 898 (“If the agency disavows any reliance on Chevron, who are we to second-guess it?”) (Tymkovich, C.J., dissenting); Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari) (“[C]ourts must equally respect the Executive’s decision not to make policy choices in the interpretation of Congress’s handiwork.”). Raising the issue sua sponte, the D.C. Circuit assumed that the Government’s actions were consistent with waiver, but held that Chevron cannot be waived. As an initial matter, that conclusion is likely inconsistent with Supreme Court precedent. Our highest Court “has often declined to apply Chevron deference when the government fails to invoke it.” Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari) (collecting cases); see, e.g., HollyFrontier Cheyenne Refining, LLC v. Renewable Fuels Ass’n, 141 S. Ct. 2172, 2180(2021) (“[T]he government is not invoking Chevron. We therefore decline to consider whether any deference might be due its regulation.”) (quotation omitted). The argument against waiver is based on the premise that Chevron is a standard of review. Guedes, 920 F.3d at 163; see also Gun Owners of America, 992 F.3d at 477–78, vacated, 2 F.4th 576. It is certainly true that parties in litigation cannot waive the applicable standard of review. E.g., United States v. Escobar, 866 F.3d 333, 339 n.13 (5th Cir. 2017) (quotation omitted). But Chevron is not a standard of review. The APA specifically sets forth standards by which courts must review agency actions—arbitrary and capricious, abuse of discretion, in excess of statutory authority, and so on. See 5 U.S.C. § 706. Chevron is merely a legal argument that the Government can make to contend that its interpretation satisfies the relevant standard of review.[10] B The Chevron framework does not apply for a second, independent reason: the statute which the Final Rule interprets imposes criminal penalties. As noted above, the primary reason for Chevron is that it allows the executive branch to make policy decisions through the accrued expertise of administrative agencies. But in exchange, Chevron deference shifts the responsibility for lawmaking from the Congress to the Executive, at least in part. That tradeoff cannot be justified for criminal statutes, in which the public’s entitlement to clarity in the law is at its highest. See Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of writ of certiorari) (“Before courts may send people to prison, we owe them an independent determination that the law actually forbids their conduct.”); Esquivel-Quintana v. Lynch, 810 F.3d 1019, 1027 (6th Cir. 2016) (Sutton, J., concurring in part and dissenting in part) (Applying Chevron to criminal statutes would “permit the aggregation” of executive and legislative power “in the one area where its division matters most: the removal of citizens from society.”). No precedent compels Chevron‘s application here. To the contrary, the Supreme Court has “never held that the Government’s reading of a criminal statute is entitled to any deference.” United States v. Apel, 571 U.S. 359, 369 (2014); see also Gutierrez-Brizuela v. Lynch, 834 F.3d 1142, 1155 (10th Cir. 2016) (“The Supreme Court has expressly instructed us not to apply Chevron deference when an agency seeks to interpret a criminal statute.”) (Gorsuch, J., concurring). That is so because “criminal laws are for the courts, not for the Government, to construe.” Abramski v. United States, 573 U.S. 169, 191 (2014). For this reason, when “the Government interprets a criminal statute too broadly . . . or too narrowly . . . a court has an obligation to correct its error.” Id. We must not apply Chevron where, as here, the Government seeks to define the scope of activities that subject the public to criminal penalties. This is precisely the position we have taken before, in an unpublished decision. United States v. Garcia, 707 F. App’x 231, 234 (5th Cir. 2017) (“The Supreme Court has now resolved this uncertainty, instructing that no deference is owed to agency interpretations of criminal statutes.”). Several of our sister circuits disagree, however. See Guedes, 920 F.3d at 163–67; Aposhian, 958 F.3d at 982–84. The disagreement stems from one paragraph in the decision Babbitt v. Sweet Home Chapter of Communities for a Great Oregon, 515 U.S. 687 (1995). There, the Supreme Court considered a Department of the Interior regulation that interpreted the Endangered Species Act’s criminal prohibition on “taking” an endangered species as including modification of the species’s habitat. The Court did not conduct a Chevron analysis, but upheld the regulation, concluding that it “owe[d] some degree of deference [to] the [DOI's] reasonable interpretation,” in part because of the “latitude the ESA gives to [DOI] in enforcing the statute.” Id. at 703. It also declined to apply the rule of lenity, reasoning that an administrative regulation does not necessarily invoke the rule just because “the governing statute authorizes criminal enforcement.” Id. at 704 n.18. Several courts cite Babbitt for the proposition that the Chevron frame-work applies with equal force to criminal regulations and displaces the rule of lenity, but it does not support that conclusion. “While Babbitt certainly cited Chevron and used the word deference with regard to the DOI’s interpretation, Babbitt did not discuss or decide whether Chevron applied nor did it analyze the challenge using Chevron, just as it did not decide whether the rule of lenity applied or analyze the challenge using the rule of lenity.” Gun Owners of America, 992 F.3d at 457, vacated, 2 F.4th 576. As explained by two members of the Supreme Court, Babbitt did not purport to set forth a general rule respecting the interpretation of criminal regulations: “The best that one can say . . . is that in Babbitt[] [the Court] deferred, with scarcely any explanation, to an agency’s interpretation of a law that carried criminal penalties. . . . Babbitt‘s drive-by ruling, in short, deserves little weight.” Whitman v. United States, 574 U.S. 1003, 135 S. Ct. 352, 353 (Scalia, J., joined by Thomas, J., respecting the denial of certiorari). This is confirmed by subsequent Supreme Court precedent, addressing the rule of lenity in relation to Chevron and declining to defer to agency interpretations of criminal statutes. See Aposhian, 989 F.3d at 901 (Tym-kovich, C.J., dissenting) (collecting cases). Babbitt does not require us to apply Chevron in these circumstances. Indeed, most proximate sources suggest otherwise. See Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari) (“[W]hatever else one thinks about Chevron, it has no role to play when liberty is at stake.”). As such, Chevron does not apply here because the statutory language at issue implicates criminal penalties. C Finally, we note a third reason why Chevron deference does not apply in these circumstances: that ATF has adopted an interpretive position that is inconsistent with its prior position. To apply Chevron here would contravene one of the rule’s central purposes: “to promote fair notice to those subject to criminal laws.” United States v. Kozminski, 487 U.S. 931, 952 (1988); see also United States v. Kaluza, 780 F.3d 647, 669 (5th Cir. 2015) (“The rule ‘vindicates the fundamental principle that no citizen should be held accountable for a violation of a statute whose commands are uncertain, or subjected to punishment that is not clearly prescribed.’”) (quoting United States v. Santos, 553 U.S. 507, 514 (2008) (plurality op.)).[11] If we were required to defer to the Government’s position, the Government could change the scope of criminal liability at any time. Indeed, that is exactly what it has done here. Until 2017, the ATF had never classified non-mechanical bump stocks as machineguns. But now the interpretation is reversed, and the Government would criminalize behavior that it long recognized was lawful. In considering one of the other cases involving the regulation, one member of the Supreme Court explained the problem as such: Chevron‘s application in this case may be doubtful for other reasons too. The agency used to tell everyone that bump stocks don’t qualify as ‘machineguns.’ Now it says the opposite. The law hasn’t changed, only an agency’s interpretation of it. How, in all this, can ordinary citizens be expected to keep up? . . . And why should courts, charged with the independent and neutral interpretation of the laws Congress has enacted, defer to such bureaucratic pirouetting? Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari); see also Aposhian, 989 F.3d at 900 (“When an agency plays pinball with a statute’s interpretation, as the ATF has here, fair notice cannot be said to exist.”) (Tymkovich, C.J., dissenting); Guedes, 920 F.3d at 181 (“TheATF’s interpretation of ‘machinegun’ gives anything but fair warning—instead, it does a volte-face of its almost eleven years’ treatment of a non-mechanical bump stock as not constituting a ‘machinegun.’”) (Henderson, J., concurring in part and dissenting in part).[12] The concern respecting the consistency of agency regulations is nothing new. Indeed, the Supreme Court has long recognized that an agency interpretation that “conflicts with the agency’s earlier interpretation is entitled to considerably less deference than a consistently held agency view.” INS v. Cardoza-Fonseca, 480 U.S. 421, 446 n.30 (1987) (internal quotations omitted); see also Cargill, 502 F. Supp. 3d at 1189 (quoting Cardoza-Fonseza). The concern is only magnified where, as here, the Government’s interpretation of the underlying statute carries implications for criminal liability. As such, Chevron does not apply  because the Government has construed the same statute in two, inconsistent ways at different points in time. V Turning to the rule of lenity, and assuming arguendo that the relevant statute is ambiguous, we now consider whether that “ambiguity concerning the ambit of criminal statutes should be resolved in favor of lenity.” Rewis v. United States, 401 U.S. 808, 812 (1971) (citation omitted). We conclude that the rule of lenity applies if the statute is ambiguous. We recognize that courts have considered two standards for whether a statute is sufficiently ambiguous to trigger the rule of lenity. One standard asks whether there is a “reasonable doubt” as to the statute’s meaning. See Reading Law at 299 (quoting Moskal v. United States, 498 U.S. 103, 108 (1990)). The other inquires whether there is a “grievous ambiguity” in the statute. See, e.g., Chapman v. United States, 500 U.S. 453, 463 (1991) (quoting Huddleston v. United States, 415 U.S. 814, 831 (1974)). The Supreme Court does not appear to have decided which of these standards governs the rule of lenity. See Wooden v. United States, 142 S. Ct. 1063, 1075 (2022) (Kavanaugh, J., concurring) (arguing in favor of the grievous-ambiguity standard); id. at 142 S. Ct. at 1084 (Gorsuch, J., concurring in the judgment) (arguing in favor of the reasonable-doubt standard). But it does not matter which standard applies because the rule of lenity applies even under the more stringent “grievously ambiguous” condition. One formulation of that standard provides that lenity applies if a court cannot discern the statute’s meaning even “after seizing everything from which aid can be derived.” Wooden, 142 S. Ct. at 1075 (Kavanaugh, J., concurring) (quoting Ocasio v. United States, 578 U.S. 282, 295 n.8 (2016)). Assuming that the statute at issue here is ambiguous, we can only “guess” at its definitive meaning. Suchowolski, 838 F.3d at 534. We have availed ourselves of all traditional tools of statutory construction, and in this circumstance, they fail to provide meaningful guidance. That is sufficient to require application of the rule of lenity irrespective of whether the reasonable doubt or grievous ambiguity standard applies. The dissenting opinion objects to our application of lenity, arguing that we fail to explain why the statute at issue here is grievously ambiguous. With this understanding, it concludes that our holding implies that “ambiguous statutes are always grievously ambiguous.” Post at 57. This criticism misunderstands our holding. We do not conclude that all ambiguous statutes are grievously ambiguous—only that this one is. Our conclusion fits comfortably into the dissenting opinion’s own conceptualization of the grievous-ambiguity standard. At the very least, lenity is appropriate, the dissenting opinion concedes, if after “having tried to make sense of a statute using every other tool, we face an unbreakable tie between different interpretations.” Post at 56. That is the case here in two respects. First, the parties argue whether “a single function of the trigger” refers to the firearm’s mechanics or to the shooter’s pulling of the trigger. Second, the parties argue if the process of engaging the trigger and maintaining forward pressure on the gun’s forebody produces “automatic” fire. True, the precise meaning of “grievously ambiguous” is not entirely clear. See Shular v. United States, 140 S. Ct. 779, 788 (2020) (Kavanaugh, J., concurring) (noting that “the Court has not always been perfectly consistent in its formulations [of grievous ambiguity]); Reading Law at 299 (arguing that the term grievous ambiguity “provides little more than atmospherics, since it leaves open the crucial question—almost invariably present—of how much ambiguousness constitutes an ambiguity”) (quoting United States v. Hansen, 772 F.2d 940, 948 (D.C. Cir. 1985)). But having utilized all available tools of statutory interpretation, and assuming arguendo that those two provisions are indeed ambiguous, we are unable to resolve either of the ties. See Alkazahg, 81 M.J. at 784 (expressing “genuine confusion as to what the statute means”). That is sufficient to conclude that this statute—and for purposes of this case, only this statute—is grievously ambiguous. The dissenting opinion also objects that applying lenity in this case wrongfully takes Congress’s prerogative to establish federal crimes and transfers that power to the Judiciary, violating the separation of powers. But this is really just a repetition of the dissenting opinion’s disagreement regarding the correct interpretation of the statute at issue here. As we understand it, the National Firearms Act and Gun Control Act do not unambiguously criminalize the possession of a non-mechanical bump stock. To apply lenity in this case preserves the separation of powers “by maintaining the legislature as the creator of crimes.” Esquivel-Quintana, 810 F.3d at 1019. If ATF could change the scope of criminal liability by issuing a regulation—free from the taxing obligations of bicameralism and presentment—the Executive could wield power that our Constitution reserves to the Legislature. The rule of lenity also prevents the possibility whereby Congress passes an ambiguous criminal statute, only to be interpreted later by a federal agency. See id. (“By applying lenity in this setting, last of all, courts would avoid incentivizing Congress to enact hybrid statutes that duck under lenity’s imperatives, to say nothing of other imperatives in construing criminal laws.”). To be sure, it would be inconsistent with our constitutional structure to apply lenity in an unprincipled manner to statutes that are not really ambiguous. But here, we are wholly persuaded that if the definition of “machinegun” does not unambiguously exclude non-mechanical bump stocks, its inclusion of the latter is at the very least ambiguous. Given that conclusion, our separation of powers is aided, rather than impeded, by applying the rule of lenity. The rule of lenity is a “time-honored interpretive guideline.” Liparota v. United States, 471 U.S. 419, 429 (1985). We have applied it many times to construe ambiguous statutes against imposing criminal liability. See e.g., United States v. Cooper, 38 F.4th 428, 434 (5th Cir. 2022); Kaluza, 780 F.3d at 669; United State v. Orellano, 405 F.3d 360, 370 (5th Cir. 2005). This case is no different: assuming the definition of machinegun is ambiguous, we are bound to apply the rule of lenity. That is, we are bound to construe the  definition of machinegun to exclude a semi-automatic weapon equipped with a non-mechanical bump stock. See Alkazahg, 81 M.J. at 784 (“We decline to step into the role of the legislature when the legislature has not been clear about whether Appellant’s conduct was criminal. Judge Henry Friendly described the rule of lenity as ‘the instinctive distaste against men languishing in prison unless the lawmaker has clearly said they should.’ Here, we express that distaste.”) (quoting United States v. Bass, 404 U.S. 336, 348 (1971)). Therefore, assuming arguendo that the statute is ambiguous, we conclude that the rule of lenity demands that we resolve that ambiguity in favor of Cargill, and in turn conclude that a non-mechanical bump stock is not a machinegun for purposes of the National Firearms Act and Gun Control Act. VI Cargill also argues that the passage of the Final Rule is an exercise of legislative power, in violation of the Constitution’s vesting all such power in Congress. U.S. Const. art. I, § 1. Some have expressed serious concern at the ATF’s lack of explicit authorization to interpret criminal statutes as such. [W]e should feel deep discomfort at allowing an agency to define the very criminal rules it will enforce by implicit delegation. Such a delegation “turn[s] the normal construction of criminal statutes upside down, replacing the doctrine of lenity with a doctrine of severity.” Carter v. Welles-Bowen Realty, Inc., 736 F.3d 722, 730 (6th Cir. 2013) (Sutton, J., concurring). The delegation raises serious constitutional concerns by making ATF the expositor, executor, and interpreter of criminal laws. Aposhian, 989 F.3d at 900 (Tymkovich, C.J., dissenting). We acknowledge this concern, especially in light of statements made by several members the Supreme Court calling into question the relevant standards for legislative-power-delegation issues. See Gundy v. United States, 139 S. Ct. 2116, 2131–42 (2019) (Gorsuch, J., dissenting, joined by Roberts, C.J., and Thomas, J.); id. at 2130–31 (Alito, J., concurring in the judgment); Paul v. United States, 140 S. Ct. 342 (2019) (Kavanaugh, J., statement respecting the denial of certiorari). We need not decide this question because multiple independent reasons compel us to hold the Final Rule to be unlawful. But if more were needed, this issue may well implicate the canon of constitutional avoidance. Under that well-established doctrine, “where an otherwise acceptable construction of a statute would raise serious constitutional problems, the Court will construe the statute to avoid such problems unless such construction is plainly contrary to the intent of Congress.” Hersh v. United States, 553 F.3d 743, 754–55 (5th Cir. 2008) (quoting Edward J. DeBartolo Corp. v. Florida Gulf Coast Building & Constr. Trades Council, 485 U.S. 568, 575 (1988)); see also Reading Law at 250. For many jurists, the question of Congress’s delegating legislative power to the Executive in the context of criminal statutes raises serious constitutional concerns. See Guedes, 140 S. Ct. at 790 (Gorsuch, J., statement respecting denial of certiorari); Aposhian, 989 F.3d at 900 (Tymkovich, C.J., dissenting); Esquivel-Quintana, 810 F.3d at 1027 (Sutton, J., concurring in part and dissenting in part). We do not reach this issue because we do not have to. But if we did, it would only provide more support for the conclusion that a semi-automatic rifle equipped with a non-mechanical bump stock is not a machinegun for purposes of federal law. VII Having determined that the judgment of the district court must be reversed, we remand this case to the district court to enter judgment for Cargill and to determine the proper scope of relief. It is well-established that “[a] plaintiff’s remedy must be tailored to redress the plaintiff’s particular injury.” Gill v. Whitford, 138 S. Ct. 1916, 1934 (2018). And as an initial matter,  vacatur of an agency action is the default rule in this Circuit. See, e.g., Data Mktg. Partnership, LP v. United States Dept. of Labor, 45 F.4th 846, 859 (5th Cir. 2022) (“The default rule is that vacatur is the appropriate remedy.”); see also Franciscan Alliance, Inc. v. Becerra, 47 F.4th 368, 374–75 (5th Cir. 2022) (“Vacatur is the only statutorily prescribed remedy for a successful APA challenge to a regulation.”). But the parties have not briefed the remedial-scope question, and it may be the case that a more limited remedy is appropriate in these circumstances. We express no opinion on that question other than to observe that the district court is well-placed to answer the question in the first instance. We therefore remand this case to the district court with the instruction that it enter judgment for Cargill and determine what remedy—injunctive, declarative, or otherwise—is appropriate to effectuate that judgment. VIII Many commentators argue that non-mechanical bump stocks contribute to firearm deaths and that the Final Rule is good public policy. We express no opinion on those arguments because it is not our job to determine our nation’s public policy. That solemn responsibilitylies with the Congress, and our task is confined to deciding cases and controversies, which requires us to apply the law as Congress has written it.[13] In defining the term machinegun, Congress referred to the mechanism by which the gun’s trigger causes bullets to be fired. Policy judgments aside, we are bound to apply that mechanical definition. And applying that definition to a semi-automatic rifle equipped with a non-mechanical bump stock, we conclude that such a weapon is not a machinegun for purposes of the Gun Control Act and National Firearms Act. Chevron deference likely has no role here either because the Government waived it or because it does not apply to the Government’s interpretation of a statute imposing criminal penalties. Finally, even if the statute were ambiguous—which it is not—the rule of lenity would require that we interpret the statute in Cargill’s favor. As Justice Holmes framed it years ago, “it is reasonable that a fair warning should be given to the world in language that the common world will understand, of what the law intends to do if a certain line is passed.” McBoyle, 283 U.S. at 27. We cannot say that the National Firearms Act and Gun Control Act give that fair warning that possession of a non-mechanical bump stock is a crime. The Final Rule promulgated by the ATF violates the APA. We therefore REVERSE the judgment of the district court and REMAND with instructions to enter judgment for Cargill. Haynes, Circuit Judge, joined by Richman, Chief Judge, concurring in the judgment: I concur in the judgment only because I reluctantly conclude that the relevant statute is ambiguous such that the rule of lenity favors the citizen in this case. James C. Ho, Circuit Judge, joined by Richman, Chief Judge, and Southwick, Circuit Judge, concurring in part and concurring in the judgment: Under the rule of lenity, “[p]enal statutes must be construed strictly.” 1 William Blackstone, Commentaries *88. “This is a rule of construction . . . as old and well established as law itself.” United States v. Wilson, 28 F. Cas. 699, 709 (C.C.E.D. Pa. 1830). Our courts have thus long recognized that “all reasonable doubts concerning [the] meaning [of a penal statute] . . . operate in favor of [the defendant].” Harrison v. Vose, 50 U.S. 372, 378 (1850). We apply the rule of lenity where “reasonable doubt persists about a [criminal] statute’s intended scope.” Moskal v. United States, 498 U.S. 103, 108 (1990). When standard principles of statutory interpretation “fail to establish that the Government’s position is unambiguously correct . . . [we] resolve the ambiguity in [the defendant's] favor.” United States v. Granderson, 511 U.S. 39, 54 (1994). The ruleof lenity rests on “the principle that the power of punishment is vested in the legislative, not in the judicial department.” United States v. Wiltberger, 18 U.S. 76, 85 (1820) (Marshall, C.J.). The rule also ensures fair notice to citizens: “To make the warning fair, . . . the line should be clear.” McBoyle v. United States, 283 U.S. 25, 27 (1931) (Holmes, J.). In sum, it is not enough to conclude that a criminal statuteshould cover a particular act. The statute must clearly and unambiguously cover the act. Consider the disturbing problem of designer drugs. The Controlled Substances Act of 1970 prohibits drugs listed on certain schedules. See Thor v. United States, 554 F.2d 759, 762–63 (5th Cir. 1977). In response, a line of synthetically produced drugs was created “to mimic the pharmacological effects” of scheduled drugs, while evading the Act. Clayton L. Smith, The Controlled Substance Analogue Enforcement Act of 1986: The Compromising of Criminalization, 16 Am. J. Crim. L. 107, 108 (1988). Designer drugs differed in chemical composition from scheduled drugs. But they were just as lethal. So the same policy justifications for banning scheduled drugs readily applied to designer drugs. Yet all three branches agreed that existing law did not ban designer drugs. See, e.g., Controlled Substance Analogs Enforcement Act of 1985: Hearing on S. 1437 Before the S. Comm. on the Judiciary, 99th Cong. 1st Sess. 2 (1985) (opening statement of Chairman Strom Thurmond) (“[U]nlawful activity under the Controlled Substances Act is defined with regard to the precise chemical makeup of the substances described by schedules . . . . Make a minor alteration in the molecular structure of an outlawed drug . . . and you have produced a . . . dangerous narcotic that is not illegal.”); id. at 41–42 (statement of Stephen S. Trott, Assistant Attorney General, Criminal Division, Department of Justice) (“[I]f a particular substance is not included in one of the schedules, it is not unlawful to manufacture or to distribute it . . . despite the potential for abuse . . . .”); United States v. Gavrilovic, 551 F.2d 1099, 1106 (8th Cir. 1977) (reversing a conviction where a concededly dangerous narcotic had not been properly added to a schedule). So a new act of Congress was required to get at the problem of designer drugs. And that’s why Congress enacted the Controlled Substance Analogue Enforcement Act of 1986. Pub. L. No. 99-570, §§ 1201–04, 100 Stat. 3207, 3207–13 to 3207–14 (1986). See also United States v. Muhammad, 14 F.4th 352, 355 (5th Cir. 2021) (“[T]he Analogue Act is an antidote to statutory evasion: It expands the CSA’s coverage to include substances that, while technically not on the schedules, mimic those that are.”). Bump stocks present the same basic conundrum as designer drugs. Federal law criminalizes the possession of fully automatic machineguns. See 18 U.S.C. § 922(o)(1) (machinegun ban); 26 U.S.C. § 5845(b) (machinegun definition). That prohibition does not apply to semiautomatic weapons. See Gun Control Act of 1968, § 201, 82 Stat. 1231 (codified at 26 U.S.C. § 5845(b)) (amending the 1934 federal definition of “machinegun” to omit weapons that shoot “semiautomatically”). But bump stocks now allow semiautomatic weapons to mimic automatic machineguns. By attaching a bump stock to a semiautomatic firearm, the shooter can simulate the experience of firing an automatic machinegun. See 83 Fed. Reg. 66,515–16. Just as designer drugs achieve the same lethality as scheduled drugs, bump stocks allow semiautomatic weapons to achieve the same lethality as fully automatic machineguns. But once again, Congress must take action if it wishes to criminalize bump stocks, just as it did during the 1980s when it came to designer drugs. That’s because federal law defines “machinegun” as a weapon that shoots more than one shot “automatically . . . by a single function of the trigger.” 26 U.S.C. § 5845(b). There are competing theories as to whether this language is best construed to cover bump stocks. In my own view, the relevant language is at best ambiguous. That makes this an easy case for invoking the rule of lenity. Accordingly, I agree that we should reverse. I. Federal law defines a machinegun as a “weapon which shoots . . . automatically more than one shot . . . by a single function of the trigger.” 26 U.S.C. § 5845(b). The definition also includes “any part designed and intended . . . for use in converting a weapon into a machinegun.” Id. Violators of the machinegun ban risk up to ten years in federal prison. 18 U.S.C. § 924(a)(2). I see at least two challenges with reading § 5845(b) to cover bump stocks. To begin with, it’s at best ambiguous whether a semiautomatic weapon equipped with a bump stock is indeed capable of shooting more than one shot by “a single function of the trigger.” Id. And even if I could get past that problem, it’s also ambiguous at best whether the weapon does so “automatically.” Id. So the rule of lenity requires us to reverse. A. The phrase “single function of the trigger” is not a matter of common parlance. No one has identified an example of this phrase ever being used in any context other than this statute. So it’s a term that requires interpretation. And in this context, the interpretive steps it takes to get from “single function of the trigger” to the criminalization of bump stocks fall short of the fair notice that lenity requires. What does “single function of the trigger” mean? “[F]unction” means “action.” Webster’s New International Dictionary 1019 (2nd ed. 1934). But where does that leave us? If it means that there is a single action on the trigger from the shooter’s perspective, then this language might very well apply to bump stocks. That’s because a single action on the trigger by the shooter is enough to spray multiple bullets. But if the phrase means that there is a single action by the trigger from the weapon’s perspective, then this language would not apply to bump stocks. That’s because from the weapon’s perspective, a single action by the trigger releases just one bullet. See, e.g., Gun Owners of Am., Inc. v. Garland, 992 F.3d 446, 469 (6th Cir. 2021) (“the question is whether ‘function’ is referring to the mechanical process (i.e., the act of the trigger’s being depressed, released, and reset) or the human process (i.e., the shooter’s pulling, or otherwise acting upon, the trigger)”), vacated on reh’g en banc, 2 F.4th 576 (6th Cir. 2021). I conclude that grammar and syntax are at best inconclusive, and that lenity therefore requires us to side with the citizen over the government. The problem here is the distinction between what classical grammarians call the “subjective genitive” and the “objective genitive.” See, e.g., Laurel J. Brinton, The Structure of Modern English: A Linguistic Introduction 108 (2000) (“The phrase the shooting of the hunters is ambiguous between subjective and objective genitive readings because it can mean either ‘the hunters shoot X’ or ‘X shoots the hunters.’”); see also Thomas Kerchever Arnold, An English Grammar for Classical Schools 98 (1848); Sidney Greenbaum & Randolph Quirk, A Student’s Grammar of the English Language103 (1990). Take, for example, the phrase “the love of my children.” That phrase could mean that my children love me (in other words, my children are the subject of love). Or it could mean that I love my children (so my children are the object of love). Only context can clarify whether children are the subject or the object of love. Compare Emma D.E.N. Southworth, Victor’s Triumph 97 (1874) (“[S]he could not win the love of children.”) (subjective genitive), with Henry James, Stephen Dewhurst’s Autobiography 54 The Atlantic Monthly 649, 650 (1884) (“[S]he had a most vivacious love of children.”) (objective genitive). Consider another example: “the fear of the soldiers.” Without clarifying context, this could mean that the soldiers are fearful (the soldiers are the subject of fear). Or it could mean that someone else is fearful of the soldiers (the soldiers are the object of fear). Compare Jan Palmper Fear: Soldiers and Emotion in Early Twentieth-Century Russian Military Psychology, 68 Slavic Rev.259, 270 n.38 (2009) (“[T]he fear that officers inspired in soldiers could also cause harm[] . . . . Only just, law-abiding behavior on the part of officers could keep this particular fear of soldiers in check.”) (subjective genitive), with C. Stanley Smith, Five Days, 1927 The Atlantic Monthly 836, 841 (“The fact that I also felt absolutely no fear of the soldiers contributed largely to my safety.”) (objective genitive). Naturally, then, this same ambiguity can also exist when a genitive construction modifies the term “function.” For example, the phrase “function of the polls” could refer to what polls do (polls are the subject). See A. Stuart, Norman L. Webb & D. Butler, Public Opinion Polls, 142 J. Royal Stat. Soc’y 443, 447 (1979) (“The second and most obvious function of polls is that they show the progress of electoral campaigns in a purely informative sense.”) (subjective genitive). Or it could refer to what voters do at the polls (polls are the object). See H.W. Warner, The Republic: Things as They Are at Present, Compared with the Past, 4 Am. Rev.278, 279 (1849) (early American lawmakers established age, residency, and property requirements as “necessary qualifications for the function of the polls”) (objective genitive). By the same token, a “function of the trigger” could mean what the trigger does (the trigger is the subject). Or it could mean what the shooter does to the trigger (the trigger is the object). A “single function of the trigger” could mean that the trigger acts once—or that the shooter acts once on the trigger. The government’s reading prevails if it’s clear that the trigger is the object, and the shooter is the implied subject. But there’s nothing inherent about the phrase “function of the trigger” that tells us that the trigger is the object rather than the subject. So grammar alone can’t dictate whether we should read the statute from the machine’s perspective (subjective genitive) or the shooter’s perspective (objective genitive). As the D.C. Circuit explained: “[T]he text is silent on the crucial question of which perspective is relevant. A mechanical perspective . . . might focus on the trigger’s release of the hammer, which causes the release of a round. From that perspective, a ‘single function of the trigger’ yields a single round of fire when a bump-stock device moves the trigger back and forth. By contrast, from the perspective of the shooter’s action, the function of pulling the trigger a single time results in repeated shots when a bump-stock device is engaged. From that perspective, then, a ‘single function of the trigger’ yields multiple rounds of fire.” Guedes v. Bureau of Alcohol, Tobacco, Firearms & Explosives, 920 F.3d 1, 29 (D.C. Cir. 2019). In the examples provided above, the history or context of a particular phrase sheds light on whether it contains a subjective or objective genitive. Take another example: “The fear of the Lord is the beginning of wisdom.” Psalm 111:10. Construed as a subjective genitive, “the fear of the lord” could refer to an anxious aristocrat, afraid of an overweening monarch or unruly populace. But we know that the Bible means something very different—that the wise man fears God. Unlike “fear of the Lord,” a “single function of the trigger” does not offer much by way of history or context. No example of its usage in any other context has been provided by anyone in this litigation. Congress appears to have coined the phrase just for this statute.[14] As between construing “single function of the trigger” from the weapon’s perspective or the shooter’s perspective, then, the statute appears to be in equipoise. And statutory equipoise is a textbook case for lenity.[15]